{
  "$schema": "https://seogeo.org/datasets/schema.json",
  "format": "seogeo.dataset/1",
  "dataset": "evidence",
  "title": "SEOGEO Evidence Dataset",
  "description": "Dated AI answer observations with the recorded question, product context, citation labels and every visible source link.",
  "site": "https://seogeo.org/",
  "collection_url": "https://seogeo.org/evidence",
  "language": "en",
  "record_count": 120,
  "content_updated_at": "2026-11-15T13:59:00.000Z",
  "terms": {
    "name": "SEOGEO public data terms",
    "url": "https://seogeo.org/terms#acceptable-use",
    "attribution": "SEOGEO"
  },
  "analysis_metadata": {
    "collection_round": "2026-11-15",
    "totals": {
      "observations": 120,
      "series": 40,
      "products": 3,
      "pairwiseComparisons": 120,
      "capturedCitationRows": 718,
      "uniqueDomains": 270,
      "totalSourceAppearances": 673,
      "answerSimilarityMean": 0.3554050361004182,
      "answerSimilarityMedian": 0.34870990903162485,
      "domainOverlapMean": 0.6512875465178097,
      "domainOverlapMedian": 1,
      "seriesWithRecurringDomain": 23,
      "partialCaptureSeries": 2
    },
    "coverage": [
      {
        "productId": "perplexity",
        "product": "Perplexity",
        "target": 60,
        "publishedRuns": 60,
        "capturedDrafts": 0,
        "pending": 0
      },
      {
        "productId": "copilot",
        "product": "Microsoft Copilot",
        "target": 60,
        "publishedRuns": 9,
        "capturedDrafts": 18,
        "pending": 33
      },
      {
        "productId": "chatgpt",
        "product": "ChatGPT",
        "target": 51,
        "publishedRuns": 51,
        "capturedDrafts": 0,
        "pending": 0
      }
    ],
    "product_summaries": [
      {
        "productId": "perplexity",
        "product": "Perplexity",
        "completeSeries": 20,
        "observations": 60,
        "answerSimilarityMean": 0.3912321941139174,
        "domainOverlapMean": 0.8823099415204678,
        "recurringDomainSeries": 19,
        "partialCaptureSeries": 2,
        "citationMean": 9.816666666666666
      },
      {
        "productId": "copilot",
        "product": "Microsoft Copilot",
        "completeSeries": 3,
        "observations": 9,
        "answerSimilarityMean": 0.20078014051850776,
        "domainOverlapMean": 0.023989898989898988,
        "recurringDomainSeries": 0,
        "partialCaptureSeries": 0,
        "citationMean": 5.5555555555555545
      },
      {
        "productId": "chatgpt",
        "product": "ChatGPT",
        "completeSeries": 17,
        "observations": 51,
        "answerSimilarityMean": 0.34054218471663833,
        "domainOverlapMean": 0.4901960784313725,
        "recurringDomainSeries": 4,
        "partialCaptureSeries": 0,
        "citationMean": 1.5490196078431373
      }
    ],
    "platform_comparisons": [
      {
        "panelId": "official-documentation-preference",
        "question": "Do AI search engines prefer official documentation?",
        "leftProductId": "chatgpt",
        "leftProduct": "ChatGPT",
        "rightProductId": "perplexity",
        "rightProduct": "Perplexity",
        "leftSeries": "chatgpt-official-documentation-preference",
        "rightSeries": "perplexity-official-documentation-preference",
        "answerSimilarity": 0.08414353657043132,
        "domainOverlap": 0,
        "sharedDomains": [],
        "leftCitationMean": 0,
        "rightCitationMean": 10
      },
      {
        "panelId": "get-cited-chatgpt",
        "question": "How can a new website get cited by ChatGPT?",
        "leftProductId": "chatgpt",
        "leftProduct": "ChatGPT",
        "rightProductId": "perplexity",
        "rightProduct": "Perplexity",
        "leftSeries": "chatgpt-get-cited-chatgpt",
        "rightSeries": "perplexity-get-cited-chatgpt",
        "answerSimilarity": 0.22641293773722815,
        "domainOverlap": 0,
        "sharedDomains": [],
        "leftCitationMean": 1.3333333333333333,
        "rightCitationMean": 10
      },
      {
        "panelId": "get-cited-perplexity",
        "question": "How can a new website get cited by Perplexity?",
        "leftProductId": "chatgpt",
        "leftProduct": "ChatGPT",
        "rightProductId": "perplexity",
        "rightProduct": "Perplexity",
        "leftSeries": "chatgpt-get-cited-perplexity",
        "rightSeries": "perplexity-get-cited-perplexity",
        "answerSimilarity": 0.1931902306362818,
        "domainOverlap": 0,
        "sharedDomains": [],
        "leftCitationMean": 1.6666666666666667,
        "rightCitationMean": 10
      },
      {
        "panelId": "verify-ai-citation",
        "question": "How can a publisher verify that an AI answer cited its source correctly?",
        "leftProductId": "chatgpt",
        "leftProduct": "ChatGPT",
        "rightProductId": "perplexity",
        "rightProduct": "Perplexity",
        "leftSeries": "chatgpt-verify-ai-citation",
        "rightSeries": "perplexity-verify-ai-citation",
        "answerSimilarity": 0.09325553247999621,
        "domainOverlap": 0,
        "sharedDomains": [],
        "leftCitationMean": 1.3333333333333333,
        "rightCitationMean": 10
      },
      {
        "panelId": "comparison-page-citations",
        "question": "How do AI search engines choose comparison pages as sources?",
        "leftProductId": "chatgpt",
        "leftProduct": "ChatGPT",
        "rightProductId": "perplexity",
        "rightProduct": "Perplexity",
        "leftSeries": "chatgpt-comparison-page-citations",
        "rightSeries": "perplexity-comparison-page-citations",
        "answerSimilarity": 0.08536938560738566,
        "domainOverlap": 0,
        "sharedDomains": [],
        "leftCitationMean": 0.3333333333333333,
        "rightCitationMean": 10
      },
      {
        "panelId": "how-ai-search-selects-sources",
        "question": "How do AI search engines choose which sources to cite?",
        "leftProductId": "copilot",
        "leftProduct": "Microsoft Copilot",
        "rightProductId": "perplexity",
        "rightProduct": "Perplexity",
        "leftSeries": "copilot-how-ai-search-selects-sources",
        "rightSeries": "perplexity-how-ai-search-selects-sources",
        "answerSimilarity": 0.1941247706486712,
        "domainOverlap": 0,
        "sharedDomains": [],
        "leftCitationMean": 5.333333333333333,
        "rightCitationMean": 5.333333333333333
      },
      {
        "panelId": "source-freshness",
        "question": "How do AI search engines decide whether information is fresh enough?",
        "leftProductId": "chatgpt",
        "leftProduct": "ChatGPT",
        "rightProductId": "perplexity",
        "rightProduct": "Perplexity",
        "leftSeries": "chatgpt-source-freshness",
        "rightSeries": "perplexity-source-freshness",
        "answerSimilarity": 0.0942040184253758,
        "domainOverlap": 0,
        "sharedDomains": [],
        "leftCitationMean": 2.3333333333333335,
        "rightCitationMean": 10
      },
      {
        "panelId": "source-change-answer-drift",
        "question": "How do source changes affect future AI citations?",
        "leftProductId": "chatgpt",
        "leftProduct": "ChatGPT",
        "rightProductId": "perplexity",
        "rightProduct": "Perplexity",
        "leftSeries": "chatgpt-source-change-answer-drift",
        "rightSeries": "perplexity-source-change-answer-drift",
        "answerSimilarity": 0.0898875478416721,
        "domainOverlap": 0,
        "sharedDomains": [],
        "leftCitationMean": 0.6666666666666666,
        "rightCitationMean": 10
      },
      {
        "panelId": "chatgpt-search-source-selection",
        "question": "How does ChatGPT Search choose the sources it cites?",
        "leftProductId": "chatgpt",
        "leftProduct": "ChatGPT",
        "rightProductId": "perplexity",
        "rightProduct": "Perplexity",
        "leftSeries": "chatgpt-chatgpt-search-source-selection",
        "rightSeries": "perplexity-chatgpt-search-source-selection",
        "answerSimilarity": 0.07864419049695826,
        "domainOverlap": 0,
        "sharedDomains": [],
        "leftCitationMean": 2,
        "rightCitationMean": 10
      },
      {
        "panelId": "google-ai-mode-source-selection",
        "question": "How does Google AI Mode choose the sources it cites?",
        "leftProductId": "chatgpt",
        "leftProduct": "ChatGPT",
        "rightProductId": "perplexity",
        "rightProduct": "Perplexity",
        "leftSeries": "chatgpt-google-ai-mode-source-selection",
        "rightSeries": "perplexity-google-ai-mode-source-selection",
        "answerSimilarity": 0.09986854648218588,
        "domainOverlap": 0,
        "sharedDomains": [],
        "leftCitationMean": 4.666666666666667,
        "rightCitationMean": 10
      },
      {
        "panelId": "b2b-saas-visibility",
        "question": "How should a B2B SaaS company measure AI search visibility?",
        "leftProductId": "chatgpt",
        "leftProduct": "ChatGPT",
        "rightProductId": "perplexity",
        "rightProduct": "Perplexity",
        "leftSeries": "chatgpt-b2b-saas-visibility",
        "rightSeries": "perplexity-b2b-saas-visibility",
        "answerSimilarity": 0.11126571573685741,
        "domainOverlap": 0,
        "sharedDomains": [],
        "leftCitationMean": 5,
        "rightCitationMean": 10
      },
      {
        "panelId": "measure-chatgpt-citations",
        "question": "How should a new website measure citations in ChatGPT and Perplexity?",
        "leftProductId": "copilot",
        "leftProduct": "Microsoft Copilot",
        "rightProductId": "perplexity",
        "rightProduct": "Perplexity",
        "leftSeries": "copilot-measure-chatgpt-citations",
        "rightSeries": "perplexity-measure-chatgpt-citations",
        "answerSimilarity": 0.16170546346223985,
        "domainOverlap": 0.007407407407407408,
        "sharedDomains": [
          "blog.hubspot.com"
        ],
        "leftCitationMean": 6.333333333333333,
        "rightCitationMean": 10
      },
      {
        "panelId": "audit-ai-answer",
        "question": "How should a publisher audit an AI-generated answer?",
        "leftProductId": "chatgpt",
        "leftProduct": "ChatGPT",
        "rightProductId": "perplexity",
        "rightProduct": "Perplexity",
        "leftSeries": "chatgpt-audit-ai-answer",
        "rightSeries": "perplexity-audit-ai-answer",
        "answerSimilarity": 0.0850098452516946,
        "domainOverlap": 0,
        "sharedDomains": [],
        "leftCitationMean": 2.6666666666666665,
        "rightCitationMean": 10
      },
      {
        "panelId": "track-citations-over-time",
        "question": "How should a team track AI citations over time?",
        "leftProductId": "chatgpt",
        "leftProduct": "ChatGPT",
        "rightProductId": "perplexity",
        "rightProduct": "Perplexity",
        "leftSeries": "chatgpt-track-citations-over-time",
        "rightSeries": "perplexity-track-citations-over-time",
        "answerSimilarity": 0.0800330449486996,
        "domainOverlap": 0,
        "sharedDomains": [],
        "leftCitationMean": 0,
        "rightCitationMean": 10
      },
      {
        "panelId": "inaccurate-ai-citations",
        "question": "What causes inaccurate AI citations?",
        "leftProductId": "chatgpt",
        "leftProduct": "ChatGPT",
        "rightProductId": "perplexity",
        "rightProduct": "Perplexity",
        "leftSeries": "chatgpt-inaccurate-ai-citations",
        "rightSeries": "perplexity-inaccurate-ai-citations",
        "answerSimilarity": 0.09717722253525131,
        "domainOverlap": 0,
        "sharedDomains": [],
        "leftCitationMean": 0,
        "rightCitationMean": 10
      },
      {
        "panelId": "product-page-evidence",
        "question": "What evidence makes a product page citable by AI search?",
        "leftProductId": "chatgpt",
        "leftProduct": "ChatGPT",
        "rightProductId": "perplexity",
        "rightProduct": "Perplexity",
        "leftSeries": "chatgpt-product-page-evidence",
        "rightSeries": "perplexity-product-page-evidence",
        "answerSimilarity": 0.07510016742954971,
        "domainOverlap": 0,
        "sharedDomains": [],
        "leftCitationMean": 1,
        "rightCitationMean": 10
      },
      {
        "panelId": "mention-vs-citation",
        "question": "What is the difference between an AI mention and an AI citation?",
        "leftProductId": "chatgpt",
        "leftProduct": "ChatGPT",
        "rightProductId": "perplexity",
        "rightProduct": "Perplexity",
        "leftSeries": "chatgpt-mention-vs-citation",
        "rightSeries": "perplexity-mention-vs-citation",
        "answerSimilarity": 0.09013334589663326,
        "domainOverlap": 0,
        "sharedDomains": [],
        "leftCitationMean": 0,
        "rightCitationMean": 10
      },
      {
        "panelId": "citation-worthy-pages",
        "question": "What makes a web page more likely to be cited by AI search?",
        "leftProductId": "copilot",
        "leftProduct": "Microsoft Copilot",
        "rightProductId": "perplexity",
        "rightProduct": "Perplexity",
        "leftSeries": "copilot-citation-worthy-pages",
        "rightSeries": "perplexity-citation-worthy-pages",
        "answerSimilarity": 0.09856502644011754,
        "domainOverlap": 0.06216931216931217,
        "sharedDomains": [
          "blog.hubspot.com",
          "developers.google.com",
          "searchable.com"
        ],
        "leftCitationMean": 5,
        "rightCitationMean": 11
      },
      {
        "panelId": "local-business-ai-search",
        "question": "Which pages should a local business optimize for AI search?",
        "leftProductId": "chatgpt",
        "leftProduct": "ChatGPT",
        "rightProductId": "perplexity",
        "rightProduct": "Perplexity",
        "leftSeries": "chatgpt-local-business-ai-search",
        "rightSeries": "perplexity-local-business-ai-search",
        "answerSimilarity": 0.075109383447953,
        "domainOverlap": 0,
        "sharedDomains": [],
        "leftCitationMean": 0.3333333333333333,
        "rightCitationMean": 10
      },
      {
        "panelId": "source-types-cited-most",
        "question": "Which source types do AI search engines cite most often?",
        "leftProductId": "chatgpt",
        "leftProduct": "ChatGPT",
        "rightProductId": "perplexity",
        "rightProduct": "Perplexity",
        "leftSeries": "chatgpt-source-types-cited-most",
        "rightSeries": "perplexity-source-types-cited-most",
        "answerSimilarity": 0.24427879429090962,
        "domainOverlap": 0.060606060606060615,
        "sharedDomains": [
          "bavior"
        ],
        "leftCitationMean": 3,
        "rightCitationMean": 10
      }
    ]
  },
  "records": [
    {
      "id": "evidence/chatgpt-audit-ai-answer-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-audit-ai-answer-repeat-2",
      "published_at": "2026-11-15T13:59:00.000Z",
      "updated_at": "2026-11-15T13:59:00.000Z",
      "revision": 1,
      "slug": "chatgpt-audit-ai-answer-repeat-2",
      "panelId": "audit-ai-answer",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-audit-ai-answer",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "How should a publisher audit an AI-generated answer?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:59 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9bab1-913c-83ea-9e2d-30fdbbe6d163?temporary-chat=true",
      "mentions": "Cited",
      "summary": "A publisher should audit an AI-generated answer as though it were a draft submitted by an unfamiliar contributor: assume nothing is correct until the important claims have been independently verified.",
      "answerSnapshot": "：\n\nA publisher should audit an AI-generated answer as though it were a draft submitted by an unfamiliar contributor: assume nothing is correct until the important claims have been independently verified.\n\nThe goal is not simply to check whether the answer sounds convincing. It is to establish whether it is accurate, properly sourced, fair, legally appropriate and suitable for publication.\n\n1. A six-step editorial audit\n\nStage\n\n\t\n\nWhat the publisher should check\n\n\n\n\n1. Factual accuracy\n\n\t\n\nAre names, dates, statistics, quotations and events correct?\n\n\n\n\n2. Source verification\n\n\t\n\nDo the cited sources exist, and do they actually support the claims?\n\n\n\n\n3. Completeness\n\n\t\n\nHas the AI omitted important context, qualifications or opposing viewpoints?\n\n\n\n\n4. Bias and fairness\n\n\t\n\nDoes the answer contain stereotypes, misleading framing or unsupported assumptions?\n\n\n\n\n5. Legal and ethical risks\n\n\t\n\nAre there copyright, privacy, defamation or confidentiality concerns?\n\n\n\n\n6. Editorial quality\n\n\t\n\nIs the answer clear, relevant, appropriately written and consistent with editorial standards?\n\n2. Verify claims individually, not just the overall answer\n\nThis is arguably the most important step.\n\nA publisher should break the AI answer into individual claims and create a claim-verification record.\n\nFor example, consider an AI-generated statement:\n\n\"The UK publishing industry grew by 12% in 2025.\"\n\nThe auditor should establish:\n\nWhere did the figure originate?\n\nDoes the original report actually contain that figure?\n\nDoes it refer to the whole publishing industry or a particular segment?\n\nIs it a year-on-year comparison?\n\nIs the figure presented in the right context?\n\nCan an independent source corroborate it?\n\nImportant: A citation is not proof. The auditor must open the source and check that it supports the exact claim.\n\nPrioritise:\n\nOriginal research and official statistics.\n\nGovernment and regulatory publications.\n\nOriginal documents and transcripts.\n\nReputable independent reporting.\n\nNamed experts with relevant expertise.\n\nAvoid treating other AI-generated answers as independent corroboration.\n\n3. Apply a risk-based review\n\nNot every AI-generated answer needs the same level of scrutiny.\n\nRisk level\n\n\t\n\nExample\n\n\t\n\nRecommended review\n\n\n\n\nLow\n\n\t\n\nHeadline suggestions, general summaries\n\n\t\n\nEditorial review\n\n\n\n\nMedium\n\n\t\n\nIndustry analysis, market commentary\n\n\t\n\nClaim-by-claim fact-check\n\n\n\n\nHigh\n\n\t\n\nMedical, financial, legal or political information\n\n\t\n\nSpecialist review and independent verification\n\n\n\n\nCritical\n\n\t\n\nAllegations about individuals, public safety or sensitive investigations\n\n\t\n\nSenior editorial and, where appropriate, legal approval\n\nThe higher the potential consequences of an error, the stronger the verification and approval process should be.\n\nThis reflects the Council of Europe's emphasis on meaningful editorial oversight, particularly for sensitive and consequential material. \n---Adopted by the Steering Committee\n+1\n\n4. Check for AI-specific problems\n\nPublishers should specifically look for:\n\nHallucinations: Invented facts, references, statistics or quotations.\n\nFalse attribution: Real people being credited with statements they never made.\n\nOutdated information: Previously accurate information presented as current.\n\nContext distortion: Accurate facts arranged to imply something misleading.\n\nBias: Unfair representation of individuals or groups.\n\nOverconfidence: Speculation presented as established fact.\n\nPlagiarism: Material reproduced without appropriate attribution or permission.\n\nA useful principle from Reuters' journalistic standards is that AI-generated facts, sources and claims must be independently verified. \nreutersagency.com\n+1\n\n5. Maintain an audit trail\n\nA publisher should retain a record of the checks performed, especially for content that carries significant editorial risk.\n\nA practical audit record might look like this:\n\nAudit field\n\n\t\n\nWhat to record\n\n\n\n\nAI tool\n\n\t\n\nName of the AI system\n\n\n\n\nContent\n\n\t\n\nThe generated answer or article\n\n\n\n\nDate\n\n\t\n\nDate generated and reviewed\n\n\n\n\nClaims\n\n\t\n\nIndividual factual statements\n\n\n\n\nEvidence\n\n\t\n\nSources used to verify each claim\n\n\n\n\nProblems\n\n\t\n\nErrors, omissions and unsupported statements\n\n\n\n\nCorrections\n\n\t\n\nChanges made by the editor\n\n\n\n\nApproval\n\n\t\n\nName or role of approving editor\n\n\n\n\nDisclosure\n\n\t\n\nWhether AI assistance needs to be acknowledged\n\nThis makes the process traceable and allows the publisher to demonstrate that appropriate editorial checks took place.\n\n6. Establish transparency and accountability\n\nBefore publication, the publisher should establish:\n\nWhether readers need to be informed that AI was used.\n\nWhether AI generated the entire answer or assisted with drafting, research or editing.\n\nWho is responsible for the final published content.\n\nHow readers can report inaccuracies.\n\nHow corrections will be made and communicated.\n\nAI involvement does not transfer editorial responsibility away from the publisher. In the UK, IPSO makes clear that its Editors' Code applies regardless of the tools used to produce content. \nIPSO\n\n7. A practical publisher's audit checklist\n\nThe following can be used as a pre-publication quality-control checklist.\n\nPre-publication audit\n\n0/10 checked\nEvery factual claim has been checked against reliable evidence.\nAll citations and hyperlinks have been opened and verified.\nStatistics, dates, names and quotations are accurate.\nSources support the precise claims attributed to them.\nImportant context and alternative viewpoints have not been omitted.\nPotential bias, misleading framing and unsupported assumptions have been examined.\nCopyright, privacy and confidentiality risks have been assessed.\nThe answer meets the publication's editorial and legal standards.\nAI involvement has been disclosed where appropriate.\nA responsible human editor has approved the final version.\nReset checklist\n8. Recommended reference frameworks\n\nPublishers can use these established standards to develop their own audit procedures:\n\nIPSO – Does AI-generated content change editorial standards? — particularly relevant to UK publishers.\n\nReuters Journalistic Standards — useful for independent verification and editorial accountability.\n\nCouncil of Europe – Guidelines on responsible AI in journalism — useful for governance, human oversight and transparency.\n\nThomson Reuters Foundation – A Journalist's Guide to Safer Reporting — includes practical AI-related editorial checks.\n\nThe fundamental principle\n\nA publisher should distinguish between three separate questions:\n\nIs the answer plausible? Does it sound convincing?\n\nIs the answer verifiable? Can its claims be supported by reliable evidence?\n\nIs the answer publishable? Does it meet the publisher's standards for accuracy, fairness, legality and accountability?\n\nOnly the third question determines whether the content is ready for publication.\n\nMy recommendation: Adopt a human-led, claim-by-claim verification process, supported by a documented audit trail and risk-based editorial approval. AI can help produce content, but it should never be treated as the authority validating its own answers.",
      "citations": [
        {
          "title": "IPSO – Does AI-generated content change editorial standards?",
          "host": "ipso.co.uk",
          "url": "https://www.ipso.co.uk/news-analysis/does-ai-generated-content-change-editorial-standards/?utm_source=chatgpt.com",
          "position": 1
        },
        {
          "title": "Reuters Journalistic Standards",
          "host": "reutersagency.com",
          "url": "https://reutersagency.com/about/standards-values/?utm_source=chatgpt.com",
          "position": 2
        },
        {
          "title": "Council of Europe – Guidelines on responsible AI in journalism",
          "host": "edoc.coe.int",
          "url": "https://edoc.coe.int/en/module/ec_addformat/download?cle=856b1cacaea25f6ebbf01b0003030a0b&k=5004f45bea8dc1a3b6276d3b08cafe2a&utm_source=chatgpt.com",
          "position": 3
        },
        {
          "title": "Thomson Reuters Foundation – A Journalist's Guide to Safer Reporting",
          "host": "trust.org",
          "url": "https://www.trust.org/wp-content/uploads/2025/07/A-Journalists-Guide-to-Safer-Reporting.pdf?utm_source=chatgpt.com",
          "position": 4
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 4 sources and 4 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-audit-ai-answer-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-audit-ai-answer-repeat-3",
      "published_at": "2026-11-15T13:59:00.000Z",
      "updated_at": "2026-11-15T13:59:00.000Z",
      "revision": 1,
      "slug": "chatgpt-audit-ai-answer-repeat-3",
      "panelId": "audit-ai-answer",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-audit-ai-answer",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "How should a publisher audit an AI-generated answer?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:59 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9bac3-c65c-83ea-b9dc-b4379d895b14?temporary-chat=true",
      "mentions": "Not observed",
      "summary": "A publisher should audit an AI-generated answer as if it were a draft produced by an inexperienced researcher: check every important claim, verify the evidence, assess the context, and take editorial responsibility before publication.",
      "answerSnapshot": "：\n\nA publisher should audit an AI-generated answer as if it were a draft produced by an inexperienced researcher: check every important claim, verify the evidence, assess the context, and take editorial responsibility before publication.\n\nThe goal is not simply to determine whether the answer sounds convincing. It is to establish whether it is accurate, fair, properly sourced, legally appropriate and safe for readers.\n\n1. A publisher's AI answer audit\n\nA robust audit can be divided into eight checks.\n\nAudit area\n\n\t\n\nWhat the publisher should check\n\n\n\n\nFactual accuracy\n\n\t\n\nAre names, dates, statistics, quotations and events correct?\n\n\n\n\nSource verification\n\n\t\n\nDo cited sources actually exist and support the claims attributed to them?\n\n\n\n\nCompleteness\n\n\t\n\nHas the AI omitted important facts, counterarguments or qualifications?\n\n\n\n\nContext\n\n\t\n\nDoes the answer represent the original source and subject fairly?\n\n\n\n\nBias and fairness\n\n\t\n\nDoes it contain stereotypes, loaded language or unbalanced perspectives?\n\n\n\n\nOriginality\n\n\t\n\nDoes it reproduce copyrighted material or closely imitate another publication?\n\n\n\n\nSafety and legal risk\n\n\t\n\nCould it be defamatory, misleading, privacy-invasive or harmful?\n\n\n\n\nEditorial quality\n\n\t\n\nIs it clear, relevant, appropriately toned and suitable for the intended audience?\n\n2. Verify claims against original sources\n\nThis is arguably the most important part of the audit.\n\nDo not assume that an AI-generated citation is genuine or that a real source supports the statement.\n\nFor each material claim:\n\nIdentify the original source, rather than relying on another AI summary.\n\nOpen the source and locate the relevant passage.\n\nCheck that the source supports the exact claim being made.\n\nVerify dates, figures, quotations and names independently.\n\nCheck whether the source is authoritative, current and relevant.\n\nRecord the source URL and the date it was checked.\n\nA useful distinction: A source can be genuine but still be misrepresented by an AI-generated answer.\n\n3. Assess the quality of the answer\n\nPublishers should look beyond individual factual errors.\n\nConsider these questions:\n\nDoes the answer address the actual question?\n\nDoes it distinguish established facts from opinions and predictions?\n\nDoes it communicate uncertainty where appropriate?\n\nDoes it present competing evidence fairly?\n\nDoes it exaggerate the importance of particular findings?\n\nDoes it make unsupported conclusions from limited evidence?\n\nDoes it give readers enough context to understand the subject?\n\nAn answer can contain entirely accurate individual statements while creating a misleading overall impression.\n\n4. Check copyright, attribution and originality\n\nBefore publication, check whether the answer:\n\nReproduces substantial passages from existing articles or books.\n\nUses quotations without appropriate attribution.\n\nPresents another author's distinctive analysis as original.\n\nIncludes images or other material without appropriate permissions.\n\nMakes claims about original reporting or research that was never conducted.\n\nPlagiarism-detection software can help identify similarities, but it cannot establish copyright compliance on its own.\n\n5. Apply a risk-based editorial review\n\nNot every AI-generated answer requires the same level of scrutiny.\n\nContent type\n\n\t\n\nRecommended review\n\n\n\n\nGeneral informational content\n\n\t\n\nStandard editorial fact-check\n\n\n\n\nNews and current affairs\n\n\t\n\nIndependent verification of material claims\n\n\n\n\nHealth and scientific content\n\n\t\n\nSpecialist review and authoritative evidence\n\n\n\n\nFinancial or legal information\n\n\t\n\nQualified expert review\n\n\n\n\nAllegations about individuals\n\n\t\n\nSenior editorial and legal review\n\n\n\n\nSensitive personal information\n\n\t\n\nPrivacy and data-protection review\n\nThe higher the potential consequences of an error, the more rigorous the review should be.\n\n6. Use a documented audit checklist\n\nHere is a practical checklist that an editorial team can use before approving an AI-generated answer.\n\nPre-publication audit\n\n0/11 complete\nThe answer addresses the original question.\nAll material factual claims have been verified.\nEvery citation has been opened and checked.\nStatistics, dates, names and quotations are accurate.\nSources are authoritative and sufficiently current.\nImportant context and alternative perspectives are included.\nPotential bias and misleading language have been assessed.\nCopyright, attribution and originality have been checked.\nPrivacy, legal and reputational risks have been considered.\nThe answer meets the publication's editorial standards.\nA human editor has approved the final version.\nReset checklist\n7. Maintain an audit trail\n\nFor professional publishing, the audit should be documented.\n\nRecord:\n\nThe original AI-generated answer.\n\nThe AI system and model used, where known.\n\nThe original prompt and relevant instructions.\n\nThe sources used to verify claims.\n\nChanges made by the editor.\n\nAny specialist reviews undertaken.\n\nThe name or role of the approving editor.\n\nThe date of approval.\n\nThis makes it easier to investigate complaints, correct errors and demonstrate editorial accountability.\n\n8. Establish clear editorial responsibility\n\nAI should assist the editorial process, not replace editorial judgement.\n\nThe publisher remains responsible for what it publishes, regardless of whether the original text was written by a journalist, an AI system or a combination of both.\n\nA sound policy should establish:\n\nWhen AI-generated content requires disclosure.\n\nWhich content categories require human review.\n\nWho is authorised to approve publication.\n\nHow corrections and complaints are handled.\n\nHow AI tools and editorial procedures are periodically reassessed.\n\nThe recommended standard\n\nI would summarise the process in five stages:\n\nInspect\n\nIdentify factual claims, opinions, quotations and potentially sensitive material.\n\nVerify\n\nCheck claims against original, authoritative sources.\n\nEvaluate\n\nAssess context, completeness, bias, copyright and potential harm.\n\nEdit\n\nCorrect inaccuracies, add missing context and improve clarity.\n\nApprove\n\nObtain documented human editorial sign-off before publication.\n\nThe central principle: Never publish an AI-generated answer simply because it is fluent, plausible or accompanied by citations. Publish it only when its material claims have been independently verified and an accountable human editor is satisfied with the result.\n\nFor a publisher, this is the difference between merely proofreading AI output and conducting a genuine editorial audit.",
      "citations": [],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 0 sources and 0 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-audit-ai-answer",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-audit-ai-answer",
      "published_at": "2026-11-15T13:58:00.000Z",
      "updated_at": "2026-11-15T13:58:00.000Z",
      "revision": 1,
      "slug": "chatgpt-audit-ai-answer",
      "panelId": "audit-ai-answer",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-audit-ai-answer",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "How should a publisher audit an AI-generated answer?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:58 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9ba8c-9910-83ea-b907-6c99a47da1f4?temporary-chat=true",
      "mentions": "Cited",
      "summary": "A publisher should audit an AI-generated answer as if it were a draft submitted by an inexperienced researcher: nothing should be published simply because it sounds convincing.",
      "answerSnapshot": "：\n\nA publisher should audit an AI-generated answer as if it were a draft submitted by an inexperienced researcher: nothing should be published simply because it sounds convincing.\n\nThe goal is to establish that the answer is accurate, verifiable, appropriately sourced, unbiased, legally safe and suitable for the intended audience.\n\n1. A seven-step AI answer audit\n\nCheck factual accuracy\n\nBreak the answer into individual factual claims. Verify each claim against reliable, independent sources. Pay particular attention to:\n\nNames, dates, figures and statistics.\n\nQuotations and attributed statements.\n\nScientific, legal and historical claims.\n\nClaims that may have changed recently.\n\nVerify every source\n\nCheck that references actually exist and support the claims attributed to them.\n\nLook for fabricated citations, incorrect URLs, misleading quotations and sources that merely repeat one another.\n\nCheck completeness and context\n\nEstablish whether the answer omits important facts, qualifications, alternative explanations or relevant counterarguments.\n\nAn answer can contain only true statements and still create a misleading impression.\n\nAssess bias and neutrality\n\nExamine whether the answer:\n\nPresents opinions as established facts.\n\nUses loaded or emotionally persuasive language.\n\nUnfairly represents particular groups.\n\nGives disproportionate attention to one viewpoint.\n\nCheck originality and intellectual property\n\nAssess whether the answer reproduces distinctive passages from existing publications, misattributes material or uses copyrighted content inappropriately.\n\nReview safety, privacy and legal risks\n\nCheck for personal information, defamatory statements, sensitive material, unsupported allegations and potentially harmful advice.\n\nReview readability and editorial quality\n\nCheck clarity, grammar, tone, structure, repetition and whether the answer actually addresses the reader's question.\n\n2. Use a claim-by-claim verification matrix\n\nA practical publisher audit should record evidence rather than simply give the answer an overall quality score.\n\nAudit item\n\n\t\n\nWhat the editor should establish\n\n\n\n\nFactual claims\n\n\t\n\nIs each claim correct?\n\n\n\n\nCitations\n\n\t\n\nDoes the source exist and support the claim?\n\n\n\n\nStatistics\n\n\t\n\nAre figures accurate and current?\n\n\n\n\nQuotations\n\n\t\n\nAre they genuine and correctly attributed?\n\n\n\n\nContext\n\n\t\n\nAre important qualifications missing?\n\n\n\n\nBalance\n\n\t\n\nAre relevant perspectives represented fairly?\n\n\n\n\nOriginality\n\n\t\n\nIs the content sufficiently original?\n\n\n\n\nPrivacy\n\n\t\n\nIs confidential or personal information exposed?\n\n\n\n\nEditorial quality\n\n\t\n\nIs the answer clear, relevant and appropriate?\n\n\n\n\nTransparency\n\n\t\n\nHas AI use been disclosed where required?\n\n3. Introduce a risk-based approval system\n\nNot every AI-generated answer requires the same level of scrutiny.\n\nLow risk\n\nStandard review\n\nGeneral explanations, simple summaries and non-sensitive informational content.\n\nEditor checks factual claims, sources and presentation.\n\nMedium risk\n\nEnhanced review\n\nFinancial explanations, scientific claims, statistics and current affairs.\n\nEditor checks primary sources and obtains subject-expert review where appropriate.\n\nHigh risk\n\nSpecialist approval\n\nMedical, legal, political, reputational or safety-critical material.\n\nRequire specialist verification, senior editorial approval and legal review where appropriate.\n\n4. Maintain an audit trail\n\nFor each answer, retain:\n\nOriginal AI-generated output.\n\nAI tool and model used, where known.\n\nDate of generation.\n\nSources used to verify claims.\n\nCorrections and editorial changes.\n\nName or role of the approving editor.\n\nAny required AI disclosure.\n\nThis makes the process repeatable and helps publishers investigate errors after publication.\n\n5. Apply a publication decision\n\nUse three possible outcomes:\n\nDecision\n\n\t\n\nMeaning\n\n\n\n\nApprove\n\n\t\n\nClaims verified and editorial standards satisfied.\n\n\n\n\nRevise\n\n\t\n\nSome claims need correction, clarification or additional evidence.\n\n\n\n\nReject\n\n\t\n\nSerious inaccuracies, fabricated sources, unacceptable risks or fundamental reliability problems.\n\n6. Recommended reference standards\n\nSeveral established organisations provide useful foundations for an AI audit:\n\nReuters — Journalistic Standards. Particularly useful for accuracy, sourcing, impartiality and human accountability.\n\nThomson Reuters Foundation — A Journalist's Guide to Safer Reporting. Includes practical AI-related editorial checks.\n\nUK Parliament — Working with AI and spotting AI-generated text. Useful guidance on hallucinations and independent verification.\n\nEuropean Broadcasting Union — News Integrity in AI Assistants Toolkit. Particularly useful for auditing omissions, sourcing, bias, context and misleading editorialisation.\n\n7. A simple publisher audit scorecard\n\nA publisher can use the following checklist for every answer.\n\nPublication audit\n\n0/10 completed\nAll factual claims independently verified\nAll references checked against original sources\nStatistics, dates and quotations verified\nNo material omissions or misleading context\nFairness and impartiality assessed\nOriginality and attribution checked\nPrivacy, legal and safety risks assessed\nLanguage and editorial quality reviewed\nAI use disclosed where required\nA human editor has approved publication\n\nCompletion is not a substitute for editorial judgement.\n\nReset\n\nThe fundamental principle: AI can assist with producing information, but the publisher retains responsibility for everything it publishes. Human verification, editorial judgement and accountability must remain central to the process. This is consistent with Reuters' AI standards and established publishing ethics. \nreutersagency.com\n+2\n\nA particularly useful next step would be to turn this framework into a formal AI-generated content audit policy and editorial approval template for a publishing organisation.",
      "citations": [
        {
          "title": "Journalistic Standards",
          "host": "reutersagency.com",
          "url": "https://reutersagency.com/about/standards-values/?utm_source=chatgpt.com",
          "position": 1
        },
        {
          "title": "A Journalist's Guide to Safer Reporting",
          "host": "trust.org",
          "url": "https://www.trust.org/wp-content/uploads/2025/07/A-Journalists-Guide-to-Safer-Reporting.pdf?utm_source=chatgpt.com",
          "position": 2
        },
        {
          "title": "Working with AI and spotting AI-generated text",
          "host": "commonslibrary.parliament.uk",
          "url": "https://commonslibrary.parliament.uk/research-briefings/cbp-10823/?utm_source=chatgpt.com",
          "position": 3
        },
        {
          "title": "News Integrity in AI Assistants Toolkit",
          "host": "ebu.ch",
          "url": "https://www.ebu.ch/files/live/sites/ebu/files/Publications/MIS/open/EBU-MIS-BBC_News_Integrity_in_AI_Assistants_Toolkit_2025.pdf?utm_source=chatgpt.com",
          "position": 4
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 4 sources and 4 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-inaccurate-ai-citations-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-inaccurate-ai-citations-repeat-2",
      "published_at": "2026-11-15T13:58:00.000Z",
      "updated_at": "2026-11-15T13:58:00.000Z",
      "revision": 1,
      "slug": "chatgpt-inaccurate-ai-citations-repeat-2",
      "panelId": "inaccurate-ai-citations",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-inaccurate-ai-citations",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "What causes inaccurate AI citations?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:58 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9ba77-2090-83ea-acfc-4c5a584c15c1?temporary-chat=true",
      "mentions": "Not observed",
      "summary": "Inaccurate AI citations happen when an AI system provides references that do not correctly support, match, or even exist in relation to the claims it makes.",
      "answerSnapshot": "：\n\nInaccurate AI citations happen when an AI system provides references that do not correctly support, match, or even exist in relation to the claims it makes.\n\nThis is often called citation hallucination or citation error.\n\n8 common causes of inaccurate AI citations\n\n1. Hallucination\nAI generates plausible-looking references, including fictional authors, article titles, publication dates, or URLs.\n\n2. Poor source retrieval\nThe AI retrieves irrelevant, outdated, or low-quality sources that do not directly address the question.\n\n3. Misinterpretation of sources\nThe AI finds a legitimate source but misunderstands its findings or draws conclusions the source does not support.\n\n4. Incorrect attribution\nInformation from one source is mistakenly attributed to another author, publication, or organisation.\n\n5. Outdated training data\nAI may rely on older information when newer research, statistics, or publications are available.\n\n6. Overgeneralisation\nThe AI uses a source that supports a narrow claim to justify a much broader statement.\n\n7. Poor context understanding\nThe AI misses important qualifications, limitations, or distinctions in the original source.\n\n8. Citation formatting errors\nIncorrect author names, publication years, page numbers, DOIs, or links make otherwise legitimate references difficult to verify.\n\nWhy does this happen?\n\nThree underlying technical issues are particularly important:\n\nTechnical issue\n\n\t\n\nExplanation\n\n\n\n\nProbabilistic generation\n\n\t\n\nAI predicts likely sequences of words rather than inherently verifying every factual claim.\n\n\n\n\nRetrieval limitations\n\n\t\n\nSearch and retrieval systems may return incomplete or irrelevant evidence.\n\n\n\n\nWeak grounding\n\n\t\n\nThe model may generate an answer without ensuring that each individual claim is supported by its cited source.\n\nExample of an inaccurate AI citation\n\nImagine an AI produces this statement:\n\nResearch shows that 85% of students improve their academic performance when using AI tools (Smith, 2023).\n\nPotential problems:\n\nThe Smith (2023) publication might not exist.\n\nThe publication might exist but contain no such statistic.\n\nThe statistic might refer to a different population.\n\nThe study might show correlation rather than causation.\n\nThe citation looks credible, but that does not mean it is accurate.\n\nHow can inaccurate AI citations be prevented?\n\nVerify that every cited publication actually exists.\n\nOpen the original source rather than relying on AI-generated summaries.\n\nCheck whether the source directly supports the associated claim.\n\nCross-check statistics against the original research.\n\nPrefer peer-reviewed research and authoritative publications.\n\nUse tools that retrieve sources and ground AI-generated answers in them.\n\nCheck publication dates and DOI information.\n\nKey takeaway\n\nThe central cause of inaccurate AI citations is the disconnect between generating plausible information and verifying evidence.\n\nAn AI can produce a convincing answer and a professionally formatted reference without either being factually correct.\n\nThis is why citation accuracy requires checking three separate things:\n\nExistence: Does the source actually exist?\n\nAuthenticity: Is the reference information correct?\n\nRelevance: Does the source genuinely support the claim?\n\nAll three are necessary for a reliable AI-generated citation.",
      "citations": [],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 0 sources and 0 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-inaccurate-ai-citations-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-inaccurate-ai-citations-repeat-3",
      "published_at": "2026-11-15T13:58:00.000Z",
      "updated_at": "2026-11-15T13:58:00.000Z",
      "revision": 1,
      "slug": "chatgpt-inaccurate-ai-citations-repeat-3",
      "panelId": "inaccurate-ai-citations",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-inaccurate-ai-citations",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "What causes inaccurate AI citations?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:58 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9ba60-6a28-83ea-b8ce-bc7bd88ebce1?temporary-chat=true",
      "mentions": "Not observed",
      "summary": "Inaccurate AI citations happen when an AI system provides sources that don't actually support the claims it makes, misrepresents source content, or references sources that do not exist.",
      "answerSnapshot": "：\n\nInaccurate AI citations happen when an AI system provides sources that don't actually support the claims it makes, misrepresents source content, or references sources that do not exist.\n\nThis can happen for several reasons:\n\n1. Hallucinations\n\nAI models can generate convincing-looking but fabricated references, including:\n\nNon-existent academic papers.\n\nIncorrect author names or publication dates.\n\nFake URLs, DOIs, or journal titles.\n\nQuotes that do not appear in the cited source.\n\nThis happens because AI predicts plausible text rather than inherently verifying every reference.\n\n2. Poor source retrieval\n\nAI systems often retrieve information from search engines or databases. Citation errors can occur when:\n\nThe wrong source is selected.\n\nSearch results are outdated.\n\nThe source discusses a similar topic but not the specific claim.\n\nImportant context is missing from the retrieved material.\n\n3. Misinterpretation of sources\n\nEven when a source is genuine, AI may:\n\nMisunderstand its findings.\n\nTake statements out of context.\n\nConfuse correlation with causation.\n\nOverstate conclusions beyond what the research supports.\n\n4. Incorrect citation matching\n\nSometimes an AI identifies a legitimate source but attaches it to the wrong statement.\n\nFor example:\n\nClaim: A study found that sleep improves academic performance.\n\nCitation: A genuine study about sleep duration and mental health.\n\nThe citation exists, but it does not directly support the claim.\n\n5. Outdated information\n\nAI may cite sources that were once accurate but have since become outdated, particularly in rapidly changing fields such as medicine, technology, law and economics.\n\n6. Weak verification processes\n\nSome AI systems generate citations without checking:\n\nWhether the source actually exists.\n\nWhether the link works.\n\nWhether the cited passage supports the claim.\n\nWhether the source is credible and authoritative.\n\n7. Limitations in AI training data\n\nAI models learn from large collections of text that may contain:\n\nIncorrect information.\n\nInconsistent referencing styles.\n\nIncomplete bibliographic details.\n\nDuplicate or poorly attributed material.\n\nThese limitations can contribute to inaccurate references.\n\nHow to reduce inaccurate AI citations\n\nBest practice\n\n\t\n\nWhy it helps\n\n\n\n\nVerify the original source\n\n\t\n\nConfirms the reference exists\n\n\n\n\nCheck the cited passage\n\n\t\n\nEnsures it supports the claim\n\n\n\n\nUse DOI links\n\n\t\n\nHelps identify academic publications\n\n\n\n\nPrioritise primary research\n\n\t\n\nReduces reliance on second-hand interpretations\n\n\n\n\nCheck publication dates\n\n\t\n\nAvoids outdated evidence\n\n\n\n\nCross-reference claims\n\n\t\n\nIdentifies inconsistencies\n\n\n\n\nUse human review\n\n\t\n\nCatches errors automated systems miss\n\nKey takeaway\n\nThe central problem is the difference between producing a plausible citation and producing a verified citation.\n\nAn AI can generate a reference that looks academically credible without establishing that the source exists or supports the statement.\n\nFor academic research, citations should therefore be treated as claims requiring independent verification, rather than assumed to be reliable simply because an AI generated them.",
      "citations": [],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 0 sources and 0 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-inaccurate-ai-citations",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-inaccurate-ai-citations",
      "published_at": "2026-11-15T13:57:00.000Z",
      "updated_at": "2026-11-15T13:57:00.000Z",
      "revision": 1,
      "slug": "chatgpt-inaccurate-ai-citations",
      "panelId": "inaccurate-ai-citations",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-inaccurate-ai-citations",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "What causes inaccurate AI citations?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:57 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9ba2f-9b88-83e9-97ef-b99e7332cdac?temporary-chat=true",
      "mentions": "Not observed",
      "summary": "Inaccurate AI citations happen when an AI system provides sources that don't correctly support the claims it makes, cite the wrong information, or even generate references that do not exist.",
      "answerSnapshot": "：\n\nInaccurate AI citations happen when an AI system provides sources that don't correctly support the claims it makes, cite the wrong information, or even generate references that do not exist.\n\nThere are several common causes:\n\n1. Hallucinations\n\nAI models can generate plausible-looking but fictional references, including:\n\nNon-existent academic papers.\n\nIncorrect author names or publication dates.\n\nMade-up journal titles, URLs, or DOIs.\n\nThis happens because AI predicts likely sequences of text rather than inherently verifying every reference.\n\n2. Poor source retrieval\n\nAI may retrieve sources that are:\n\nOutdated or irrelevant.\n\nOnly loosely related to the question.\n\nBased on unreliable or misleading information.\n\nMissing important context.\n\nEven if a source is real, it might not support the specific claim being cited.\n\n3. Misinterpretation of sources\n\nAI can misunderstand the information in a genuine source.\n\nFor example:\n\nConfusing correlation with causation.\n\nMisrepresenting an author's conclusion.\n\nTaking a statement out of context.\n\nOvergeneralising findings from a limited study.\n\n4. Incorrect citation matching\n\nAI may find a legitimate source but attach it to the wrong statement.\n\nFor example, a research paper might discuss the effects of social media on adults, but the AI cites it to support a claim about teenagers.\n\n5. Outdated training data\n\nAI may rely on information learned during training that is no longer accurate.\n\nThis can result in:\n\nOutdated statistics.\n\nSuperseded research.\n\nIncorrect publication details.\n\nBroken or changed URLs.\n\n6. Ambiguous prompts\n\nVague questions can lead AI to make assumptions about which sources are relevant.\n\nFor example, asking for \"research on AI in education\" could produce citations about very different topics, populations, and educational settings.\n\n7. Lack of citation verification\n\nSome AI systems generate citations without checking whether:\n\nThe source actually exists.\n\nThe cited information appears in the source.\n\nThe source supports the claim.\n\nThe quotation or page number is accurate.\n\nHow can inaccurate AI citations be prevented?\n\nSolution\n\n\t\n\nHow it helps\n\n\n\n\nSource verification\n\n\t\n\nConfirms that references exist\n\n\n\n\nCross-checking\n\n\t\n\nCompares claims against original sources\n\n\n\n\nReliable databases\n\n\t\n\nUses Google Scholar, PubMed or academic journals\n\n\n\n\nClaim-level citations\n\n\t\n\nLinks individual claims to supporting evidence\n\n\n\n\nHuman review\n\n\t\n\nIdentifies misleading interpretations\n\n\n\n\nUp-to-date retrieval\n\n\t\n\nReduces reliance on outdated information\n\nKey takeaway\n\nThe fundamental problem is that generating a convincing citation is not the same as verifying a citation.\n\nAI citation errors generally arise from a combination of hallucination, poor information retrieval, source misinterpretation, and insufficient verification.\n\nAn important distinction is that a citation can be:\n\nFabricated – the source does not exist.\n\nInaccurate – the source exists, but its bibliographic details are wrong.\n\nMisattributed – the source exists but does not support the attached claim.\n\nMisleading – the source supports the claim only partially or without the context presented.\n\nThese distinctions are particularly useful when researching the reliability of AI-generated academic references.",
      "citations": [],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 0 sources and 0 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-source-change-answer-drift",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-source-change-answer-drift",
      "published_at": "2026-11-15T13:56:00.000Z",
      "updated_at": "2026-11-15T13:56:00.000Z",
      "revision": 1,
      "slug": "chatgpt-source-change-answer-drift",
      "panelId": "source-change-answer-drift",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-source-change-answer-drift",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "How do source changes affect future AI citations?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:56 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b9f8-beb0-83ea-8ea1-b4462fb70cfd?temporary-chat=true",
      "mentions": "Not observed",
      "summary": "Source changes can significantly affect future AI citations because AI systems may retrieve, evaluate, and cite updated versions of web pages rather than relying permanently on the version they originally encountered.",
      "answerSnapshot": "：\n\nSource changes can significantly affect future AI citations because AI systems may retrieve, evaluate, and cite updated versions of web pages rather than relying permanently on the version they originally encountered.\n\nThe key principle is: changing a source today can change what AI systems cite tomorrow, but the effect depends on when they discover and process the changes.\n\n1. How different source changes affect AI citations\n\nSource change\n\n\t\n\nPotential effect on future AI citations\n\n\n\n\nUpdating facts or statistics\n\n\t\n\nAI may cite the updated information.\n\n\n\n\nAdding new research or evidence\n\n\t\n\nThe source may become more useful and relevant.\n\n\n\n\nImproving clarity and structure\n\n\t\n\nAI may find relevant information easier to extract.\n\n\n\n\nChanging the page title\n\n\t\n\nMay affect how AI identifies and selects the page.\n\n\n\n\nChanging the URL\n\n\t\n\nExisting citations may break unless redirects are configured.\n\n\n\n\nRemoving content\n\n\t\n\nThe source may become less relevant or disappear from future citations.\n\n\n\n\nAdding authoritative references\n\n\t\n\nMay strengthen the source's credibility.\n\n\n\n\nCorrecting outdated information\n\n\t\n\nMay improve its usefulness for future answers.\n\n2. How quickly do AI citations respond to changes?\n\nThere are several stages:\n\nSource updated\n\nThe website owner changes the original content.\n\nCrawling\n\nSearch engines or AI search systems discover the changes.\n\nIndexing and retrieval\n\nThe revised content becomes available to search systems.\n\nCitation selection\n\nAI systems may select the revised page for relevant questions.\n\nFuture answers\n\nThe updated information may appear in generated answers and citations.\n\nA 2026 study by Parse found a median interval of 15.1 days between crawls of pages that were crawled repeatedly. This illustrates why changes aren't necessarily reflected immediately. \nParse\n+1\n\n3. What makes a source more likely to retain citations?\n\nConsider these factors:\n\nAccuracy: Information is verifiable and factually correct.\n\nFreshness: Statistics and claims reflect current evidence.\n\nOriginality: The source contributes useful information rather than simply repeating others.\n\nAuthority: The author and publisher have relevant expertise.\n\nConsistency: Information agrees with reliable independent sources.\n\nAccessibility: AI search crawlers can access the relevant pages.\n\nImportantly, no single factor guarantees an AI citation. Different AI platforms use different retrieval and source-selection systems. \nNamedrop\n+1\n\n4. How to measure the impact of source changes\n\nIf you're managing a website or researching AI visibility, track:\n\nMetric\n\n\t\n\nWhat it tells you\n\n\n\n\nCitation frequency\n\n\t\n\nHow often AI cites the source\n\n\n\n\nCitation accuracy\n\n\t\n\nWhether AI represents the source correctly\n\n\n\n\nCitation retention\n\n\t\n\nWhether existing citations continue\n\n\n\n\nNew citations\n\n\t\n\nWhether updated content attracts citations\n\n\n\n\nSource substitution\n\n\t\n\nWhether AI switches to another source\n\n\n\n\nReferral traffic\n\n\t\n\nWhether citations generate visitors\n\nCompare results before and after a source change, using the same questions across multiple AI platforms.\n\n5. The important distinction\n\nThere are two different types of AI knowledge:\n\nTraining knowledge: Information learned during model training. Editing a website does not automatically change an already-trained model.\n\nRetrieved knowledge: Information fetched from websites during an AI search. This can reflect website changes once the relevant systems discover and retrieve them.\n\nThis distinction is particularly important when discussing ChatGPT Search, Google AI features and other AI-powered search products. \nThe SEO Handbook\n+1\n\nBottom line: Source changes influence future AI citations through discovery, indexing, retrieval and source selection. Improving a source can increase its usefulness, but it does not guarantee more citations.",
      "citations": [],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 0 sources and 0 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-source-change-answer-drift-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-source-change-answer-drift-repeat-2",
      "published_at": "2026-11-15T13:56:00.000Z",
      "updated_at": "2026-11-15T13:56:00.000Z",
      "revision": 1,
      "slug": "chatgpt-source-change-answer-drift-repeat-2",
      "panelId": "source-change-answer-drift",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-source-change-answer-drift",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "How do source changes affect future AI citations?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:56 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9ba08-21f4-83ea-908e-3a93e67f9eaa?temporary-chat=true",
      "mentions": "Cited",
      "summary": "Changes to a source can significantly affect whether, how often, and in what context AI systems cite it in future answers.",
      "answerSnapshot": "：\n\nChanges to a source can significantly affect whether, how often, and in what context AI systems cite it in future answers.\n\nThe key idea is that AI citations are based on the information available to the system when it retrieves or generates an answer, not necessarily on what a source said when it was first published.\n\nFor example, if you update a webpage, research paper, or online article, future AI-generated answers may reflect the updated version once the system discovers and indexes the changes.\n\n1. What happens when a source changes?\n\nType of source change\n\n\t\n\nPotential effect on future AI citations\n\n\n\n\nCorrecting factual errors\n\n\t\n\nAI may retrieve and cite the corrected information\n\n\n\n\nAdding new research or data\n\n\t\n\nThe source may become relevant to more questions\n\n\n\n\nUpdating outdated information\n\n\t\n\nFuture answers may reflect newer facts\n\n\n\n\nChanging the page title\n\n\t\n\nThe source may match different search queries\n\n\n\n\nRemoving important content\n\n\t\n\nThe source may become less useful or stop being cited\n\n\n\n\nDeleting the page\n\n\t\n\nExisting citations may disappear once systems detect the removal\n\n\n\n\nChanging the URL\n\n\t\n\nThe source may become harder to retrieve unless redirects are maintained\n\n2. Why don't citations change immediately?\n\nThere are several stages:\n\nSource update: The publisher changes the original content.\n\nCrawling: Search systems discover the changes.\n\nIndexing: The updated content becomes available in a search index.\n\nRetrieval: An AI system finds the updated source when answering a question.\n\nCitation: The AI may cite that source in a future response.\n\nThese stages don't necessarily happen simultaneously. Google, for example, says recrawling can take days or weeks and does not guarantee immediate indexing. \nGoogle for Developers\n+1\n\n3. Does updating a source guarantee more AI citations?\n\nNo. This is an important distinction.\n\nAn updated source might:\n\nGain citations because it contains more relevant information.\n\nLose citations because important supporting information was removed.\n\nContinue receiving the same citations.\n\nBe replaced by another, more relevant source.\n\nAI citation patterns also depend on the question, retrieval systems, source credibility and competing information.\n\n4. What should publishers monitor?\n\nIf the goal is to understand how source changes influence AI visibility, monitor:\n\nCitation frequency: How often AI systems cite the source.\n\nCitation accuracy: Whether the AI represents the source correctly.\n\nCitation position: Whether the source appears prominently.\n\nContent freshness: Whether AI answers reflect recent updates.\n\nCitation persistence: Whether the source continues to be cited over time.\n\n5. The bigger picture\n\nThe most important distinction is between two things:\n\nHistorical citations: What an AI system cited before a source changed.\n\nFuture citations: What an AI system retrieves and cites after discovering the updated source.\n\nA source change does not automatically rewrite historical AI answers or citations. Its influence is primarily on future retrieval and responses.\n\nIn short: Source changes can reshape future AI citations, but the effect depends on discovery, indexing, relevance and retrieval—not simply on how recently a page was edited.\n\nFor further reading: Google's crawling and indexing guidance.",
      "citations": [
        {
          "title": "Google's crawling and indexing guidance",
          "host": "developers.google.com",
          "url": "https://developers.google.com/search/help/crawling-index-faq?hl=en&utm_source=chatgpt.com",
          "position": 1
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 1 sources and 1 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-source-change-answer-drift-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-source-change-answer-drift-repeat-3",
      "published_at": "2026-11-15T13:56:00.000Z",
      "updated_at": "2026-11-15T13:56:00.000Z",
      "revision": 1,
      "slug": "chatgpt-source-change-answer-drift-repeat-3",
      "panelId": "source-change-answer-drift",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-source-change-answer-drift",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "How do source changes affect future AI citations?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:56 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9ba1e-2eac-83ea-8970-caf76a5457d2?temporary-chat=true",
      "mentions": "Cited",
      "summary": "Changes to a source can significantly affect how AI systems cite it in future answers. The key point is that AI citations depend on the information available to the AI when it retrieves or processes sources, not necessarily on what a source contained when it was first published.",
      "answerSnapshot": "：\n\nChanges to a source can significantly affect how AI systems cite it in future answers. The key point is that AI citations depend on the information available to the AI when it retrieves or processes sources, not necessarily on what a source contained when it was first published.\n\nFor example, if you update a website article, research paper, or documentation page, future AI-generated answers may reflect the revised information.\n\nHow source changes affect AI citations\n\nUpdated information: If a source changes its facts, statistics, or explanations, future citations may reflect the new version.\n\nChanged URLs: If a page moves or its URL changes, AI systems may stop finding the original source unless redirects and indexing are handled properly.\n\nRemoved content: If information is deleted, future AI responses may no longer cite that page for the removed information.\n\nImproved credibility: Adding references, evidence, author details, and publication dates can make a source more useful and trustworthy.\n\nConflicting information: If a source changes and contradicts other sources, AI systems may select different sources or present conflicting interpretations.\n\nWhat happens in practice?\n\nRecent research suggests that the relationship is more nuanced than simply updating a page and receiving more citations.\n\nFor example, a 2026 study examining more than 2.2 million AI-cited pages found:\n\nAI systems do not necessarily stop citing a page just because its content changes.\n\nAfter a substantive update, 33.4% of examined citation pairs continued citing the same page.\n\nWhen a page became permanently unavailable, only 13.6% of examined citation pairs cited it again within the following 60 days. \nParse\n\nHow to make sources more useful for future AI citations\n\nIf you're managing a website or publishing content, consider these practices:\n\nAction\n\n\t\n\nPotential benefit\n\n\n\n\nCorrect outdated facts\n\n\t\n\nImproves accuracy\n\n\n\n\nAdd recent evidence\n\n\t\n\nMakes claims easier to verify\n\n\n\n\nInclude authoritative references\n\n\t\n\nStrengthens credibility\n\n\n\n\nMaintain stable URLs\n\n\t\n\nPreserves discoverability\n\n\n\n\nUpdate publication metadata accurately\n\n\t\n\nHelps communicate content freshness\n\n\n\n\nRemove obsolete information\n\n\t\n\nReduces conflicting information\n\nThe important distinction\n\nThere are three separate stages:\n\nSource change: The original publisher updates the content.\n\nDiscovery: Search engines and AI retrieval systems find the updated version.\n\nCitation: An AI system retrieves that version and chooses to cite it in an answer.\n\nThese stages don't happen simultaneously. A change to a source does not guarantee an immediate change in AI citations.\n\nBottom line: Updating sources can influence future AI citations, but accuracy, relevance, credibility, accessibility and successful re-indexing matter more than simply changing a publication date.\n\nFurther reading: Research on how quickly AI systems notice website changes.",
      "citations": [
        {
          "title": "Research on how quickly AI systems notice website changes",
          "host": "parse.gl",
          "url": "https://parse.gl/research/how-long-ai-takes-to-notice-website-changes?utm_source=chatgpt.com",
          "position": 1
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 1 sources and 1 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-track-citations-over-time-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-track-citations-over-time-repeat-2",
      "published_at": "2026-11-15T13:55:00.000Z",
      "updated_at": "2026-11-15T13:55:00.000Z",
      "revision": 1,
      "slug": "chatgpt-track-citations-over-time-repeat-2",
      "panelId": "track-citations-over-time",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-track-citations-over-time",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "How should a team track AI citations over time?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:55 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b9aa-6c88-83ea-b9dc-8ff2fe8ae0cf?temporary-chat=true",
      "mentions": "Not observed",
      "summary": "A team should track AI citations as an ongoing visibility and authority measurement programme, rather than simply counting how often its brand appears in ChatGPT or other AI answers.",
      "answerSnapshot": "：\n\nA team should track AI citations as an ongoing visibility and authority measurement programme, rather than simply counting how often its brand appears in ChatGPT or other AI answers.\n\nThe goal is to understand three things:\n\nVisibility: Are AI platforms mentioning and citing your brand?\n\nAuthority: Are they using your website as a trusted source?\n\nConsistency: Is your visibility improving, declining, or changing across AI platforms over time?\n\nHere's a practical framework.\n\n1. Establish a baseline\n\nStart by identifying the AI platforms you want to monitor.\n\nPlatform\n\n\t\n\nWhat to track\n\n\n\n\nChatGPT\n\n\t\n\nBrand mentions and linked sources\n\n\n\n\nGoogle AI Overviews\n\n\t\n\nCitations and source URLs\n\n\n\n\nGoogle AI Mode\n\n\t\n\nBrand recommendations and citations\n\n\n\n\nPerplexity\n\n\t\n\nReferenced domains and pages\n\n\n\n\nMicrosoft Copilot\n\n\t\n\nBrand mentions and sources\n\n\n\n\nGemini\n\n\t\n\nBrand mentions and supporting sources\n\nCreate a fixed set of 30–100 prompts representing the questions your customers actually ask.\n\nFor example:\n\nWhat are the best [product category] providers?\n\nWhich companies specialise in [service]?\n\nHow does [your product] compare with competitors?\n\nWhat should I consider when choosing [solution]?\n\nWhat are the leading alternatives to [competitor]?\n\nKeep these prompts consistent between reporting periods.\n\n2. Track the right metrics\n\nI recommend five core metrics.\n\nAI citation tracking dashboard\n\n1. AI mention rate\n\nVisibility\n\nPercentage of monitored responses that mention your brand.\n\nResponses mentioning brand\nTotal responses\n×\n100\nTotal responses\nResponses mentioning brand\n\t​\n\n×100\n\n2. Citation rate\n\nAuthority\nPercentage of responses citing your website or content.\n\n3. Share of AI voice\n\nCompetition\nYour brand's share of total brand mentions across monitored responses.\n\n4. Citation diversity\n\nCoverage\nNumber of unique pages and domains cited as sources.\n\n5. Citation accuracy\n\nQuality\nWhether AI platforms accurately describe your brand, products and expertise.\n\nOne important distinction: a brand mention isn't necessarily a citation. Track them separately.\n\n3. Build a historical tracking database\n\nRecord every monitoring run, rather than overwriting previous results.\n\nA useful data structure:\n\nField\n\n\t\n\nExample\n\n\n\n\nDate checked\n\n\t\n\n15 November 2026\n\n\n\n\nAI platform\n\n\t\n\nChatGPT\n\n\n\n\nPrompt\n\n\t\n\nBest CRM platforms for SMEs\n\n\n\n\nBrand mentioned\n\n\t\n\nYes\n\n\n\n\nBrand recommended\n\n\t\n\nYes\n\n\n\n\nWebsite cited\n\n\t\n\nYes\n\n\n\n\nCited URL\n\n\t\n\n/crm-platform-guide\n\n\n\n\nCompetitors mentioned\n\n\t\n\nCompetitor A, B\n\n\n\n\nPosition\n\n\t\n\n2\n\n\n\n\nAccuracy\n\n\t\n\nAccurate\n\nThis gives you an auditable record of how AI visibility evolves.\n\nI would organise the data into four levels:\n\nPlatform: ChatGPT, Gemini, Perplexity, etc.\n\nPrompt: the question asked.\n\nBrand: yours and your competitors.\n\nSource: the specific URL or domain cited.\n\n4. Report trends monthly\n\nAvoid relying on individual snapshots. AI-generated answers can vary between runs.\n\nA monthly report should show:\n\nExample monthly reporting layout\n\nBrand mention rate\n\n42%\n↑ 6 percentage points\n\nCitation rate\n\n28%\n↑ 4 percentage points\n\nShare of AI voice\n\n18%\n↑ 3 percentage points\n\nUnique cited pages\n\n34\n↑ 8 pages\n\nIllustrative data, not actual company performance.\n\nCitation rate over six months\n\n0%\n10%\n20%\n30%\n40%\nJun\nJul\nAug\nSep\nOct\nNov\nJun: Citation rate 12%\nJul: Citation rate 15%\nAug: Citation rate 18%\nSep: Citation rate 21%\nOct: Citation rate 24%\nNov: Citation rate 28%\n\nLook for:\n\nWhich prompts are generating more citations?\n\nWhich pages are repeatedly cited?\n\nWhich competitors are gaining visibility?\n\nWhich AI platforms are underperforming?\n\nWhat content improvements correlate with increased citations?\n\n5. Identify which content earns citations\n\nThis is where tracking becomes actionable.\n\nAnalyse your cited URLs and group them by content type.\n\nContent type\n\n\t\n\nWhat to investigate\n\n\n\n\nResearch and statistics\n\n\t\n\nOriginal data and evidence\n\n\n\n\nGuides\n\n\t\n\nDepth and usefulness\n\n\n\n\nComparison pages\n\n\t\n\nClear, objective comparisons\n\n\n\n\nProduct pages\n\n\t\n\nSpecific features and facts\n\n\n\n\nCase studies\n\n\t\n\nDemonstrable results\n\n\n\n\nFAQs\n\n\t\n\nDirect answers to common questions\n\nBuild a list of:\n\nMost frequently cited pages.\n\nPages losing citations.\n\nImportant pages never cited.\n\nThird-party websites that AI platforms frequently reference.\n\nUse those insights to prioritise content improvements.\n\n6. Choose a monitoring tool or workflow\n\nThere are three practical approaches:\n\nApproach\n\n\t\n\nBest suited to\n\n\n\n\nManual prompt testing + spreadsheet\n\n\t\n\nSmall teams and initial baselines\n\n\n\n\nAI visibility monitoring software\n\n\t\n\nRegular, multi-platform reporting\n\n\n\n\nCustom API-based monitoring\n\n\t\n\nLarger teams and bespoke analytics\n\nTools worth evaluating include:\n\nProfound – AI search visibility and citation analytics.\n\nPeec AI – AI visibility and source monitoring.\n\nOtterly.AI – AI search mentions and citations.\n\nSE Ranking – SEO and AI visibility monitoring.\n\nEvaluate platforms on their ability to retain historical data, track individual URLs, compare competitors and export results.\n\n7. My recommended operating rhythm\n\nFrequency\n\n\t\n\nActivity\n\n\n\n\nWeekly\n\n\t\n\nRun fixed prompts and capture citations\n\n\n\n\nMonthly\n\n\t\n\nAnalyse visibility and competitor movements\n\n\n\n\nQuarterly\n\n\t\n\nReview content gaps and update strategy\n\n\n\n\nHalf-yearly\n\n\t\n\nReassess prompts, platforms and KPIs\n\nOne important measurement rule: keep prompts, platform settings, geography and collection methodology consistent. Record changes in model versions where possible. Otherwise, you risk interpreting changes in AI behaviour as changes in your marketing performance.\n\nThe key recommendation\n\nBuild an AI citation tracker around this simple relationship:\n\nPrompts\n\nWhat customers ask AI platforms\n\nAI responses\n\nMentions, recommendations and citations\n\nCitation database\n\nPlatforms, prompts, brands, URLs and timestamps\n\nHistorical analytics\n\nVisibility, authority, competitors and trends\n\nContent optimisation\n\nImprove the pages and sources AI systems rely on\n\nThe real objective isn't simply to get more AI citations. It's to become a consistently trusted and referenced source when potential customers research your category.\n\nA useful next step: I can help design a practical AI citation tracking spreadsheet and dashboard, including the metrics, database structure, monthly reporting and competitor benchmarking.",
      "citations": [],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 0 sources and 0 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-track-citations-over-time-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-track-citations-over-time-repeat-3",
      "published_at": "2026-11-15T13:55:00.000Z",
      "updated_at": "2026-11-15T13:55:00.000Z",
      "revision": 1,
      "slug": "chatgpt-track-citations-over-time-repeat-3",
      "panelId": "track-citations-over-time",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-track-citations-over-time",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "How should a team track AI citations over time?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:55 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b9d1-03e4-83e9-82e8-69e5b75b7b12?temporary-chat=true",
      "mentions": "Not observed",
      "summary": "A team should track AI citations as an ongoing visibility and reputation metric, much like traditional SEO rankings. The goal is to understand when, where, how often, and why AI platforms mention or recommend your brand.",
      "answerSnapshot": "：\n\nA team should track AI citations as an ongoing visibility and reputation metric, much like traditional SEO rankings. The goal is to understand when, where, how often, and why AI platforms mention or recommend your brand.\n\nThis is increasingly important as people use tools such as ChatGPT, Perplexity, Google AI Overviews and Gemini to discover products, services and information.\n\n1. Establish what you want to measure\n\nI recommend tracking six core metrics.\n\nMetric\n\n\t\n\nWhat it tells you\n\n\n\n\nCitation frequency\n\n\t\n\nHow often AI platforms cite your website\n\n\n\n\nBrand mentions\n\n\t\n\nHow often your brand appears in AI answers\n\n\n\n\nCitation share\n\n\t\n\nYour share of citations compared with competitors\n\n\n\n\nSource URLs\n\n\t\n\nWhich pages AI systems reference\n\n\n\n\nCitation accuracy\n\n\t\n\nWhether information attributed to you is correct\n\n\n\n\nSentiment/context\n\n\t\n\nWhether your brand is recommended, neutrally mentioned or criticised\n\nOne important distinction: a brand mention is not necessarily a citation. Track them separately.\n\n2. Monitor multiple AI platforms\n\nChatGPT\n\nTrack\n\nGoogle Gemini\n\nTrack\n\nPerplexity\n\nTrack\n\nGoogle AI Overviews\n\nTrack\n\nMicrosoft Copilot\n\nTrack\n\nDon't assume that visibility on one platform means visibility on another. Each can produce different answers and cite different sources.\n\n3. Build a consistent prompt-tracking library\n\nCreate a fixed set of questions that represent how your target audience searches.\n\nFor example, a software company might track:\n\nWhat are the best project management platforms?\n\nWhich project management tools are best for small businesses?\n\nWhat are the alternatives to [competitor]?\n\nWhich platform offers the best value?\n\nWhat should businesses consider when choosing project management software?\n\nOrganise prompts into categories:\n\nBrand-specific searches\n\nCategory searches\n\nComparison searches\n\nProblem-solving searches\n\nIndustry and informational searches\n\nKeep the core prompts consistent month to month so you can identify meaningful trends.\n\n4. Create a central citation database\n\nA spreadsheet or dashboard should capture the following information for every observation.\n\nField\n\n\t\n\nExample\n\n\n\n\nDate checked\n\n\t\n\n15 November 2026\n\n\n\n\nAI platform\n\n\t\n\nChatGPT\n\n\n\n\nPrompt\n\n\t\n\nBest project management tools\n\n\n\n\nBrand mentioned\n\n\t\n\nYes\n\n\n\n\nWebsite cited\n\n\t\n\nexample.com\n\n\n\n\nCitation URL\n\n\t\n\n/product/features\n\n\n\n\nCompetitor cited\n\n\t\n\nCompetitor A\n\n\n\n\nPosition\n\n\t\n\nFirst recommendation\n\n\n\n\nContext\n\n\t\n\nPositive\n\n\n\n\nAccuracy\n\n\t\n\nCorrect\n\nAlso record the model or search mode where available, because AI answers can change depending on the configuration.\n\n5. Build a monthly performance dashboard\n\nI would use five main dashboard sections.\n\nAI citation performance\n\nIllustrative dashboard — sample figures, not actual results\n\n245\n\nTotal citations\n\n+18% month-on-month\n\n68%\n\nBrand visibility\n\n+6 percentage points\n\n42\n\nUnique cited URLs\n\n12\n\nCompetitors monitored\n\nCitation trend\n\nIllustrative monthly citation counts.\n\n90\n135\n180\n225\n270\nJun\nJul\nAug\nSep\nOct\nNov\nJun: Citations 120\nJul: Citations 145\nAug: Citations 138\nSep: Citations 175\nOct: Citations 207\nNov: Citations 245\n\nPlatform breakdown\n\nIllustrative citations by AI platform.\n\n0\n25\n50\n75\n100\nChatGPT\nPerplexity\nGemini\nGoogle AI\nChatGPT: Citations 95\nPerplexity: Citations 65\nGemini: Citations 50\nGoogle AI: Citations 35\n6. Establish a measurement cadence\n\nA practical operating rhythm:\n\nFrequency\n\n\t\n\nActivity\n\n\n\n\nWeekly\n\n\t\n\nRun core prompts and record citations\n\n\n\n\nMonthly\n\n\t\n\nAnalyse visibility and competitor movements\n\n\n\n\nQuarterly\n\n\t\n\nReview content gaps and citation quality\n\n\n\n\nEvery 6 months\n\n\t\n\nReassess prompts, platforms and KPIs\n\nAvoid drawing conclusions from one-off AI responses. AI outputs can fluctuate, so use repeated observations and consistent methodology.\n\n7. Turn insights into actions\n\nThe real value is understanding what drives citations.\n\nFor example:\n\nFinding\n\n\t\n\nRecommended action\n\n\n\n\nCompetitors cited more frequently\n\n\t\n\nAnalyse their content and authority\n\n\n\n\nProduct pages rarely cited\n\n\t\n\nImprove clarity and supporting information\n\n\n\n\nThird-party websites cited more often\n\n\t\n\nDevelop relevant industry coverage\n\n\n\n\nOutdated information cited\n\n\t\n\nUpdate factual content\n\n\n\n\nBrand mentioned but not linked\n\n\t\n\nImprove discoverability and source attribution\n\n8. Consider dedicated tracking tools\n\nSome platforms worth evaluating include:\n\nProfound – AI search visibility and citation monitoring.\n\nPeec AI – brand visibility and competitor monitoring.\n\nOtterly.AI – AI search mentions and citations.\n\nSemrush – SEO and AI visibility capabilities.\n\nCheck their current platform coverage, reporting frequency and export functionality before choosing.\n\nMy recommended approach\n\nI would structure the programme around three levels:\n\nMeasurement: How often are we cited, and on which platforms?\n\nAnalysis: Which prompts, topics and pages generate citations?\n\nImprovement: What content and authority-building activities will increase relevant citations?\n\nThe most important KPI shouldn't simply be total citations. I'd prioritise citation share of voice across strategically important prompts, supported by citation accuracy and brand visibility.\n\nOne question that would help tailor the framework:\n\nAre you looking to track AI citations for:\n\nYour own company's website and brand\n\nAn SEO or digital marketing agency managing multiple clients\n\nA corporate marketing team monitoring competitors\n\nAn AI search visibility reporting product\n\nThe answer changes how I'd design the tracking database, KPIs and reporting dashboard.",
      "citations": [],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 0 sources and 0 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-product-page-evidence-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-product-page-evidence-repeat-3",
      "published_at": "2026-11-15T13:54:00.000Z",
      "updated_at": "2026-11-15T13:54:00.000Z",
      "revision": 1,
      "slug": "chatgpt-product-page-evidence-repeat-3",
      "panelId": "product-page-evidence",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-product-page-evidence",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "What evidence makes a product page citable by AI search?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:54 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b978-8b48-83e9-87a3-fbd509988a39?temporary-chat=true",
      "mentions": "Not observed",
      "summary": "A product page becomes citable by AI search when it provides specific, verifiable, clearly attributable information that an AI system can retrieve and use to support an answer.",
      "answerSnapshot": "：\n\nA product page becomes citable by AI search when it provides specific, verifiable, clearly attributable information that an AI system can retrieve and use to support an answer.\n\nThe key distinction is that a page being crawlable is not the same as being worth citing.\n\nFor product pages, I'd look at six types of evidence.\n\n1. Product facts that can be independently verified\nHigh priority\n\nAI systems need concrete facts rather than vague marketing claims.\n\nExamples:\n\nExact product name, model and manufacturer.\n\nTechnical specifications and dimensions.\n\nMaterials, ingredients or components, where relevant.\n\nCompatibility and supported features.\n\nPricing and availability, with clear dates or context.\n\nWarranty, certifications and compliance information.\n\nWeak evidence:\n\"Our product offers outstanding performance.\"\n\nStrong evidence:\n\"The product supports Bluetooth 5.3, has a 40-hour stated battery life and weighs 280g.\"\n\nThe second statement is much easier to extract, compare and reference.\n\n2. Original evidence and first-hand information\nHigh priority\n\nThis is one of the strongest opportunities to differentiate a product page.\n\nExamples include:\n\nOriginal product testing.\n\nMeasured performance results.\n\nTest methodology and conditions.\n\nOriginal product photography.\n\nDemonstrations and technical documentation.\n\nComparisons against clearly identified alternatives.\n\nResearch conducted by the manufacturer.\n\nGoogle explicitly recommends original information, research, analysis and demonstrable effort rather than content that simply repeats existing information. \nGoogle for Developers\n\nA useful test:\n\nDoes this page contain information that another retailer could not simply copy from the manufacturer's product feed?\n\nIf yes, it has potentially valuable citation material.\n\n3. Independent corroboration\n\nAI systems can encounter competing claims from different sources.\n\nA product page becomes more credible when its claims are supported by:\n\nIndependent product reviews.\n\nRecognised certification bodies.\n\nPublished test results.\n\nTechnical documentation.\n\nCustomer experiences.\n\nThird-party comparisons.\n\nReputable industry publications.\n\nFor example, a manufacturer claiming a product is durable is one thing. A published independent durability test is additional evidence.\n\nThe distinction is between what the brand claims and what other sources substantiate.\n\n4. Structured, machine-readable product information\nTechnical foundation\n\nUse Schema.org structured data, particularly:\n\nSchema property/type\n\n\t\n\nWhat it communicates\n\n\n\n\nProduct\n\n\t\n\nProduct identity\n\n\n\n\nBrand\n\n\t\n\nManufacturer or brand\n\n\n\n\nOffer\n\n\t\n\nPrice and availability\n\n\n\n\nAggregateRating\n\n\t\n\nOverall review rating\n\n\n\n\nReview\n\n\t\n\nIndividual review information\n\n\n\n\nSKU / GTIN\n\n\t\n\nProduct identification\n\n\n\n\nImage\n\n\t\n\nProduct imagery\n\nAlso ensure that structured data agrees with the information visible to customers.\n\nImportant qualification: schema makes information easier to interpret, but it does not guarantee AI citations. Research into schema and AI citation effects remains mixed. \naixiv.science\n+1\n\n5. Content that directly answers buying questions\n\nA strong product page should answer questions that people actually ask AI search engines.\n\nFor example:\n\nCustomer question\n\n\t\n\nEvidence needed\n\n\n\n\nWhat does it do?\n\n\t\n\nClear product description\n\n\n\n\nHow does it compare?\n\n\t\n\nComparative specifications\n\n\n\n\nWho is it suitable for?\n\n\t\n\nSupported use cases\n\n\n\n\nIs it compatible with X?\n\n\t\n\nCompatibility documentation\n\n\n\n\nHow much does it cost?\n\n\t\n\nCurrent, visible pricing\n\n\n\n\nIs it reliable?\n\n\t\n\nTesting and independent reviews\n\n\n\n\nWhat are its limitations?\n\n\t\n\nHonest limitations and exclusions\n\nThis matters because recent research suggests that matching the language and meaning of users' questions is important to whether pages are retrieved and cited. \narXiv\n\n6. Trust, provenance and freshness\n\nAI systems need to be able to establish where information comes from and whether it is current.\n\nLook for:\n\nNamed manufacturer or publisher.\n\nClear company contact and About information.\n\nIdentifiable authors or technical reviewers where appropriate.\n\nSources linked to supporting claims.\n\nPublication or update dates.\n\nCurrent stock and pricing information.\n\nConsistent product information across the website.\n\nGoogle's guidance specifically highlights trustworthiness, expertise, accuracy and clear sourcing. \nGoogle for Developers\n\nA practical AI citation-readiness scorecard\n\nThis is a proposed auditing framework, not an established search-engine ranking formula.\n\nAssess a product page\n\nScore each area from 0 (missing) to 5 (excellent).\n\nProduct identity and specifications\n\n0/5 · 20%\n\nOriginal testing and evidence\n\n0/5 · 20%\n\nIndependent corroboration\n\n0/5 · 15%\n\nStructured product data\n\n0/5 · 10%\n\nQuestion-answering content\n\n0/5 · 20%\n\nTrust, provenance and freshness\n\n0/5 · 15%\n\nReadiness score\n\n0 / 100\n\nSignificant evidence gaps\n\nReset assessment\nThe most important distinction: three stages\n\nI would separate AI visibility into three measurable outcomes:\n\nDiscoverability\n\nCan the AI search engine access and retrieve the product page?\n\nEvidence usability\n\nCan it extract specific facts and use them to answer a question accurately?\n\nCitation\n\nDoes it actually reference the product page as a source in its answer?\n\nThese are separate outcomes. Improving one does not guarantee the next.\n\nMy central takeaway: The strongest product pages for AI citation aren't necessarily the pages with the most SEO copy or the most schema markup. They're the pages that provide the clearest, most useful and most independently supportable evidence about a product.\n\nA useful next step would be to turn this into a formal AI Product Page Citation Audit, with measurable criteria for ChatGPT, Google AI Overviews and Perplexity.",
      "citations": [],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 0 sources and 0 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-track-citations-over-time",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-track-citations-over-time",
      "published_at": "2026-11-15T13:54:00.000Z",
      "updated_at": "2026-11-15T13:54:00.000Z",
      "revision": 1,
      "slug": "chatgpt-track-citations-over-time",
      "panelId": "track-citations-over-time",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-track-citations-over-time",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "How should a team track AI citations over time?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:54 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b98a-bf00-83e8-b774-f0fd205a23a4?temporary-chat=true",
      "mentions": "Not observed",
      "summary": "A team should track AI citations as an ongoing visibility and reputation metric, much like traditional SEO rankings. The goal is to understand how often AI platforms mention your brand, whether they cite your website as a source, which competitors appear alongside you, and how those patterns change over time.",
      "answerSnapshot": "：\n\nA team should track AI citations as an ongoing visibility and reputation metric, much like traditional SEO rankings. The goal is to understand how often AI platforms mention your brand, whether they cite your website as a source, which competitors appear alongside you, and how those patterns change over time.\n\nI'd recommend building an AI Citation Tracking Framework with six components.\n\n1. Establish a baseline\n\nStart by identifying the AI platforms and topics that matter to your business.\n\nPlatforms to monitor:\n\nChatGPT\n\nGoogle AI Overviews / AI Mode\n\nPerplexity\n\nMicrosoft Copilot\n\nGemini\n\nClaude\n\nCreate a prompt library covering the questions your target audience might ask.\n\nFor example:\n\nWhat are the best providers of [product category]?\n\nWhich companies specialise in [service]?\n\nWhat is the difference between [Brand A] and [Brand B]?\n\nWhat should I consider when choosing a [product]?\n\nWho are the market leaders in [industry]?\n\nKeep the prompts consistent so you can compare results over time.\n\n2. Track the right metrics\n\nI would use these eight core metrics:\n\nMetric\n\n\t\n\nWhat it measures\n\n\n\n\nAI mention rate\n\n\t\n\nPercentage of prompts mentioning your brand\n\n\n\n\nCitation frequency\n\n\t\n\nHow often your website is cited\n\n\n\n\nShare of AI voice\n\n\t\n\nYour mentions relative to competitors\n\n\n\n\nCitation position\n\n\t\n\nWhere your brand appears in responses\n\n\n\n\nSource diversity\n\n\t\n\nNumber of different URLs cited\n\n\n\n\nCompetitor visibility\n\n\t\n\nHow often competitors are mentioned\n\n\n\n\nCitation accuracy\n\n\t\n\nWhether information about your brand is correct\n\n\n\n\nSentiment\n\n\t\n\nWhether brand references are positive, neutral or negative\n\nAn important distinction: being mentioned is not the same as being cited. Track them separately.\n\n3. Build a central tracking dashboard\n\nA shared dashboard should give marketing, SEO, content and leadership teams one view of performance.\n\nAI visibility dashboard\n\nIllustrative\nAI mention rate\n42%\n+8 percentage points\nCitation frequency\n126\nAcross tracked responses\nShare of AI voice\n28%\nTracked competitors\nUnique cited URLs\n34\nAcross all platforms\n\nCitation trend\n\n21%\n28%\n35%\n42%\n49%\nJun\nJul\nAug\nSep\nOct\nNov\nJun: Your brand 24%\nJul: Your brand 29%\nAug: Your brand 27%\nSep: Your brand 34%\nOct: Your brand 38%\nNov: Your brand 42%\nIllustrative data, not actual platform measurements.\n\nI would also include filters for:\n\nAI platform\n\nTopic or prompt category\n\nCountry or market\n\nCompetitor\n\nLanding page\n\nDate range\n\n4. Establish a consistent monitoring schedule\n\nDifferent reporting frequencies serve different purposes.\n\nFrequency\n\n\t\n\nActivity\n\n\n\n\nDaily\n\n\t\n\nAutomated data collection, where available\n\n\n\n\nWeekly\n\n\t\n\nReview unusual citation changes\n\n\n\n\nMonthly\n\n\t\n\nAnalyse trends and competitor movements\n\n\n\n\nQuarterly\n\n\t\n\nReview strategy and content priorities\n\nOne important consideration: AI answers can vary between runs. Use multiple repeated observations per prompt and compare averages rather than relying on a single response.\n\n5. Turn citation insights into action\n\nTracking is only valuable if it informs decisions.\n\nI'd create a workflow like this:\n\n1. Monitor\n\nCollect AI answers, mentions and cited URLs\n\n2. Analyse\n\nIdentify missing citations, competitors and content gaps\n\n3. Prioritise\n\nIdentify pages and topics with the greatest opportunity\n\n4. Optimise\n\nImprove content quality, evidence, structure and authoritative references\n\n5. Measure again\n\nCompare new results against the original baseline\n\nFor example, if competitors are consistently cited for a particular question but your brand isn't, investigate:\n\nWhether you have a relevant page covering the topic.\n\nWhether your content provides clear, factual answers.\n\nWhether your claims are supported by credible evidence.\n\nWhether authoritative third-party sources reference your organisation.\n\nWhether your website is accessible to search engines and AI crawlers.\n\n6. Assign clear ownership\n\nAI citation tracking should not sit exclusively within the SEO team.\n\nTeam\n\n\t\n\nResponsibility\n\n\n\n\nSEO\n\n\t\n\nTechnical visibility and citation monitoring\n\n\n\n\nContent\n\n\t\n\nContent gaps and optimisation\n\n\n\n\nPR / Communications\n\n\t\n\nThird-party mentions and authority\n\n\n\n\nBrand\n\n\t\n\nBrand accuracy and sentiment\n\n\n\n\nAnalytics\n\n\t\n\nReporting and measurement\n\n\n\n\nLeadership\n\n\t\n\nStrategic priorities and investment\n\nRecommended tracking template\n\nI'd start with a simple central spreadsheet or database with these fields:\n\nField\n\n\t\n\nExample\n\n\n\n\nDate checked\n\n\t\n\n15 Nov 2026\n\n\n\n\nAI platform\n\n\t\n\nChatGPT\n\n\n\n\nPrompt\n\n\t\n\nBest providers of X\n\n\n\n\nBrand mentioned\n\n\t\n\nYes\n\n\n\n\nWebsite cited\n\n\t\n\nYes\n\n\n\n\nCited URL\n\n\t\n\n/solutions/example\n\n\n\n\nCompetitors mentioned\n\n\t\n\nCompetitor A, B\n\n\n\n\nCitation position\n\n\t\n\n2\n\n\n\n\nSentiment\n\n\t\n\nPositive\n\n\n\n\nNotes\n\n\t\n\nStrong product recommendation\n\nMy recommendation: Start with 30–50 high-value prompts, five or six AI platforms, and monthly reporting. Build a reliable baseline before investing in more sophisticated tracking.\n\nThe biggest mistake is focusing exclusively on the number of citations. A smaller number of highly relevant citations on commercially important questions can be more valuable than hundreds of incidental mentions.\n\nTools worth evaluating\n\nSome established platforms in this space include:\n\nProfound – AI search visibility and citation analytics.\n\nPeec AI – AI visibility and competitor monitoring.\n\nOtterly.AI – AI search mentions and monitoring.\n\nSemrush – SEO and emerging AI visibility capabilities.\n\nOne strategic question: Are you building this tracking framework for an in-house marketing team monitoring its own brand, or for an agency monitoring multiple clients?\n\nThat distinction affects the dashboard structure, reporting model and tooling I would recommend.",
      "citations": [],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 0 sources and 0 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-product-page-evidence",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-product-page-evidence",
      "published_at": "2026-11-15T13:53:00.000Z",
      "updated_at": "2026-11-15T13:53:00.000Z",
      "revision": 1,
      "slug": "chatgpt-product-page-evidence",
      "panelId": "product-page-evidence",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-product-page-evidence",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "What evidence makes a product page citable by AI search?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:53 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b940-70d4-83ea-a5e6-13fb61cd3f82?temporary-chat=true",
      "mentions": "Cited",
      "summary": "A product page becomes citable by AI search when it provides clear, specific, verifiable evidence that an AI system can retrieve, understand, and confidently use to answer a question.",
      "answerSnapshot": "：\n\nA product page becomes citable by AI search when it provides clear, specific, verifiable evidence that an AI system can retrieve, understand, and confidently use to answer a question.\n\nThe key distinction is that being indexed is not the same as being cited.\n\nIndexed: AI search can discover the page.\n\nUnderstood: AI can extract its product information.\n\nTrusted: AI can establish that the information is credible.\n\nCited: AI selects the page as supporting evidence in an answer.\n\n1. The seven types of evidence that make a product page citable\n\n1. Factual product specifications\n\nConcrete, extractable facts:\n\nDimensions, weight and materials\n\nCompatibility and supported features\n\nPerformance measurements\n\nModel numbers and variants\n\nEvidence: Manufacturer specifications, technical documents and measurable attributes.\n\n2. Original testing and performance data\n\nEvidence that the product has actually been evaluated:\n\nIndependent test results\n\nComparative performance measurements\n\nTesting methodology\n\nReal-world usage observations\n\nEvidence: Published test reports, photographs, measurement data and reproducible results.\n\n3. Customer experience\n\nEvidence from people who have used the product:\n\nVerified purchaser reviews\n\nSpecific use cases\n\nRecurring strengths and limitations\n\nCustomer questions and answers\n\nEvidence: Authentic, attributable customer feedback rather than generic testimonials.\n\n4. Certifications and compliance\n\nVerifiable proof of product claims:\n\nRelevant safety certifications\n\nRegulatory compliance\n\nEnvironmental certifications\n\nTesting laboratory credentials\n\nEvidence: Named certification bodies, certificate references and supporting documentation.\n\n5. Comparative evidence\n\nHelp AI answer questions such as:\n\nHow does this product compare with alternatives?\n\nWhat makes it different?\n\nWhich use case is it suitable for?\n\nWhat are its limitations?\n\nEvidence: Transparent comparison tables, consistent measurements and clearly explained trade-offs.\n\n6. Brand and manufacturer authority\n\nEvidence establishing who stands behind the product:\n\nIdentifiable manufacturer\n\nOfficial product documentation\n\nAuthoritative brand website\n\nProduct history and support information\n\nEvidence: Consistent brand and product information across trustworthy sources.\n\n7. Machine-readable product information\n\nMake factual information easy for search systems to interpret:\n\nProduct schema\n\nGTIN, SKU and brand identifiers\n\nPrice and availability\n\nReview and aggregate rating data\n\nCrawlable HTML content\n\nEvidence: Valid structured data that matches the information visible on the page.\n\n2. What does the strongest evidence look like?\n\nConsider a hypothetical product: a wireless headset.\n\nClaim\n\n\t\n\nWeak evidence\n\n\t\n\nStrong evidence\n\n\n\n\nBattery life\n\n\t\n\nLong-lasting battery\n\n\t\n\nTested playback duration with methodology\n\n\n\n\nComfort\n\n\t\n\nExtremely comfortable\n\n\t\n\nDocumented user testing\n\n\n\n\nDurability\n\n\t\n\nBuilt to last\n\n\t\n\nPublished durability test results\n\n\n\n\nCompatibility\n\n\t\n\nWorks with everything\n\n\t\n\nNamed compatible devices and operating systems\n\n\n\n\nSound quality\n\n\t\n\nAmazing audio\n\n\t\n\nIndependent measurements and listening tests\n\n\n\n\nCustomer satisfaction\n\n\t\n\nCustomers love it\n\n\t\n\nVerified reviews with specific experiences\n\nThe important principle: AI needs evidence behind claims, not just claims written in persuasive language.\n\n3. What makes evidence particularly valuable?\n\nI would assess evidence against five dimensions:\n\nDimension\n\n\t\n\nWhat to look for\n\n\n\n\nVerifiability\n\n\t\n\nCan the claim be independently checked?\n\n\n\n\nSpecificity\n\n\t\n\nDoes it contain measurable detail?\n\n\n\n\nProvenance\n\n\t\n\nIs the source identifiable?\n\n\n\n\nIndependence\n\n\t\n\nIs there corroboration beyond the seller?\n\n\n\n\nConsistency\n\n\t\n\nDoes the same fact appear consistently across sources?\n\n4. Technical requirements: making the evidence accessible\n\nA product page should also have:\n\nA unique, indexable URL.\n\nImportant product information available in crawlable HTML.\n\nClear product names and descriptive headings.\n\nProduct structured data using Schema.org.\n\nConsistent Product, Offer, Brand and Review information.\n\nAccurate price, stock and variant information.\n\nInternal links connecting relevant product and supporting information.\n\nGoogle explicitly explains that structured data helps search engines understand page content, but it does not guarantee AI citations. Its November 2026 guidance also says there is no special AI schema requirement. \nGoogle for Developers\n+2\n\n5. A practical product-page citation audit\n\nEvidence readiness checklist\n\n0/10 complete\nProduct claims are supported by measurable facts\nSpecifications match manufacturer documentation\nOriginal testing or independent evidence is available\nCustomer reviews are authentic and attributable\nCertifications link to verifiable sources\nProduct comparisons include transparent criteria\nManufacturer and brand identity are clear\nProduct schema is valid and matches visible content\nImportant information is crawlable without interaction\nProduct information is consistent across external sources\nReset checklist\n6. The most important strategic distinction\n\nThere are three separate things to optimise:\n\nDiscoverability: Can AI search find and access the product page?\n\nExtractability: Can it accurately extract the product facts?\n\nCitation-worthiness: Does the page contain sufficiently useful, distinctive and trustworthy evidence to justify referencing it?\n\nThe third is where many product pages fall short. They contain descriptions and marketing claims, but very little original evidence.\n\nMy recommendation is to prioritise original testing, verifiable specifications, authentic customer experiences and independent corroboration before investing heavily in additional AI-specific markup.\n\nUseful primary references:\n\nGoogle's AI search optimisation guidance\n\nGoogle's helpful, reliable content guidance\n\nGoogle Product structured data documentation\n\nThe ultimate test is simple: Could an AI answer a specific product-related question using your page as evidence, and could another person independently verify that answer? If yes, you're building a genuinely citable product page.",
      "citations": [
        {
          "title": "Google's AI search optimisation guidance",
          "host": "developers.google.com",
          "url": "https://developers.google.com/search/docs/fundamentals/ai-optimization-guide?utm_source=chatgpt.com",
          "position": 1
        },
        {
          "title": "Google's helpful, reliable content guidance",
          "host": "developers.google.com",
          "url": "https://developers.google.com/search/docs/fundamentals/creating-helpful-content?utm_source=chatgpt.com",
          "position": 2
        },
        {
          "title": "Google Product structured data documentation",
          "host": "developers.google.com",
          "url": "https://developers.google.com/search/docs/appearance/structured-data/product-snippet?utm_source=chatgpt.com",
          "position": 3
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 3 sources and 3 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-product-page-evidence-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-product-page-evidence-repeat-2",
      "published_at": "2026-11-15T13:53:00.000Z",
      "updated_at": "2026-11-15T13:53:00.000Z",
      "revision": 1,
      "slug": "chatgpt-product-page-evidence-repeat-2",
      "panelId": "product-page-evidence",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-product-page-evidence",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "What evidence makes a product page citable by AI search?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:53 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b954-bea4-83e9-9776-270a788af94f?temporary-chat=true",
      "mentions": "Not observed",
      "summary": "A product page becomes citable by AI search when it provides clear, extractable, verifiable evidence that directly answers a question a shopper might ask.",
      "answerSnapshot": "：\n\nA product page becomes citable by AI search when it provides clear, extractable, verifiable evidence that directly answers a question a shopper might ask.\n\nThe important distinction is that being indexed, being mentioned in an AI answer, and being cited as a source are three different outcomes.\n\nThere are six main evidence signals worth examining.\n\n1. Product facts that can be independently verified\n\nAI systems need specific facts rather than vague marketing language.\n\nWeak evidence\n\n\t\n\nStronger evidence\n\n\n\n\nPremium quality\n\n\t\n\nNamed materials and specifications\n\n\n\n\nLong-lasting\n\n\t\n\nPublished durability test results\n\n\n\n\nCompatible with most devices\n\n\t\n\nExplicit compatibility list\n\n\n\n\nEco-friendly\n\n\t\n\nNamed certification and supporting evidence\n\n\n\n\nFast charging\n\n\t\n\nCharging speed in watts and test conditions\n\n\n\n\nBest-selling\n\n\t\n\nPublished sales data and timeframe\n\nFor example:\n\nWeak: \"A powerful, long-lasting battery.\"\n\nStrong: \"A 5,000 mAh battery with up to 18 hours of stated operating time under the manufacturer's test conditions.\"\n\nThe second statement is easier to extract, quote and verify.\n\n2. Structured product information\n\nMachine-readable data helps search systems understand exactly what the page describes.\n\nImportant Schema.org properties include:\n\nProduct\n\nOffer\n\nBrand\n\nReview\n\nAggregateRating\n\nProductGroup (where applicable)\n\nUseful product attributes include:\n\nAttribute\n\n\t\n\nEvidence provided\n\n\n\n\nProduct name\n\n\t\n\nExact product identity\n\n\n\n\nBrand\n\n\t\n\nManufacturer identity\n\n\n\n\nGTIN / MPN\n\n\t\n\nUnique product identification\n\n\n\n\nPrice\n\n\t\n\nCurrent commercial offer\n\n\n\n\nAvailability\n\n\t\n\nStock status\n\n\n\n\nMaterial\n\n\t\n\nProduct composition\n\n\n\n\nDimensions\n\n\t\n\nPhysical specifications\n\n\n\n\nRating\n\n\t\n\nCustomer feedback\n\nGoogle specifically recommends accurate product identification, including valid GTINs and manufacturer details. \nGoogle for Developers\n+1\n\nImportant: Schema makes information easier to interpret. It does not automatically make a page more likely to receive an AI citation. \naixiv.science\n+1\n\n3. Original evidence rather than manufacturer copy\n\nThis is one of the most important distinctions.\n\nA product page containing only generic manufacturer descriptions offers relatively little unique information.\n\nStronger evidence includes:\n\nOriginal product testing.\n\nIndependent laboratory results.\n\nMeasured performance.\n\nOriginal product photography.\n\nComparison against clearly identified alternatives.\n\nExpert explanations.\n\nDocumented limitations.\n\nSpecific use-case recommendations.\n\nFor example, instead of saying:\n\n\"Suitable for professional photographers.\"\n\nExplain which camera systems it supports, what performance was tested, and under which conditions.\n\nThe more independently useful the information, the more valuable the page becomes as a source.\n\n4. Third-party corroboration\n\nAI systems can encounter the same product information across multiple sources.\n\nEvidence becomes more credible when product claims are corroborated by:\n\nIndependent product reviews.\n\nRecognised industry publications.\n\nTesting organisations.\n\nManufacturer documentation.\n\nRetailer specifications.\n\nVerified customer feedback.\n\nThe important factor is consistency and source quality, not simply the number of websites repeating a claim.\n\n5. Content that directly answers buyer questions\n\nA product page should answer questions that people actually ask AI assistants.\n\nFor example:\n\nBuyer question\n\n\t\n\nEvidence the page should contain\n\n\n\n\nWhat is this product?\n\n\t\n\nClear product definition\n\n\n\n\nWho is it suitable for?\n\n\t\n\nIntended audience and use cases\n\n\n\n\nWhat makes it different?\n\n\t\n\nSpecific differentiators\n\n\n\n\nIs it compatible?\n\n\t\n\nCompatibility information\n\n\n\n\nWhat are its limitations?\n\n\t\n\nExplicit exclusions and constraints\n\n\n\n\nIs it worth the price?\n\n\t\n\nFeatures, performance and price context\n\n\n\n\nHow does it compare?\n\n\t\n\nComparable specifications\n\nThis is particularly important because recent research suggests that semantic alignment between search queries and page content is a major factor in AI citation selection. \narXiv\n+1\n\n6. Technical accessibility and freshness\n\nEven excellent evidence cannot be cited if an AI search system cannot retrieve it.\n\nCheck:\n\nThe page is publicly accessible.\n\nImportant product information is available in crawlable HTML.\n\nSearch engine indexing is permitted.\n\nRelevant AI search crawlers are not unnecessarily blocked.\n\nProduct information is consistent between page content, structured data and feeds.\n\nPrices and availability are maintained.\n\nCanonical URLs identify the correct product page.\n\nFor ChatGPT Search specifically, OpenAI documents the role of OAI-SearchBot in making websites discoverable through search.\n\nA practical product-page citation scorecard\n\nHere is a useful framework for auditing a product detail page.\n\nProduct-page evidence audit\n\n0/8 assessed\n\nProduct identity\n\nExact name, brand, GTIN or MPN\n\nVerifiable specifications\n\nSpecific, measurable product facts\n\nOriginal evidence\n\nTesting, measurements and original content\n\nStructured data\n\nAccurate Product and Offer markup\n\nBuyer question coverage\n\nAnswers to practical purchase questions\n\nIndependent corroboration\n\nCredible external reviews and references\n\nTechnical accessibility\n\nCrawlable, indexable and accessible HTML\n\nInformation freshness\n\nCurrent pricing, availability and specifications\n\nReset audit\n\nThis is an operational checklist, not a validated AI citation prediction model.\n\nWhat does the research actually tell us?\n\nThere are some important nuances:\n\nA 2026 study examining structured data on four products found improvements in citation positioning after implementation, but no increase in the number of responses featuring the webshop. \ncs.ru.nl\n\nAnother 2026 study found that structured-data presence alone did not independently predict AI citations after controlling for other factors. \naixiv.science\n\nResearch published in October 2026 suggests that matching page titles and paragraphs to the questions people ask is particularly important for retrieval. \narXiv\n\nThis means there is no proven magic combination of schema, FAQs or word count that guarantees citations.\n\nThe key distinction\n\nI would separate product-page AI performance into three measurable stages:\n\nStage 1\n\nDiscoverability\n\nCan the AI search system find and retrieve the product page?\n\nStage 2\n\nEvidence extraction\n\nCan it extract accurate, specific and useful product facts?\n\nStage 3\n\nCitation\n\nDoes it actually reference the product page as evidence in its answer?\n\nMy conclusion: The strongest product pages for AI search are not simply SEO-optimised product descriptions. They are verifiable product-information sources, containing original evidence, explicit specifications, trustworthy references and answers to real buyer questions.\n\nThat is the distinction between a product page that AI can read and one that AI has a reason to cite.",
      "citations": [],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 0 sources and 0 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-comparison-page-citations",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-comparison-page-citations",
      "published_at": "2026-11-15T13:52:00.000Z",
      "updated_at": "2026-11-15T13:52:00.000Z",
      "revision": 1,
      "slug": "chatgpt-comparison-page-citations",
      "panelId": "comparison-page-citations",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-comparison-page-citations",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "How do AI search engines choose comparison pages as sources?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:52 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b8f4-e918-83ea-ae39-5b24634a2895?temporary-chat=true",
      "mentions": "Cited",
      "summary": "AI search engines choose comparison pages based on how useful, trustworthy, relevant, and easy to extract information from those pages.",
      "answerSnapshot": "：\n\nAI search engines choose comparison pages based on how useful, trustworthy, relevant, and easy to extract information from those pages.\n\nUnlike traditional search engines, which primarily return a list of links ranked by relevance, AI search engines often retrieve several sources, extract information from them, compare the claims, and use selected sources to construct an answer.\n\nFor example, if someone searches \"Best project management software compared\", an AI search engine might look for:\n\nIndependent comparison articles.\n\nProduct-versus-product reviews.\n\nDetailed feature comparison tables.\n\nOfficial product documentation.\n\nExpert reviews and customer feedback.\n\nThe important distinction is that ranking highly in traditional search does not automatically guarantee being cited by an AI search engine.\n\n1. The six stages of source selection\n\nQuery understanding\n\nThe engine interprets the question and identifies what needs comparing: price, features, performance, limitations, use cases, etc.\n\nQuery expansion\n\nIt may generate several smaller searches. For example:\n\nProduct A vs Product B\n\nProduct A pricing\n\nProduct B integrations\n\nBest alternatives to Product A\n\nCandidate retrieval\n\nIt retrieves potentially relevant webpages from its search index or other available sources.\n\nPassage evaluation\n\nIt examines individual sections of pages to determine whether they contain useful information answering the question.\n\nEvidence selection\n\nIt favours useful, credible, relevant and sufficiently current information, while potentially excluding duplicate or weak sources.\n\nCitation selection\n\nIt generates the answer and chooses which sources to display as supporting citations.\n\nThis is a simplified model. Individual search engines use different systems, and their precise selection algorithms are not public.\n\nThis distinction matters: a page can be discovered without being selected, and a page can influence an answer without appearing as a visible citation. \nThe AI Search Handbook\n+2\n\n2. What makes comparison pages attractive sources?\n\nHere are the main characteristics that can make a comparison page useful to AI search systems.\n\nCharacteristic\n\n\t\n\nWhy it matters\n\n\n\n\nDirect relevance\n\n\t\n\nMatches the exact comparison being requested\n\n\n\n\nDetailed comparisons\n\n\t\n\nProvides evidence about multiple alternatives\n\n\n\n\nClear tables\n\n\t\n\nMakes differences easier to extract\n\n\n\n\nSpecific facts\n\n\t\n\nIncludes prices, features, specifications and limitations\n\n\n\n\nIndependent evidence\n\n\t\n\nSupports claims with verifiable information\n\n\n\n\nTopical authority\n\n\t\n\nDemonstrates knowledge of the subject\n\n\n\n\nFreshness\n\n\t\n\nKeeps changing information current\n\n\n\n\nClear structure\n\n\t\n\nMakes individual passages understandable outside their original context\n\nOne particularly important factor is semantic relevance.\n\nA page titled \"The Complete Guide to Business Software\" might be less useful for a particular query than a page titled \"Product A vs Product B: Features, Pricing and Limitations\".\n\nThe second page more directly addresses the information need.\n\nResearch published in October 2026 also suggests that matching the search engine's rewritten queries can be more important than superficial page-formatting features. \narXiv\n+1\n\n3. Do AI engines prefer comparison websites over official websites?\n\nNot necessarily. They serve different purposes.\n\nFor a comparison query, an AI engine may combine several types of sources:\n\nSource type\n\n\t\n\nWhat it contributes\n\n\n\n\nIndependent comparison sites\n\n\t\n\nSide-by-side evaluations\n\n\n\n\nOfficial product websites\n\n\t\n\nFeatures, specifications and pricing\n\n\n\n\nReview publications\n\n\t\n\nHands-on testing and expert opinions\n\n\n\n\nCustomer reviews\n\n\t\n\nReal-world experiences\n\n\n\n\nTechnical documentation\n\n\t\n\nVerified technical capabilities\n\nFor example, an AI engine comparing two software products might use an independent review for its evaluation, but consult both manufacturers' websites to verify their pricing and features.\n\nThis helps reduce reliance on a single source.\n\nHowever, comparison websites can have commercial incentives, particularly where affiliate commissions are involved. A well-presented comparison is not automatically an unbiased one.\n\n4. What does an ideal AI-citable comparison page look like?\n\nConsider this example structure:\n\nExample structure\nProduct A vs Product B (2026)\n\nAn independent comparison of features, pricing, integrations and limitations.\n\n1. Direct answer\n\nA short explanation of which product is better suited to different needs.\n\n2. Side-by-side comparison\n\nFeature\n\n\t\n\nProduct A\n\n\t\n\nProduct B\n\n\n\n\nPricing\n\n\t\n\n£20/month\n\n\t\n\n£30/month\n\n\n\n\nIntegrations\n\n\t\n\n40\n\n\t\n\n65\n\n\n\n\nFree plan\n\n\t\n\nYes\n\n\t\n\nNo\n\n\n\n\nAPI access\n\n\t\n\nYes\n\n\t\n\nYes\n\nIllustrative figures, not actual product data.\n\n3. Individual product analysis\n\nMain strengths\n\nWeaknesses\n\nImportant features\n\nIdeal use cases\n\n4. Evidence and methodology\n\nLinks to official documentation\n\nDate pricing was checked\n\nExplanation of testing methodology\n\nSources for factual claims\n\n5. Final verdict\n\nExplain which product is suitable for which type of user, rather than declaring one universal winner.\n\nThis format makes the information easier to interpret, verify and reuse.\n\n5. Do different AI search engines select sources differently?\n\nYes.\n\nSearch engine\n\n\t\n\nGeneral approach\n\n\n\n\nChatGPT Search\n\n\t\n\nRetrieves web information and uses relevant sources to support answers\n\n\n\n\nGoogle AI Overviews\n\n\t\n\nUses Google's search infrastructure and AI-generated summaries\n\n\n\n\nPerplexity\n\n\t\n\nSearches the web and presents answers with supporting citations\n\n\n\n\nMicrosoft Copilot\n\n\t\n\nUses web search and grounding to support generated responses\n\nTheir exact source-ranking and citation-selection formulas are not publicly disclosed. \nSEORecheck\n+2\n\n6. The most important distinction: ranking vs citation\n\nThere are three separate outcomes:\n\nVisibility: Does the comparison page appear in the search engine's candidate results?\n\nSelection: Does the AI retrieve and use information from that page?\n\nCitation: Does the AI actually display the page as a source?\n\nA website can achieve the first without achieving the second or third.\n\nThis is why measuring traditional SEO rankings alone doesn't tell you how visible a comparison page is in AI-generated answers.\n\n7. What this means for comparison-page publishers\n\nThe practical lesson is to create pages that function as useful evidence, not simply pages that mention lots of competing products.\n\nThe strongest approach is generally to:\n\nAnswer specific comparison questions directly.\n\nInclude genuinely useful side-by-side data.\n\nMake factual claims independently verifiable.\n\nExplain how comparisons were conducted.\n\nKeep time-sensitive information updated.\n\nClearly disclose commercial relationships.\n\nBuild a reputation for reliable coverage of the subject.\n\nOne final caveat: No publisher can guarantee that ChatGPT, Gemini, Perplexity or another AI engine will cite a particular page. Their retrieval and citation processes are dynamic.\n\nFor further reading, the Ofcom research paper on the era of answer engines provides useful background on how generative search systems retrieve and present information.\n\nThe key takeaway: AI search engines don't simply choose the best-looking comparison page. They try to find retrievable information that helps answer the particular question, then decide which evidence is useful enough to support the generated response.",
      "citations": [
        {
          "title": "Ofcom research paper on the era of answer engines",
          "host": "ofcom.org.uk",
          "url": "https://www.ofcom.org.uk/siteassets/resources/documents/research-and-data/online-research/other/the-era-of-answer-engines---discussion-paper.pdf?v=407803&utm_source=chatgpt.com",
          "position": 1
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 1 sources and 1 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-comparison-page-citations-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-comparison-page-citations-repeat-2",
      "published_at": "2026-11-15T13:52:00.000Z",
      "updated_at": "2026-11-15T13:52:00.000Z",
      "revision": 1,
      "slug": "chatgpt-comparison-page-citations-repeat-2",
      "panelId": "comparison-page-citations",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-comparison-page-citations",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "How do AI search engines choose comparison pages as sources?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:52 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b908-0a4c-83e9-98b4-827ee455a74d?temporary-chat=true",
      "mentions": "Not observed",
      "summary": "AI search engines choose comparison pages based on how well they help answer a user's question, rather than simply selecting the pages with the highest traditional Google rankings.",
      "answerSnapshot": "：\n\nAI search engines choose comparison pages based on how well they help answer a user's question, rather than simply selecting the pages with the highest traditional Google rankings.\n\nFor example, if someone searches \"Best project management software for small businesses\", an AI search engine might look for pages that compare several products, explain their differences, discuss pricing, and make recommendations.\n\nThe process generally involves five stages:\n\n1. Understanding the search intent\n\nThe AI first interprets what the user wants.\n\nFor example:\n\n\"X vs Y\" → direct product comparison.\n\n\"Best alternatives to X\" → alternatives and competitor lists.\n\n\"Best software for small businesses\" → ranked recommendations.\n\n\"Which is cheaper, X or Y?\" → pricing comparisons.\n\nThis helps determine which types of pages are relevant.\n\n2. Finding potentially relevant pages\n\nAI search systems retrieve candidate sources from search indexes and other available information sources.\n\nThey may consider:\n\nTraditional search relevance.\n\nHow closely the page matches the question.\n\nWhether the page covers the products or brands mentioned.\n\nWhether the content is accessible and understandable.\n\nWhether the page contains information that can answer the query.\n\nA comparison page doesn't necessarily need to rank first in traditional search to be considered.\n\n3. Evaluating the comparison page\n\nOnce a page is retrieved, the AI needs to determine whether its content is useful.\n\nComparison pages are particularly useful when they contain:\n\nSignal\n\n\t\n\nWhy it matters\n\n\n\n\nDirect comparisons\n\n\t\n\nMakes differences easy to identify\n\n\n\n\nComparison tables\n\n\t\n\nPresents structured, extractable information\n\n\n\n\nSpecific product features\n\n\t\n\nHelps answer detailed questions\n\n\n\n\nPricing information\n\n\t\n\nSupports cost-related answers\n\n\n\n\nStrengths and limitations\n\n\t\n\nHelps explain trade-offs\n\n\n\n\nEvidence and references\n\n\t\n\nMakes claims easier to verify\n\n\n\n\nRecent updates\n\n\t\n\nHelps with current product information\n\nFor example, consider two pages:\n\nPage A: \"The 10 best CRM platforms\" followed by short descriptions.\n\nPage B: A detailed comparison of 10 CRM platforms, including pricing, integrations, features, limitations and ideal use cases.\n\nFor a question about which CRM has the best integrations, Page B has a clearer answer to extract.\n\nHowever, Page A could still be selected if it better matches the particular question.\n\n4. Extracting useful information\n\nAI systems don't necessarily use an entire article.\n\nThey can identify individual passages that answer particular questions.\n\nFor example, an AI might extract:\n\nA paragraph explaining the differences between two products.\n\nA table comparing subscription plans.\n\nA section explaining which product suits beginners.\n\nA statement about limitations or compatibility.\n\nThis is important because a page can be cited for one specific fact without being used as the main source for the entire answer.\n\n5. Deciding which sources to cite\n\nFinally, the AI synthesises information from the selected sources.\n\nIt may prioritise sources that:\n\nDirectly support the answer.\n\nProvide specific, verifiable facts.\n\nAgree with other credible sources.\n\nOffer information not adequately covered elsewhere.\n\nIt may also use several different comparison websites rather than relying on one publisher.\n\nImportantly, being retrieved, being read and being cited are three different things.\n\nWhat makes one comparison website more likely to be cited?\n\nThere is no publicly documented universal scoring formula used by ChatGPT, Perplexity, Gemini or Google AI Overviews.\n\nHowever, research into AI citations suggests that some factors are particularly important:\n\nRelevance: Does the page answer the actual question?\n\nTopical coverage: Does it cover the products, brands and attributes being compared?\n\nInformation quality: Are the claims specific, accurate and supported?\n\nExtractability: Can the AI easily identify and understand the relevant information?\n\nSearch visibility: Can the engine retrieve the page in the first place?\n\nFreshness: Is the information current enough for the question?\n\nA 2026 research paper analysing more than 158,000 AI-generated answers found that retrieval and query relevance were major influences on which pages received citations. It also found that matching page titles and paragraphs to users' questions improved citation likelihood. \narXiv\n+1\n\nDo different AI search engines choose different comparison pages?\n\nYes.\n\nAI search engine\n\n\t\n\nMain retrieval environment\n\n\n\n\nChatGPT Search\n\n\t\n\nOpenAI search and external search providers\n\n\n\n\nGoogle AI Overviews\n\n\t\n\nGoogle Search\n\n\n\n\nGemini\n\n\t\n\nGoogle Search grounding\n\n\n\n\nMicrosoft Copilot\n\n\t\n\nBing\n\n\n\n\nPerplexity\n\n\t\n\nIts own search infrastructure and other sources\n\nConsequently, a comparison website might be cited frequently by one AI engine but rarely by another. \nCitable\n+1\n\nAn important distinction: ranking versus citation\n\nTraditional SEO asks:\n\n\"How can my comparison page rank higher in search results?\"\n\nAI search visibility asks a slightly different question:\n\n\"When an AI needs information about the products I'm comparing, will it retrieve my page, understand its information and use it as evidence?\"\n\nThese overlap, but they aren't identical.\n\nMy key takeaway: A comparison page is potentially valuable to AI search engines when it acts as a useful, relevant and trustworthy information source, rather than simply being a list of products with promotional descriptions.\n\nOne useful way to understand this is to distinguish three stages:\n\nDiscovery\n\nCan the AI find and retrieve the comparison page?\n\nUnderstanding\n\nCan the AI extract the relevant comparisons and supporting facts?\n\nCitation\n\nDoes the AI actually use the page to support its answer?\n\nThis distinction is particularly useful when analysing why some comparison websites become frequently cited AI sources while others don't.",
      "citations": [],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 0 sources and 0 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-comparison-page-citations-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-comparison-page-citations-repeat-3",
      "published_at": "2026-11-15T13:52:00.000Z",
      "updated_at": "2026-11-15T13:52:00.000Z",
      "revision": 1,
      "slug": "chatgpt-comparison-page-citations-repeat-3",
      "panelId": "comparison-page-citations",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-comparison-page-citations",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "How do AI search engines choose comparison pages as sources?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:52 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b91d-1bd4-83ea-abff-8de56644e0da?temporary-chat=true",
      "mentions": "Not observed",
      "summary": "AI search engines choose comparison pages based on how useful, trustworthy, relevant, and easy to extract information from they appear to be.",
      "answerSnapshot": "：\n\nAI search engines choose comparison pages based on how useful, trustworthy, relevant, and easy to extract information from they appear to be.\n\nThe important distinction is that AI search engines don't necessarily choose the page that ranks highest on Google. They try to identify sources that can help them construct a reliable answer to a user's question.\n\nFor example, if someone searches:\n\n\"Best CRM software for small businesses\"\n\n\"HubSpot vs Salesforce\"\n\n\"Best project management tools compared\"\n\n\"Which accounting software is easiest to use?\"\n\nAn AI search engine may look for comparison articles, product reviews, official product documentation, customer feedback, and independent research.\n\n1. The five main selection criteria\n\nFactor\n\n\t\n\nWhat the AI looks for\n\n\n\n\nRelevance\n\n\t\n\nDoes the page directly answer the user's question?\n\n\n\n\nAuthority\n\n\t\n\nIs the publisher recognised as a credible source?\n\n\n\n\nEvidence\n\n\t\n\nDoes it contain facts, data, examples and supporting references?\n\n\n\n\nFreshness\n\n\t\n\nAre product details, prices and comparisons up to date?\n\n\n\n\nExtractability\n\n\t\n\nCan the AI easily understand and extract useful information?\n\nThese aren't published universal ranking weights. Each search engine has its own systems, and the exact formulas are proprietary. \ndiscussion paper\n+2\n\n2. Why comparison pages are particularly useful\n\nComparison pages are valuable because they do much of the research and organisation that an AI answer needs.\n\nFor example, consider an article titled:\n\n\"The 10 Best CRM Platforms for Small Businesses in 2026\"\n\nAn AI engine can extract:\n\nThe products being compared.\n\nThe intended audience for each product.\n\nFeatures and functionality.\n\nPricing differences.\n\nAdvantages and disadvantages.\n\nWhich product is best for particular use cases.\n\nA well-structured comparison page can potentially answer several related searches from one source.\n\n11\n3. How the selection process works\n\nA simplified version of the process looks like this:\n\n1. User asks a question\n\n\"What's the best CRM for a small business?\"\n\n2. Query expansion\n\nThe system identifies related searches about CRM features, pricing and suitability.\n\n3. Source retrieval\n\nSearch systems retrieve potentially relevant comparison pages and other sources.\n\n4. Content evaluation\n\nThe AI evaluates relevant passages, facts and supporting evidence.\n\n5. Answer generation and citations\n\nSelected information is synthesised into an answer with supporting source links.\n\nThis process is commonly described as retrieval-augmented generation (RAG). Importantly, being retrieved doesn't guarantee that a page will ultimately be cited. \nCitable\n+2\n\n4. What makes one comparison website more likely to be cited?\n\nConsider two hypothetical comparison pages:\n\nPage A\n\n\t\n\nPage B\n\n\n\n\nGeneric product descriptions\n\n\t\n\nSpecific product comparisons\n\n\n\n\nNo supporting references\n\n\t\n\nLinks to original evidence\n\n\n\n\nOutdated pricing\n\n\t\n\nRecently verified pricing\n\n\n\n\nBroad marketing language\n\n\t\n\nSpecific facts and measurements\n\n\n\n\nNo clear methodology\n\n\t\n\nExplains how products were evaluated\n\nPage B offers more independently verifiable information and more useful passages for an AI system.\n\nOther useful characteristics include:\n\nClear comparison tables.\n\nExplicit evaluation criteria.\n\nOriginal research and testing.\n\nNamed authors and relevant expertise.\n\nTransparent commercial relationships.\n\nClear publication and update dates.\n\nConsistency with independent sources.\n\nOne caveat: a comparison page isn't automatically trustworthy just because it has a table or lists lots of products.\n\n5. Does each AI search engine behave differently?\n\nYes.\n\nSearch engine\n\n\t\n\nImportant distinction\n\n\n\n\nChatGPT Search\n\n\t\n\nUses web search and retrieves sources to support answers.\n\n\n\n\nGoogle AI Overviews\n\n\t\n\nBuilds on Google's existing Search systems.\n\n\n\n\nMicrosoft Copilot\n\n\t\n\nUses web retrieval and Bing-related search infrastructure.\n\n\n\n\nPerplexity\n\n\t\n\nEmphasises web retrieval and cited answers.\n\nTheir precise source-selection algorithms aren't fully public. \nAmpere\n+2\n\n6. The interesting strategic implication\n\nThere is an important distinction between two types of visibility:\n\nSearch visibility: Your comparison page ranks when someone searches Google or Bing.\n\nAI citation visibility: Your comparison page is selected as evidence when an AI generates an answer.\n\nThese overlap, but they aren't identical.\n\nA particularly interesting finding from 2026 research is that the queries an AI generates during retrieval can have a major influence on which pages it eventually cites. This means that matching the actual questions people ask matters, rather than simply repeating popular keywords. \narXiv\n\nMy takeaway: The strongest comparison pages aren't necessarily the longest or most heavily optimised. They're the ones that provide relevant, independently verifiable, well-organised information that an AI can confidently use to support a specific answer.\n\nA useful next step: I can also explain how a company can get its comparison pages selected and cited by ChatGPT, Perplexity and Google AI Overviews — essentially, how to approach this as an AI search optimisation (GEO) strategy.",
      "citations": [],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 0 sources and 0 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-local-business-ai-search-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-local-business-ai-search-repeat-2",
      "published_at": "2026-11-15T13:51:00.000Z",
      "updated_at": "2026-11-15T13:51:00.000Z",
      "revision": 1,
      "slug": "chatgpt-local-business-ai-search-repeat-2",
      "panelId": "local-business-ai-search",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-local-business-ai-search",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "Which pages should a local business optimize for AI search?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:51 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b8b8-eca0-83ea-8201-ba98e3276c5b?temporary-chat=true",
      "mentions": "Not observed",
      "summary": "A local business should optimize the pages that clearly explain what it does, where it operates, which customers it helps, and why people should choose it.",
      "answerSnapshot": "：\n\nA local business should optimize the pages that clearly explain what it does, where it operates, which customers it helps, and why people should choose it.\n\nFor AI search (ChatGPT, Google AI Overviews, Perplexity and other AI-powered search experiences), the goal is to make your business easy to understand, verify, recommend and reference.\n\nI would prioritize the following pages.\n\nThe 10 most important pages to optimize\n\n1. Homepage\n\nEstablish the business identity:\n\nWhat the business does.\n\nWho it serves.\n\nWhere it operates.\n\nMain products or services.\n\nWhat makes it different.\n\nMake the opening paragraph particularly clear.\n\n2. Individual service pages\n\nCreate a dedicated page for every major service.\n\nInclude:\n\nWhat the service involves.\n\nWho needs it.\n\nPricing or pricing guidance.\n\nProcess and timescales.\n\nCommon customer questions.\n\nRelevant experience and credentials.\n\n3. Location and service-area pages\n\nCreate pages for important towns, cities or neighbourhoods served.\n\nInclude:\n\nServices available in that location.\n\nActual coverage information.\n\nLocal examples and projects.\n\nRelevant customer testimonials.\n\nLocation-specific FAQs.\n\nAvoid mass-producing pages that simply swap town names.\n\n4. About us page\n\nHelp AI understand who is behind the business.\n\nInclude:\n\nCompany history.\n\nFounders and team.\n\nQualifications.\n\nAccreditations.\n\nYears of experience.\n\nAreas of expertise.\n\n5. Reviews and testimonials page\n\nDemonstrate customer satisfaction through:\n\nGenuine customer reviews.\n\nNamed testimonials where permission is given.\n\nCase studies.\n\nSpecific outcomes.\n\nLinks to independent review profiles.\n\n6. FAQ page\n\nAnswer questions customers might ask AI assistants.\n\nExamples:\n\nHow much does the service cost?\n\nDo you offer same-day appointments?\n\nWhich areas do you cover?\n\nHow long does the service take?\n\nWhat qualifications do you have?\n\n7. Case studies and project pages\n\nShow evidence of real-world experience.\n\nInclude:\n\nCustomer problem.\n\nService provided.\n\nLocation.\n\nApproach taken.\n\nMeasurable results where available.\n\n8. Pricing page\n\nExplain:\n\nTypical costs or price ranges.\n\nWhat's included.\n\nFactors affecting the price.\n\nOptional extras.\n\nHow to request a quote.\n\n9. Contact and individual location pages\n\nMake business details unambiguous:\n\nBusiness name.\n\nPhysical address, where applicable.\n\nPhone number.\n\nEmail.\n\nOpening hours.\n\nDirections and booking options.\n\n10. Educational guides and articles\n\nPublish genuinely useful answers to questions customers research before buying.\n\nExamples:\n\nHow to choose a reputable local provider.\n\nCommon problems and their solutions.\n\nService comparisons.\n\nMaintenance advice.\n\nWhat to expect during an appointment.\n\nPriority: which pages should you do first?\n\nIf resources are limited, I would tackle them in this order:\n\nPriority\n\n\t\n\nPage\n\n\t\n\nImportance\n\n\n\n\n1\n\n\t\n\nHomepage\n\n\t\n\nCritical\n\n\n\n\n2\n\n\t\n\nCore service pages\n\n\t\n\nCritical\n\n\n\n\n3\n\n\t\n\nLocation pages\n\n\t\n\nCritical\n\n\n\n\n4\n\n\t\n\nAbout us\n\n\t\n\nHigh\n\n\n\n\n5\n\n\t\n\nReviews/testimonials\n\n\t\n\nHigh\n\n\n\n\n6\n\n\t\n\nFAQ\n\n\t\n\nHigh\n\n\n\n\n7\n\n\t\n\nCase studies\n\n\t\n\nHigh\n\n\n\n\n8\n\n\t\n\nPricing\n\n\t\n\nHigh\n\n\n\n\n9\n\n\t\n\nContact\n\n\t\n\nHigh\n\n\n\n\n10\n\n\t\n\nEducational content\n\n\t\n\nMedium\n\nWhat should every important page contain?\n\nBeyond the page content itself, use a consistent approach:\n\nAnswer-first content: Give a direct answer near the beginning of each page.\n\nClear headings: Use headings that reflect actual customer questions.\n\nSpecific facts: Include real prices, locations, qualifications, processes and experience.\n\nTrust signals: Link to relevant credentials, genuine reviews and independent references.\n\nInternal links: Connect services, locations, FAQs and case studies.\n\nStructured data: Use appropriate Schema.org markup, such as LocalBusiness, Service and Organization.\n\nConsistent business information: Keep your business details consistent across your website and external profiles.\n\nStructured data should accurately reflect visible content. It helps clarify information but does not guarantee AI citations. \nKAP SEO Services\n+2\n\nOne important distinction\n\nI would think about AI search optimization in three layers:\n\nLayer 1\n\nBusiness understanding\n\nHomepage, About, Contact and Location pages.\n\nCan AI understand who you are and where you operate?\n\nLayer 2\n\nService understanding\n\nService, Pricing, FAQ and Educational pages.\n\nCan AI understand what you offer and answer questions about it?\n\nLayer 3\n\nTrust and authority\n\nReviews, Testimonials, Case Studies and Credentials.\n\nCan AI find evidence that your business is credible?\n\nMy biggest recommendation: Don't just optimize pages to rank for keywords. Build a website that acts as a comprehensive, trustworthy source of information about the business.\n\nThat gives search engines and AI assistants more useful information to understand, evaluate and potentially recommend your business.",
      "citations": [],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 0 sources and 0 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-local-business-ai-search-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-local-business-ai-search-repeat-3",
      "published_at": "2026-11-15T13:51:00.000Z",
      "updated_at": "2026-11-15T13:51:00.000Z",
      "revision": 1,
      "slug": "chatgpt-local-business-ai-search-repeat-3",
      "panelId": "local-business-ai-search",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-local-business-ai-search",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "Which pages should a local business optimize for AI search?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:51 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b8df-d4c8-83e9-839a-f8f5cd212e99?temporary-chat=true",
      "mentions": "Cited",
      "summary": "A local business should optimize the pages that help AI search engines understand what it offers, where it operates, who it serves, and why customers should trust it.",
      "answerSnapshot": "：\n\nA local business should optimize the pages that help AI search engines understand what it offers, where it operates, who it serves, and why customers should trust it.\n\nFor platforms such as ChatGPT, Google AI Overviews, Gemini and Perplexity, I would prioritize the following pages.\n\nThe 10 most important pages to optimize\n\nPriority\n\n\t\n\nPage\n\n\t\n\nWhat it should communicate\n\n\n\n\n1\n\n\t\n\nHomepage\n\n\t\n\nWho you are, what you do, where you operate and who you help\n\n\n\n\n2\n\n\t\n\nMain service pages\n\n\t\n\nDetailed explanations of each service or product\n\n\n\n\n3\n\n\t\n\nLocation pages\n\n\t\n\nThe specific towns, cities and areas you serve\n\n\n\n\n4\n\n\t\n\nAbout us\n\n\t\n\nBusiness history, expertise, qualifications and people\n\n\n\n\n5\n\n\t\n\nPricing pages\n\n\t\n\nCosts, pricing factors, packages and what's included\n\n\n\n\n6\n\n\t\n\nFAQ pages\n\n\t\n\nClear answers to common customer questions\n\n\n\n\n7\n\n\t\n\nReviews and testimonials\n\n\t\n\nCustomer experiences, outcomes and trust signals\n\n\n\n\n8\n\n\t\n\nCase studies / project pages\n\n\t\n\nReal examples of work, processes and results\n\n\n\n\n9\n\n\t\n\nContact page\n\n\t\n\nAddress, phone number, opening hours and service area\n\n\n\n\n10\n\n\t\n\nComparison pages\n\n\t\n\nDifferences between services, solutions or options\n\nWhat each page should contain\n\n1. Homepage: establish business identity\n\nMake it immediately clear:\n\nBusiness name and type.\n\nMain products or services.\n\nPrimary location and service areas.\n\nTypes of customers served.\n\nMain differentiators.\n\nLinks to key service and location pages.\n\n2. Service pages: establish topical relevance\n\nIdeally, create a dedicated page for every major service.\n\nFor example, an electrician might have:\n\nElectrical installations\n\nElectrical repairs\n\nElectrical inspections\n\nEV charger installation\n\nEmergency electrical services\n\nEach page should explain what the service involves, who needs it, typical costs, the process, common questions and the areas covered.\n\n3. Location pages: establish geographic relevance\n\nCreate useful pages for important towns and cities you serve.\n\nFor example:\n\nElectrician in Maidenhead\n\nElectrician in Windsor\n\nElectrician in Slough\n\nEach page should contain genuinely local information, such as:\n\nServices available in that location.\n\nLocal projects and examples.\n\nRelevant customer testimonials.\n\nAreas and neighbourhoods covered.\n\nLocal contact or availability information.\n\nAvoid creating dozens of near-identical pages that simply swap town names.\n\n4. About page: establish credibility\n\nInclude:\n\nYears in business.\n\nExperience and qualifications.\n\nRelevant certifications and memberships.\n\nTeam information.\n\nCompany history.\n\nWhat makes the business different.\n\n5. Pricing page: answer commercial questions\n\nAI search users frequently ask questions such as:\n\nHow much does an electrician cost?\n\nWhat does a boiler installation cost?\n\nHow much does a website designer charge?\n\nProvide realistic price ranges where possible, explain variables and clarify what's included.\n\n6. FAQ page: answer conversational queries\n\nOrganise questions around actual customer concerns.\n\nFor example:\n\nDo you offer same-day appointments?\n\nWhich areas do you cover?\n\nAre you available at weekends?\n\nDo you provide free quotations?\n\nAre your services guaranteed?\n\nAnswer each question directly before providing additional detail.\n\n7. Case studies: demonstrate first-hand experience\n\nInclude:\n\nCustomer requirements.\n\nThe original problem.\n\nYour recommended solution.\n\nWork completed.\n\nPhotographs.\n\nActual outcomes.\n\nOriginal experience and useful information are particularly valuable. Google's AI search guidance specifically encourages distinctive, first-hand content rather than generic material. \nGoogle for Developers\n\nDon't overlook these supporting assets\n\nOptimizing website pages alone isn't enough.\n\nAsset\n\n\t\n\nWhy it matters\n\n\n\n\nGoogle Business Profile\n\n\t\n\nConfirms business details and local presence\n\n\n\n\nReviews\n\n\t\n\nProvides external evidence of customer experience\n\n\n\n\nIndustry directories\n\n\t\n\nHelps establish business identity\n\n\n\n\nRelevant local publications\n\n\t\n\nBuilds local recognition and authority\n\n\n\n\nSocial profiles\n\n\t\n\nReinforces consistent business information\n\nGoogle specifically identifies Business Profiles as useful for local business visibility in generative AI experiences. \nGoogle for Developers\n\nTechnical optimization checklist\n\nApply these principles across the website:\n\nUse clear page titles and descriptive headings.\n\nMake important content accessible as text.\n\nUse internal links between related services and locations.\n\nImplement appropriate LocalBusiness and relevant business-type schema.\n\nKeep business name, address and phone number consistent.\n\nEnsure pages are crawlable and indexable.\n\nMake sure structured data accurately reflects visible page content.\n\nKeep opening hours and business details updated.\n\nImportantly, Google says there are no special AI-only technical requirements or mandatory AI-specific schema markups for appearing in its AI search features. \nGoogle for Developers\n+1\n\nMy recommended priority\n\nIf a local business has limited time or budget, I would build its AI search optimization in this order:\n\nHomepage\n\nHighest priority\n\nEstablish exactly who you are and what you do.\n\nCore service pages\n\nBuild relevance around the services customers search for.\n\nMain location pages\n\nConnect services to the geographical areas you serve.\n\nAbout and credibility pages\n\nDemonstrate expertise, experience and trustworthiness.\n\nPricing and FAQ pages\n\nAnswer specific questions that potential customers ask AI.\n\nCase studies and supporting content\n\nDemonstrate real-world experience and expertise.\n\nOne important strategic distinction: Don't build pages simply because you want AI to mention your business. Build pages that provide the best, most useful answer to the questions your prospective customers are asking.\n\nThat is a much stronger long-term strategy for traditional SEO, local SEO and AI search visibility.\n\nUseful official reference: Google's guide to optimizing for generative AI search.",
      "citations": [
        {
          "title": "Google's guide to optimizing for generative AI search",
          "host": "developers.google.com",
          "url": "https://developers.google.com/search/docs/fundamentals/ai-optimization-guide?version=published&utm_source=chatgpt.com",
          "position": 1
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 1 sources and 1 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-local-business-ai-search",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-local-business-ai-search",
      "published_at": "2026-11-15T13:50:00.000Z",
      "updated_at": "2026-11-15T13:50:00.000Z",
      "revision": 1,
      "slug": "chatgpt-local-business-ai-search",
      "panelId": "local-business-ai-search",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-local-business-ai-search",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "Which pages should a local business optimize for AI search?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:50 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b89d-f7dc-83ea-938e-4b9574ccefc7?temporary-chat=true",
      "mentions": "Not observed",
      "summary": "A local business should optimize the pages that answer the questions customers ask AI search engines, not just its homepage.",
      "answerSnapshot": "：\n\nA local business should optimize the pages that answer the questions customers ask AI search engines, not just its homepage.\n\nFor platforms such as ChatGPT, Google AI Overviews and Perplexity, the goal is to make your business easy to discover, understand, verify and recommend.\n\nI'd prioritize these 10 page types, roughly in order of importance.\n\n1. The essential pages to optimize\n\nPriority\n\n\t\n\nPage\n\n\t\n\nWhat it should target\n\n\n\n\n1\n\n\t\n\nHomepage\n\n\t\n\nBusiness name, services, location, expertise and unique selling points\n\n\n\n\n2\n\n\t\n\nIndividual service pages\n\n\t\n\nEach specific service you provide\n\n\n\n\n3\n\n\t\n\nLocation pages\n\n\t\n\nTowns, cities and neighbourhoods you serve\n\n\n\n\n4\n\n\t\n\nService + location pages\n\n\t\n\nSpecific services in specific locations\n\n\n\n\n5\n\n\t\n\nPricing pages\n\n\t\n\nCosts, estimates, packages and what's included\n\n\n\n\n6\n\n\t\n\nFAQ pages\n\n\t\n\nCommon customer questions and clear answers\n\n\n\n\n7\n\n\t\n\nAbout page\n\n\t\n\nExperience, qualifications, team and credibility\n\n\n\n\n8\n\n\t\n\nCase studies / project pages\n\n\t\n\nReal examples, results and customer outcomes\n\n\n\n\n9\n\n\t\n\nReviews / testimonials\n\n\t\n\nCustomer experiences and evidence of quality\n\n\n\n\n10\n\n\t\n\nContact page\n\n\t\n\nAddress, service area, phone, email and opening hours\n\n2. The most valuable pages: service + location\n\nFor many local businesses, these are the pages with the greatest commercial potential.\n\nFor example, imagine a plumbing company serving Maidenhead, Windsor and Slough.\n\nRather than having one generic plumbing page, it could have:\n\nService pages\n\n/boiler-repair/\n\n/boiler-installation/\n\n/emergency-plumber/\n\n/bathroom-installation/\n\nLocation pages\n\n/plumber-maidenhead/\n\n/plumber-windsor/\n\n/plumber-slough/\n\nService + location pages\n\n/boiler-repair-maidenhead/\n\n/boiler-installation-windsor/\n\n/emergency-plumber-slough/\n\nThe important distinction: each page needs genuinely useful, location-specific information. Simply duplicating a page and changing the town name is not a good strategy.\n\n3. Pages that answer conversational AI queries\n\nAI search users often ask detailed questions rather than typing short keywords.\n\nCreate useful content around questions such as:\n\nCustomer question\n\n\t\n\nRecommended page\n\n\n\n\nHow much does a new boiler cost?\n\n\t\n\nPricing guide\n\n\n\n\nWho offers emergency plumbing in Maidenhead?\n\n\t\n\nEmergency service page\n\n\n\n\nHow long does boiler installation take?\n\n\t\n\nService guide / FAQ\n\n\n\n\nIs it worth repairing an old boiler?\n\n\t\n\nAdvice article\n\n\n\n\nWhat should I look for in a plumber?\n\n\t\n\nExpert advice page\n\n\n\n\nDo you offer weekend appointments?\n\n\t\n\nService information / FAQ\n\nThese pages give AI systems useful, specific information to understand and potentially reference.\n\n4. What every important page should contain\n\nI'd use this structure as a starting point:\n\nClear page title: Identify the service and location.\n\nDirect answer: Explain what you offer in the opening paragraph.\n\nDetailed service information: Explain what's included.\n\nWho it's for: Describe the problems you solve.\n\nPricing information: Give prices or realistic pricing guidance where possible.\n\nLocal relevance: Include genuine local experience and coverage.\n\nEvidence: Qualifications, photographs, case studies and testimonials.\n\nFAQs: Answer 3–6 relevant customer questions.\n\nClear business details: Name, address or service area and contact information.\n\nNext step: Make contacting or booking straightforward.\n\nGoogle's own guidance reinforces that useful original content, crawlability, clear site structure and accurate business information matter for its AI search experiences. \nGoogle for Developers\n+1\n\n5. Don't overlook these external business profiles\n\nAI visibility isn't just about your website.\n\nAlso maintain:\n\nGoogle Business Profile\n\nBing Places\n\nRelevant industry directories\n\nReputable local business directories\n\nReview platforms relevant to your industry\n\nKeep your business name, address, phone number, services and opening hours consistent.\n\nGoogle specifically recommends maintaining accurate Business Profile information for visibility in its AI experiences. \nGoogle for Developers\n+1\n\n6. My recommended implementation order\n\nIf I were planning an AI search optimisation project for a local business, I'd work through this sequence:\n\nFoundation\n\nHomepage, About, Contact and Google Business Profile.\n\nCommercial intent\n\nIndividual service pages and pricing pages.\n\nLocal intent\n\nMain location pages and genuinely useful service + location pages.\n\nQuestion coverage\n\nFAQs, buying guides and problem-solving articles.\n\nTrust and authority\n\nCase studies, testimonials, qualifications and original project content.\n\nTechnical foundations\n\nIndexability, internal linking, structured data where appropriate and mobile usability.\n\nOne important point: don't create hundreds of thin pages just to target AI queries. Google's guidance explicitly favours original, helpful content over mass-produced pages. \nGoogle for Developers\n+1\n\n7. The key takeaway\n\nI'd think about local AI SEO in three layers:\n\nDiscovery pages: Homepage, service pages and location pages.\n\nDecision pages: Pricing, comparisons, FAQs and case studies.\n\nTrust pages: About, reviews, credentials and contact information.\n\nTogether, these help AI systems understand three things:\n\nWhat does this business do?\n\nWhere does it provide those services?\n\nWhy should a customer trust and choose it?\n\nA useful next step: I can build a complete local business AI SEO page map, including recommended URLs, target search queries, page titles and content sections, for a particular business type.",
      "citations": [],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 0 sources and 0 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-b2b-saas-visibility-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-b2b-saas-visibility-repeat-3",
      "published_at": "2026-11-15T13:49:00.000Z",
      "updated_at": "2026-11-15T13:49:00.000Z",
      "revision": 1,
      "slug": "chatgpt-b2b-saas-visibility-repeat-3",
      "panelId": "b2b-saas-visibility",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-b2b-saas-visibility",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "How should a B2B SaaS company measure AI search visibility?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:49 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b7dd-8e9c-83e9-a9e5-3a4d181fc02f?temporary-chat=true",
      "mentions": "Cited",
      "summary": "A B2B SaaS company should measure AI search visibility by how often AI platforms discover, mention, recommend, and cite its brand when potential buyers research solutions — and whether that visibility contributes to pipeline and revenue.",
      "answerSnapshot": "：\n\nA B2B SaaS company should measure AI search visibility by how often AI platforms discover, mention, recommend, and cite its brand when potential buyers research solutions — and whether that visibility contributes to pipeline and revenue.\n\nThe important distinction is that AI search visibility is not simply traditional SEO rankings applied to ChatGPT or Perplexity.\n\nA company can have excellent Google rankings but be almost invisible in AI-generated answers. Equally, it can be frequently recommended by AI platforms without receiving many direct website visits.\n\nI would build a measurement framework around five areas:\n\n1. The five pillars of AI search visibility\n\n1. Brand visibility\n\nHow frequently does your brand appear in AI-generated answers?\n\nBrand mention rate\n\nShare of AI voice\n\nVisibility by AI platform\n\n2. Recommendation and positioning\n\nHow does AI describe your company?\n\nRecommendation frequency\n\nPosition in comparison lists\n\nPositive, neutral or negative sentiment\n\nProduct strengths attributed to your brand\n\n3. Citation and authority\n\nDoes AI use your content as a source?\n\nCitation frequency\n\nShare of citations\n\nPages and domains cited\n\nThird-party sources influencing recommendations\n\n4. Audience and buying-intent coverage\n\nAre you visible for the questions your ideal customers actually ask?\n\nCategory discovery\n\nProblem-oriented searches\n\nCompetitor comparisons\n\nVendor selection and purchase-intent queries\n\n5. Commercial impact\n\nDoes AI visibility translate into business outcomes?\n\nAI-referred sessions\n\nDemo requests and trials\n\nQualified opportunities\n\nInfluenced pipeline\n\nCustomer acquisition\n\n2. Define the metrics and formulas\n\nI would standardise the following metrics so marketing, SEO and leadership all use the same definitions.\n\nMetric\n\n\t\n\nCalculation\n\n\n\n\nAI mention rate\n\n\t\n\nAnswers mentioning your brand ÷ valid answers tested × 100\n\n\n\n\nAI citation rate\n\n\t\n\nAnswers citing your domain ÷ valid answers tested × 100\n\n\n\n\nAI recommendation rate\n\n\t\n\nAnswers recommending your product ÷ valid answers tested × 100\n\n\n\n\nAI share of voice\n\n\t\n\nYour brand mentions ÷ all tracked brand mentions × 100\n\n\n\n\nCompetitive inclusion rate\n\n\t\n\nAnswers including a competitor ÷ valid answers tested × 100\n\n\n\n\nAnswer accuracy\n\n\t\n\nCorrect brand descriptions ÷ answers mentioning your brand × 100\n\n\n\n\nAI referral conversion rate\n\n\t\n\nConversions from AI-referred sessions ÷ AI-referred sessions × 100\n\nA few important distinctions:\n\nMention is not citation. AI can recommend your product without linking to your website.\n\nCitation is not recommendation. AI might cite your documentation while recommending a competitor.\n\nVisibility is not traffic. Someone can discover your brand in ChatGPT without ever visiting your website.\n\nTraffic is not pipeline. A visit does not necessarily indicate buying intent.\n\nThese should remain separate metrics rather than being combined into one opaque AI visibility score. \nSearch Engine Land\n+2\n\n3. Build a representative set of buyer prompts\n\nThis is probably the most important part of the measurement methodology.\n\nDon't just test prompts containing your brand name. That measures existing brand awareness, not whether AI independently discovers your company.\n\nFor a B2B SaaS business, I would create a library of approximately 30–60 prompts initially.\n\nOrganise them by buying intent:\n\nPrompt category\n\n\t\n\nExample\n\n\n\n\nCategory discovery\n\n\t\n\nBest CRM platforms for mid-market B2B companies\n\n\n\n\nProblem awareness\n\n\t\n\nHow can SaaS companies reduce customer churn?\n\n\n\n\nSolution discovery\n\n\t\n\nBest tools for automating customer onboarding\n\n\n\n\nCompetitor comparison\n\n\t\n\nAlternatives to Salesforce for smaller businesses\n\n\n\n\nProduct comparison\n\n\t\n\nHubSpot vs. Pipedrive for B2B sales teams\n\n\n\n\nFeature-specific\n\n\t\n\nSaaS platforms with native Salesforce integration\n\n\n\n\nPurchase intent\n\n\t\n\nBest enterprise customer success platforms\n\n\n\n\nBrand-specific\n\n\t\n\nWhat are the strengths and weaknesses of [Brand]?\n\nI would also segment prompts by:\n\nICP: Enterprise, mid-market and SMB.\n\nPersona: CMO, CTO, RevOps, sales leader, etc.\n\nGeography: UK, US and other priority markets.\n\nProduct category: Different products or use cases.\n\nFunnel stage: Awareness, consideration and decision.\n\nWhy this matters: A company might have strong visibility for generic educational questions but very weak visibility for high-intent vendor comparisons. Those are two very different commercial outcomes.\n\n4. Measure across multiple AI platforms\n\nDon't assume that visibility on one AI platform translates into visibility elsewhere.\n\nChatGPT\n\nBrand recommendations, category discovery and comparisons\n\nGoogle Gemini\n\nBrand mentions, product discovery and source selection\n\nPerplexity\n\nSource citations, research and vendor comparisons\n\nMicrosoft Copilot\n\nEnterprise-oriented discovery and product research\n\nGoogle AI Overviews\n\nAI-enhanced visibility within traditional search\n\nFor each platform, record:\n\nWhether your brand appears.\n\nWhether your brand is recommended.\n\nWhether your website is cited.\n\nWhich competitors appear.\n\nYour position in the response.\n\nWhether the description is accurate.\n\nWhich third-party websites are being cited.\n\nRun the same prompts repeatedly, ideally with multiple runs per prompt, and record the date, country and platform.\n\nA monthly full measurement is a sensible starting point. More frequent sampling of your highest-value prompts can help identify meaningful changes.\n\nImportantly, AI answers vary. Don't interpret a single appearance or disappearance as a genuine visibility trend. \nMiraa\n+1\n\n5. Connect AI visibility to commercial performance\n\nThis is where B2B SaaS companies should differentiate themselves.\n\nA marketing team shouldn't simply report:\n\n\"Our AI visibility increased by 25%.\"\n\nThe more interesting question is:\n\n\"Did increased AI visibility help us generate more qualified demand?\"\n\nI would measure commercial impact across four levels.\n\nAI visibility to revenue measurement\n\nLevel 1: Visibility\n\nShare of voice\n\nMentions, citations and recommendations\n\nLevel 2: Engagement\n\nAI referrals\n\nSessions, landing pages and engagement\n\nLevel 3: Demand generation\n\nQualified leads\n\nDemo requests, trials, MQLs and SQLs\n\nLevel 4: Revenue\n\nPipeline and ARR\n\nOpportunities, influenced pipeline and closed-won revenue\n\nHere are the practical measurement methods:\n\nA. GA4\n\nTrack referral traffic from AI platforms, including:\n\nChatGPT\n\nPerplexity\n\nGemini\n\nClaude\n\nCopilot\n\nMeasure sessions, engagement, landing pages and conversions.\n\nRemember that AI referrals can be misclassified as direct traffic when referrer information is lost. Therefore, referral traffic is a lower-bound estimate of AI-driven website visits, not a complete measurement of AI influence. \nSearch Engine Land\n+1\n\nB. CRM attribution\n\nAdd a question to demo and trial forms:\n\n\"How did you first hear about us?\"\n\nInclude AI assistants as an option.\n\nAlso consider recording:\n\nAI-referred first-touch leads.\n\nAI-influenced opportunities.\n\nAI-influenced pipeline.\n\nAI-influenced closed-won revenue.\n\nKeep first-touch, last-touch and self-reported attribution separate. Don't claim that every deal involving an AI-discovered prospect was caused by AI.\n\nC. Branded search demand\n\nMonitor changes in branded search impressions, clicks and search volume.\n\nIncreases can be a useful supporting indicator of brand awareness generated by AI exposure, although they cannot establish causation on their own.\n\n6. Create an executive AI visibility dashboard\n\nI would keep the executive dashboard relatively simple.\n\nHere is an illustrative example of what it could look like (the figures are hypothetical).\n\nAI Search Performance\n\nIllustrative monthly B2B SaaS dashboard\n\nAI mention rate\n\n38%\n+8 percentage points\n\nAI citation rate\n\n24%\n+5 percentage points\n\nAI share of voice\n\n21%\n+4 percentage points\n\nAI referral sessions\n\n1,250\n+18%\n\nCommercial performance\n\nMetric\n\n\t\n\nMonthly result\n\n\n\n\nAI-referred demo requests\n\n\t\n\n32\n\n\n\n\nAI-referred qualified leads\n\n\t\n\n18\n\n\n\n\nAI-influenced opportunities\n\n\t\n\n12\n\n\n\n\nAI-influenced pipeline\n\n\t\n\n£180,000\n\n\n\n\nAI-influenced closed-won ARR\n\n\t\n\n£24,000\n\nAll figures are fictional and demonstrate dashboard structure, not industry benchmarks.\n\nI would also include:\n\nAI visibility by platform.\n\nVisibility against the top five competitors.\n\nVisibility by buyer-intent category.\n\nMost frequently cited website pages.\n\nTop prompts where competitors appear but your brand doesn't.\n\nAccuracy issues and incorrect product descriptions.\n\n7. Recommended measurement cadence\n\nFrequency\n\n\t\n\nActivity\n\n\n\n\nWeekly\n\n\t\n\nMonitor priority buyer prompts and AI referrals\n\n\n\n\nMonthly\n\n\t\n\nFull prompt-set analysis and competitor benchmarking\n\n\n\n\nQuarterly\n\n\t\n\nReview AI visibility against pipeline and revenue\n\n\n\n\nQuarterly\n\n\t\n\nReassess prompt coverage, ICP and platform priorities\n\nAvoid setting arbitrary industry-wide visibility targets. Establish your own baseline and set improvement targets based on your competitive position, category and commercial priorities. \nSearchable\n+1\n\n8. My recommended KPI hierarchy for a B2B SaaS company\n\nIf I were designing this measurement framework from scratch, I'd prioritise the following:\n\nPriority\n\n\t\n\nKPI\n\n\t\n\nWhy it matters\n\n\n\n\n1\n\n\t\n\nAI share of voice\n\n\t\n\nCompetitive visibility\n\n\n\n\n2\n\n\t\n\nAI recommendation rate\n\n\t\n\nShortlist inclusion\n\n\n\n\n3\n\n\t\n\nAI citation rate\n\n\t\n\nContent authority and retrieval\n\n\n\n\n4\n\n\t\n\nBuyer-intent prompt coverage\n\n\t\n\nCommercial relevance\n\n\n\n\n5\n\n\t\n\nAI referral conversions\n\n\t\n\nMeasurable demand\n\n\n\n\n6\n\n\t\n\nAI-influenced pipeline\n\n\t\n\nBusiness impact\n\n\n\n\n7\n\n\t\n\nAI-influenced ARR\n\n\t\n\nCommercial outcome\n\nThe most important principle is to avoid treating AI search as purely an SEO measurement exercise.\n\nAI visibility is a brand discovery and demand-generation channel. Traditional SEO metrics remain useful, but they don't adequately measure how AI systems influence vendor consideration.\n\n9. Suggested measurement architecture\n\nA practical technology stack could look like this:\n\nRequirement\n\n\t\n\nPotential solution\n\n\n\n\nAI visibility tracking\n\n\t\n\nProfound / Peec AI\n\n\n\n\nSEO and competitive research\n\n\t\n\nSemrush / Ahrefs\n\n\n\n\nWebsite analytics\n\n\t\n\nGoogle Analytics 4\n\n\n\n\nSearch performance\n\n\t\n\nGoogle Search Console\n\n\n\n\nLead and pipeline attribution\n\n\t\n\nHubSpot / Salesforce\n\n\n\n\nExecutive reporting\n\n\t\n\nLooker Studio / Power BI\n\nYou don't necessarily need all of these. A prompt-tracking spreadsheet, GA4 and CRM integration are enough to establish an initial measurement programme.\n\n10. The strategic distinction I'd make\n\nI'd actually separate AI search performance into three different questions:\n\nAre we visible? Are AI platforms mentioning and citing our brand?\n\nAre we competitive? Are we being recommended ahead of, or alongside, our competitors for important buying questions?\n\nAre we commercially successful? Is AI-driven discovery contributing to qualified opportunities and revenue?\n\nThese are different dimensions, and each needs its own KPIs.\n\nMy recommendation would be to make AI share of voice the primary visibility KPI, AI recommendation rate the primary consideration KPI, and AI-influenced pipeline the primary commercial KPI.\n\nThat creates a measurement framework that SEO teams can operate, marketing leaders can report, and commercial leaders can understand.\n\nFor further reading:\n\nSearch Engine Land – Building an AI visibility dashboard\n\nMiraa – Measuring AI search visibility for B2B SaaS\n\nMeasurable – A methodology for measuring AI search visibility",
      "citations": [
        {
          "title": "How to build a meaningful AI search visibility dashboard  · 九月 14, 2026",
          "host": "Search Engine Land",
          "url": "https://searchengineland.com/guide/how-to-build-an-ai-visibility-dashboard?utm_source=chatgpt.com&__cf_chl_rt_tk=iCBR7UCmjoUwu81ScLXXVCQd1o4LXRC9s_wsHcOlnTk-1794750524-1.0.1.1-JiMnAEYPbgE5EM2Rr1SwdmnIfMNF9voDJ5DlM6FpcA0",
          "position": 1
        },
        {
          "title": "How to Measure AI Search Visibility: A B2B SaaS Setup  · 十月 14, 2026",
          "host": "Miraa",
          "url": "https://www.miraa.io/blog/measure-ai-search-visibility?utm_source=chatgpt.com",
          "position": 2
        },
        {
          "title": "How to Measure AI Search Visibility: Citation Rate, Share of Voice and Prominence 八月 28, 2026",
          "host": "Maxlytics",
          "url": "https://www.maxlytics.io/blog/measure-ai-search-visibility?utm_source=chatgpt.com",
          "position": 3
        },
        {
          "title": "How to build a meaningful AI search visibility dashboard  · 九月 14, 2026",
          "host": "Search Engine Land",
          "url": "https://searchengineland.com/guide/how-to-build-an-ai-visibility-dashboard?utm_source=chatgpt.com&__cf_chl_rt_tk=3TnGECdw94XWFZeNeDUwmyWnAH0_XV5O8iFQyF8sZAA-1794750544-1.0.1.1-iTzXxxy6INZKv7IWX1cYDjk.TTwlrdKcE93gQuIGKJk",
          "position": 4
        },
        {
          "title": "How to Measure AI Search Visibility for B2B SaaS 十月 7, 2026",
          "host": "madx.digital",
          "url": "https://www.madx.digital/learn/how-to-measure-ai-search-visibility?utm_source=chatgpt.com",
          "position": 5
        },
        {
          "title": "AI Visibility Metrics and KPIs: What to Track and Report 十一月 5, 2026",
          "host": "Searchable",
          "url": "https://www.searchable.com/blog/ai-visibility-metrics?utm_source=chatgpt.com",
          "position": 6
        },
        {
          "title": "How to Measure AI Search Visibility Chuck Price · 九月 9, 2026",
          "host": "Measurable",
          "url": "https://chuckprice.org/research/measuring-ai-search-visibility/?utm_source=chatgpt.com",
          "position": 7
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 7 sources and 7 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-b2b-saas-visibility-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-b2b-saas-visibility-repeat-2",
      "published_at": "2026-11-15T13:47:00.000Z",
      "updated_at": "2026-11-15T13:47:00.000Z",
      "revision": 1,
      "slug": "chatgpt-b2b-saas-visibility-repeat-2",
      "panelId": "b2b-saas-visibility",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-b2b-saas-visibility",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "How should a B2B SaaS company measure AI search visibility?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:47 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b793-df24-83e9-8f8c-c3daba446d1d?temporary-chat=true",
      "mentions": "Cited",
      "summary": "A B2B SaaS company should measure AI search visibility by how often AI systems discover, mention, recommend, and accurately represent its brand when potential buyers research solutions.",
      "answerSnapshot": "A B2B SaaS company should measure AI search visibility by how often AI systems discover, mention, recommend, and accurately represent its brand when potential buyers research solutions.\n\nThe key is to measure more than brand mentions. A company can appear frequently in AI-generated answers but still fail to influence buying decisions.\n\nI'd recommend building an AI Search Visibility Measurement Framework around five dimensions:\n\nPresence — does your brand appear?\n\nProminence — how prominently is it recommended?\n\nAccuracy — does AI describe your product correctly?\n\nCompetitive position — are competitors mentioned instead?\n\nBusiness impact — does AI visibility generate qualified demand?\n\n1. The core metrics to track\n\nI would use the following scorecard.\n\nMetric\n\n\t\n\nHow to calculate\n\n\t\n\nWhy it matters\n\n\n\n\nBrand mention rate\n\n\t\n\nAnswers mentioning your brand ÷ total answers\n\n\t\n\nMeasures visibility\n\n\n\n\nAI share of voice\n\n\t\n\nYour brand mentions ÷ all competitor brand mentions\n\n\t\n\nMeasures competitive presence\n\n\n\n\nCitation rate\n\n\t\n\nAnswers citing your website ÷ total answers\n\n\t\n\nMeasures source visibility\n\n\n\n\nRecommendation rate\n\n\t\n\nAnswers recommending your product ÷ relevant answers\n\n\t\n\nMeasures buying influence\n\n\n\n\nProminence\n\n\t\n\nFrequency of appearing as a leading recommendation\n\n\t\n\nMeasures strength of positioning\n\n\n\n\nAccuracy\n\n\t\n\nCorrect product descriptions ÷ total descriptions\n\n\t\n\nMeasures message consistency\n\n\n\n\nSentiment\n\n\t\n\nPositive, neutral and negative brand references\n\n\t\n\nMeasures brand perception\n\nTwo important distinctions:\n\nA brand mention is not necessarily a recommendation.\n\nA citation is not necessarily an endorsement.\n\nFor example, ChatGPT might mention your product in a comparison but cite a competitor's website as its source. Both facts should be captured separately.\n\nThese metrics should be calculated separately for each AI platform rather than hiding differences inside one aggregate score. \nSearchable\n+2\n\n2. Build a representative prompt library\n\nThis is arguably the most important part of the measurement process.\n\nDon't simply test prompts containing your brand name. You want to understand whether AI recommends your company when buyers don't already know you exist.\n\nI'd organise prompts around the B2B buying journey:\n\nBuyer intent\n\n\t\n\nExample prompt\n\n\n\n\nCategory discovery\n\n\t\n\nWhat are the best CRM platforms for mid-market businesses?\n\n\n\n\nProblem awareness\n\n\t\n\nHow can a SaaS company reduce customer churn?\n\n\n\n\nSolution discovery\n\n\t\n\nWhat tools help automate customer onboarding?\n\n\n\n\nProduct comparison\n\n\t\n\nCompare the leading customer success platforms.\n\n\n\n\nAlternatives\n\n\t\n\nWhat are the best alternatives to [competitor]?\n\n\n\n\nUse case\n\n\t\n\nWhich platforms support enterprise customer onboarding?\n\n\n\n\nIntegration\n\n\t\n\nWhich customer success platforms integrate with Salesforce?\n\n\n\n\nPurchase evaluation\n\n\t\n\nWhat should I consider when choosing a customer success platform?\n\nI would start with 50–100 prompts, split across:\n\nAwareness, consideration and decision stages.\n\nDifferent buyer personas.\n\nCore product use cases.\n\nCompetitor comparisons.\n\nIndustry-specific requirements.\n\nKeep branded and non-branded prompts separate. Otherwise, you risk overstating visibility because the prompt itself tells the AI to look for your company.\n\n3. Measure across multiple AI platforms\n\nChatGPT\n\nMeasure category recommendations, product comparisons, brand mentions and citations.\n\nGoogle AI Overviews and AI Mode\n\nMeasure visibility in AI-generated search experiences.\n\nPerplexity\n\nMeasure citations, source selection and competitive recommendations.\n\nMicrosoft Copilot\n\nMeasure visibility across research and business-oriented questions.\n\nThe platforms should be measured independently. Their answers and source selection can differ substantially. \nMiraa\n+1\n\n4. Establish a repeatable measurement methodology\n\nAI answers are variable. One test is not enough to establish visibility.\n\nI'd recommend:\n\nCreate a fixed prompt library.\n\nRun prompts across the same AI platforms every week or month.\n\nRepeat selected prompts several times to account for answer variability.\n\nCapture the complete response and all cited URLs.\n\nRecord brand mentions, recommendations, competitors and sentiment.\n\nCompare results over rolling periods.\n\nFor each response, capture:\n\nField\n\n\t\n\nExample\n\n\n\n\nPlatform\n\n\t\n\nChatGPT\n\n\n\n\nPrompt\n\n\t\n\nBest customer success platforms\n\n\n\n\nBrand mentioned\n\n\t\n\nYes\n\n\n\n\nBrand recommended\n\n\t\n\nYes\n\n\n\n\nCitation\n\n\t\n\nYes\n\n\n\n\nCited URL\n\n\t\n\nYour product comparison page\n\n\n\n\nCompetitors mentioned\n\n\t\n\nCompetitor A, B\n\n\n\n\nPosition\n\n\t\n\nFirst recommendation\n\n\n\n\nDescription accurate\n\n\t\n\nYes\n\nKeep the raw responses. This allows you to audit the numbers and understand why visibility changed.\n\n5. Connect AI visibility to commercial performance\n\nThis is where B2B SaaS measurement becomes particularly interesting.\n\nAI visibility is a marketing indicator, not a revenue metric by itself.\n\nYou need to connect three measurement layers:\n\nLayer 1\n\nAI visibility\n\nBrand mentions\n\nShare of voice\n\nRecommendations\n\nCitation frequency\n\nLayer 2\n\nDemand generation\n\nAI-referred website sessions\n\nDemo requests\n\nBranded search growth\n\nDirect traffic trends\n\nLayer 3\n\nRevenue impact\n\nMarketing-qualified leads\n\nSales-qualified opportunities\n\nPipeline influenced\n\nClosed-won revenue\n\nSome practical approaches:\n\nA. Track AI referral traffic\n\nUse GA4 to identify traffic from sources such as ChatGPT and Perplexity. Track landing pages, engagement, demo conversions and assisted conversions.\n\nRemember that some AI-influenced visits may appear as direct traffic or be attributed to another channel.\n\nB. Add an AI discovery question to demo forms\n\nFor example:\n\n\"How did you first hear about us?\"\n\nInclude an option such as:\n\nChatGPT or another AI assistant\n\nGoogle search\n\nSocial media\n\nRecommendation\n\nOther\n\nC. Track AI-influenced pipeline\n\nConnect self-reported discovery, referral data and CRM records to opportunities.\n\nDon't automatically attribute all subsequent revenue to AI. Distinguish between sourced, influenced and merely associated pipeline.\n\nmadx.digital\n+2\n6. Build an executive AI visibility dashboard\n\nI would keep the executive dashboard relatively simple.\n\nHere's an illustrative example of what it could contain.\n\nAI Search Visibility Dashboard\n\nIllustrative data — not actual company performance\n\nBrand mention rate\n\n42%\n\n+8 percentage points\n\nAI share of voice\n\n28%\n\n+5 percentage points\n\nCitation rate\n\n19%\n\n+3 percentage points\n\nRecommendation rate\n\n31%\n\n+6 percentage points\n\nVisibility by platform\n\nIllustrative percentage of tracked prompts mentioning the brand.\n\n0%\n15%\n30%\n45%\n60%\nChatGPT\nPerplexity\nGemini\nCopilot\nChatGPT: Mention rate 48%\nPerplexity: Mention rate 39%\nGemini: Mention rate 35%\nCopilot: Mention rate 26%\n\nCommercial performance\n\nMetric\n\n\t\n\nIllustrative result\n\n\n\n\nAI referral sessions\n\n\t\n\n1,250\n\n\n\n\nAI-referred demo requests\n\n\t\n\n42\n\n\n\n\nAI-sourced opportunities\n\n\t\n\n12\n\n\n\n\nAI-influenced pipeline\n\n\t\n\n£180,000\n\nAll figures are hypothetical and intended to demonstrate dashboard design.\n7. What should the reporting cadence look like?\n\nFrequency\n\n\t\n\nActivity\n\n\n\n\nWeekly\n\n\t\n\nMonitor prompts, mentions and citations\n\n\n\n\nMonthly\n\n\t\n\nAnalyse share of voice and competitors\n\n\n\n\nMonthly\n\n\t\n\nReview AI traffic and conversion trends\n\n\n\n\nQuarterly\n\n\t\n\nEvaluate pipeline influence and strategic progress\n\nI would avoid setting arbitrary industry-wide targets. Establish a baseline first, then set targets based on your category, competitors and commercial objectives. \nSearchable\n+1\n\n8. My recommended KPI hierarchy for B2B SaaS\n\nIf I were designing this for a SaaS CMO, I would prioritise the metrics as follows:\n\nPriority\n\n\t\n\nKPI\n\n\t\n\nPurpose\n\n\n\n\n1\n\n\t\n\nAI share of voice\n\n\t\n\nCompetitive visibility\n\n\n\n\n2\n\n\t\n\nRecommendation rate\n\n\t\n\nBuyer consideration\n\n\n\n\n3\n\n\t\n\nCitation rate\n\n\t\n\nSource authority\n\n\n\n\n4\n\n\t\n\nBrand mention rate\n\n\t\n\nBrand discovery\n\n\n\n\n5\n\n\t\n\nAnswer accuracy\n\n\t\n\nBrand positioning\n\n\n\n\n6\n\n\t\n\nAI referral conversions\n\n\t\n\nDemand generation\n\n\n\n\n7\n\n\t\n\nAI-influenced pipeline\n\n\t\n\nCommercial impact\n\nOne important strategic point: Don't optimise solely for getting your website cited. AI systems can recommend a brand based on information from third-party review sites, comparison articles, industry publications and other sources.\n\nTherefore, measure two distinct things:\n\nOwned visibility: How often AI references your website and content.\n\nEarned visibility: How often AI references independent sources that mention or recommend your company.\n\nBoth matter.\n\n9. A practical 90-day implementation plan\n\nDays 1–30: Establish the baseline\n\nIdentify 3–5 main competitors.\n\nBuild 50–100 buyer prompts.\n\nSelect the AI platforms to monitor.\n\nRun the initial measurement.\n\nDocument mention, citation and recommendation rates.\n\nDays 31–60: Identify visibility gaps\n\nIdentify prompts where competitors appear but you don't.\n\nAnalyse which third-party websites AI cites.\n\nIdentify inaccurate product descriptions.\n\nPrioritise content and digital PR opportunities.\n\nDays 61–90: Measure improvement\n\nRepeat the original prompt tests.\n\nCompare share of voice and recommendation rates.\n\nAnalyse referral traffic and demo conversions.\n\nReport progress to marketing leadership.\n\nFinal recommendation\n\nThe biggest mistake would be treating AI search visibility as another SEO ranking report.\n\nFor B2B SaaS, the real objective is:\n\nBecome one of the products AI recommends when a potential customer asks which solution they should buy.\n\nI would use AI share of voice as the primary visibility KPI, recommendation rate as the primary consideration KPI, and AI-influenced pipeline as the commercial outcome.\n\nTogether, these provide a much more meaningful picture than simply counting how often ChatGPT mentions your brand.",
      "citations": [
        {
          "title": "AI Visibility Metrics and KPIs: What to Track and Report 十一月 5, 2026",
          "host": "Searchable",
          "url": "https://www.searchable.com/blog/ai-visibility-metrics?utm_source=chatgpt.com",
          "position": 1
        },
        {
          "title": "How to Measure AI Search Visibility: A B2B SaaS Setup  · 十月 14, 2026",
          "host": "Miraa",
          "url": "https://www.miraa.io/blog/measure-ai-search-visibility?utm_source=chatgpt.com",
          "position": 2
        },
        {
          "title": "How to Measure AI Search Visibility a Practical Framework  · 八月 28, 2026",
          "host": "Verbatim Digital",
          "url": "https://verbatimdigital.com/articles/how-to-measure-ai-searc-visibility?utm_source=chatgpt.com",
          "position": 3
        },
        {
          "title": "Measuring AI Visibility: Metrics and Method for B2B 九月 10, 2026",
          "host": "andweekly.com",
          "url": "https://www.andweekly.com/en/perspectives/measuring-ai-visibility?utm_source=chatgpt.com",
          "position": 4
        },
        {
          "title": "How to Measure AI Search Visibility for B2B SaaS 十月 7, 2026",
          "host": "madx.digital",
          "url": "https://www.madx.digital/learn/how-to-measure-ai-search-visibility?utm_source=chatgpt.com",
          "position": 5
        },
        {
          "title": "How to Track AI Visibility: Tools and Metrics for B2B Teams  · 六月 5, 2026",
          "host": "Rudo",
          "url": "https://rudo.co.uk/insights/articles/how-to-track-ai-visibility/?utm_source=chatgpt.com",
          "position": 6
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 6 sources and 6 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-b2b-saas-visibility",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-b2b-saas-visibility",
      "published_at": "2026-11-15T13:45:00.000Z",
      "updated_at": "2026-11-15T13:45:00.000Z",
      "revision": 1,
      "slug": "chatgpt-b2b-saas-visibility",
      "panelId": "b2b-saas-visibility",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-b2b-saas-visibility",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "How should a B2B SaaS company measure AI search visibility?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:45 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b76b-c098-83ea-9b3f-a5420480286b?temporary-chat=true",
      "mentions": "Cited",
      "summary": "When a potential customer asks ChatGPT, Perplexity, Google AI Overviews, or another AI search engine about a problem your product solves, does your company appear, how is it represented, and does that visibility generate pipeline?",
      "answerSnapshot": "A B2B SaaS company should measure AI search visibility by answering one fundamental question:\n\nWhen a potential customer asks ChatGPT, Perplexity, Google AI Overviews, or another AI search engine about a problem your product solves, does your company appear, how is it represented, and does that visibility generate pipeline?\n\nThe key is to move beyond traditional SEO metrics such as rankings, impressions and organic clicks.\n\nAI search visibility needs to be measured across five dimensions:\n\nPresence: Does your brand appear in AI-generated answers?\n\nPositioning: How prominently and positively is it mentioned?\n\nCompetitive share: How often are competitors recommended instead?\n\nAuthority: Which sources influence AI-generated recommendations?\n\nBusiness impact: Does AI visibility translate into website visits, leads and revenue?\n\nHere's how I'd structure a measurement framework for a B2B SaaS business.\n\n1. Build a measurement scorecard\n\nI'd use the following metrics, grouped into four layers.\n\nMetric\n\n\t\n\nHow to measure\n\n\t\n\nWhat it tells you\n\n\n\n\nAI mention rate\n\n\t\n\nAnswers mentioning your brand ÷ answers tested\n\n\t\n\nBrand presence\n\n\n\n\nAI share of voice\n\n\t\n\nYour brand mentions ÷ all brand mentions\n\n\t\n\nCompetitive visibility\n\n\n\n\nCitation rate\n\n\t\n\nAnswers citing your website ÷ answers tested\n\n\t\n\nContent authority\n\n\n\n\nRecommendation rate\n\n\t\n\nCommercial answers recommending your product ÷ commercial answers tested\n\n\t\n\nShortlist inclusion\n\n\n\n\nBrand accuracy\n\n\t\n\nCorrect product descriptions ÷ descriptions audited\n\n\t\n\nMessage consistency\n\n\n\n\nSentiment\n\n\t\n\nPositive, neutral or negative descriptions\n\n\t\n\nBrand perception\n\n\n\n\nAI referral traffic\n\n\t\n\nSessions from identifiable AI platforms\n\n\t\n\nMeasurable demand\n\n\n\n\nAI-sourced pipeline\n\n\t\n\nQualified opportunities associated with AI discovery\n\n\t\n\nCommercial impact\n\nI would avoid rolling all these into one opaque AI visibility score. Each metric answers a different question.\n\nIn particular, distinguish three outcomes:\n\nMention: AI knows your brand exists.\n\nCitation: AI uses your website as a source.\n\nRecommendation: AI actively suggests your product as a solution.\n\nA company can perform well on one and poorly on the others.\n\n2. Measure visibility using actual buyer questions\n\nThis is arguably the most important part of the methodology.\n\nDon't simply track generic keywords such as \"CRM software\" or \"marketing automation\".\n\nBuild a repeatable prompt library based on how your target buyers research software.\n\nFor example, if you sell B2B marketing automation:\n\nBuyer journey\n\n\t\n\nExample AI search prompt\n\n\n\n\nProblem awareness\n\n\t\n\nHow can a B2B SaaS company improve marketing attribution?\n\n\n\n\nCategory research\n\n\t\n\nWhat are the best marketing automation platforms for B2B SaaS?\n\n\n\n\nProduct comparison\n\n\t\n\nHubSpot alternatives for mid-market B2B companies\n\n\n\n\nFeature evaluation\n\n\t\n\nWhich platforms offer advanced lead scoring and intent data?\n\n\n\n\nIntegration research\n\n\t\n\nMarketing automation tools that integrate with Salesforce\n\n\n\n\nPurchase intent\n\n\t\n\nBest marketing automation platform for a 100-person SaaS business\n\nI'd start with approximately 40–60 prompts, divided across:\n\nCategory and discovery questions\n\nProblem-solving questions\n\nCompetitor comparison questions\n\nAlternative and replacement questions\n\nIntegration and feature questions\n\nPricing and buying questions\n\nKeep the prompts consistent between measurement periods.\n\nWhich AI platforms should you measure?\n\nChatGPT\n\nMeasure brand mentions, product recommendations and cited sources.\n\nGoogle Gemini and AI Search\n\nMeasure discovery and visibility in Google's generative search experiences.\n\nPerplexity\n\nMeasure recommendations and source citations.\n\nMicrosoft Copilot\n\nMeasure visibility in an additional AI discovery environment relevant to business users.\n\nDon't assume visibility on one platform translates to another. Measure them separately.\n\n3. Calculate AI share of voice against competitors\n\nFor B2B SaaS, competitive share of voice is one of the most valuable metrics.\n\nFor example, imagine you track 100 relevant AI answers:\n\nIllustrative AI share of voice\n\nShare of brand mentions across a sample of AI-generated answers.\n\n0%\n8%\n16%\n24%\n32%\nYour company\nCompetitor A\nCompetitor B\nCompetitor C\nOthers\nYour company: Share of voice 24%\nCompetitor A: Share of voice 32%\nCompetitor B: Share of voice 21%\nCompetitor C: Share of voice 15%\nOthers: Share of voice 8%\n\nIllustrative data, not an industry benchmark.\n\nFormula:\n\n𝐴\n𝐼\n \n𝑆\n𝑜\n𝑉\n=\nYour brand mentions\nAll brand mentions\n×\n100\nAI SoV=\nAll brand mentions\nYour brand mentions\n\t​\n\n×100\n\nI'd also calculate share of voice separately for:\n\nCategory discovery\n\nCompetitor comparisons\n\nAlternative searches\n\nHigh-intent buying questions\n\nIndividual product features\n\nThis helps identify where you're winning and where competitors are dominating.\n\nFor example, you might have strong visibility for educational questions but almost no visibility when someone asks AI to recommend the best enterprise solution.\n\nThat is a much more actionable finding than an overall visibility score.\n\n4. Measure the sources driving AI recommendations\n\nThis is an important dimension that many SaaS companies overlook.\n\nKnowing that ChatGPT recommends your company is useful. Knowing why it recommends your company is even more valuable.\n\nTrack:\n\nSource category\n\n\t\n\nWhat to investigate\n\n\n\n\nYour website\n\n\t\n\nProduct pages, use cases, integrations, documentation\n\n\n\n\nReview platforms\n\n\t\n\nG2, Capterra and TrustRadius\n\n\n\n\nEditorial content\n\n\t\n\nIndustry publications and software comparisons\n\n\n\n\nCommunity discussions\n\n\t\n\nReddit, specialist forums and professional communities\n\n\n\n\nPartner websites\n\n\t\n\nIntegration partners and technology ecosystems\n\nFor every AI citation, capture:\n\nSource domain\n\nSpecific URL\n\nSource type\n\nCompetitor or brand referenced\n\nNumber of tracked answers citing that source\n\nThis creates a source influence map.\n\nFor example, you might discover that third-party comparison articles are cited much more frequently than your own product pages.\n\nThat suggests a distribution and digital PR opportunity, rather than simply publishing more blog content.\n\n5. Connect AI visibility to pipeline and revenue\n\nUltimately, the CMO and CEO will want to know whether AI search is contributing to commercial performance.\n\nI'd create a separate attribution dashboard.\n\nCommercial KPI\n\n\t\n\nMeasurement\n\n\n\n\nAI referral sessions\n\n\t\n\nVisits from identifiable AI platforms\n\n\n\n\nAI referral conversions\n\n\t\n\nDemo requests and sign-ups from those visits\n\n\n\n\nAI-influenced leads\n\n\t\n\nLeads reporting AI as a discovery source\n\n\n\n\nAI-influenced opportunities\n\n\t\n\nOpportunities associated with AI discovery\n\n\n\n\nAI-sourced pipeline\n\n\t\n\nPipeline from leads attributed to AI\n\n\n\n\nAI-sourced revenue\n\n\t\n\nClosed-won revenue from AI-attributed leads\n\nThere are two important attribution considerations.\n\nFirst: AI referral traffic is undercounted.\n\nSome AI experiences don't pass reliable referral information. Some visits will appear as direct traffic.\n\nUse:\n\nGA4 referral tracking\n\nLanding-page analysis\n\nCRM source attribution\n\nA self-reported attribution field on demo forms: \"How did you first hear about us?\"\n\nSecond: distinguish sourced from influenced pipeline.\n\nFor example:\n\nAI-sourced: A prospect discovers your company through an AI assistant and subsequently visits your website.\n\nAI-influenced: A prospect already knows your company but uses AI to compare vendors before requesting a demo.\n\nThese should be reported separately to avoid overstating AI's contribution.\n\n6. Build a practical monthly reporting dashboard\n\nI'd structure the executive dashboard around four sections.\n\nAI Search Visibility\n\nIllustrative monthly executive dashboard\n\nAI mention rate\n38%\n\n+6 percentage points\n\nAI share of voice\n24%\n\n+4 percentage points\n\nCitation rate\n19%\n\n+3 percentage points\n\nRecommendation rate\n27%\n\n+5 percentage points\n\nBusiness impact\n\n420\nAI referral sessions\n18\nAI referral leads\n£85k\nAI-sourced pipeline\n£22k\nAI-sourced revenue\n\nAll figures are hypothetical examples, not expected performance or benchmarks.\n\nI'd report the following:\n\nReporting frequency\n\n\t\n\nFocus\n\n\n\n\nWeekly\n\n\t\n\nVisibility changes, new citations, competitor movements\n\n\n\n\nMonthly\n\n\t\n\nShare of voice, recommendations, source analysis, referral traffic\n\n\n\n\nQuarterly\n\n\t\n\nPipeline contribution, revenue influence, strategic progress\n\n7. Which tools should you use?\n\nI'd avoid buying too many overlapping tools.\n\nA sensible measurement stack would be:\n\nTool\n\n\t\n\nPurpose\n\n\n\n\nGoogle Search Console\n\n\t\n\nGoogle organic and generative search performance\n\n\n\n\nGoogle Analytics 4\n\n\t\n\nAI referral traffic and conversions\n\n\n\n\nAhrefs\n\n\t\n\nSEO visibility and competitive research\n\n\n\n\nSemrush\n\n\t\n\nSEO and AI visibility research\n\n\n\n\nOtterlyAI\n\n\t\n\nAI answer monitoring\n\n\n\n\nProfound\n\n\t\n\nAI search visibility and brand monitoring\n\n\n\n\nHubSpot\n\n\t\n\nLead, opportunity and revenue attribution\n\nOne particularly relevant development: Google introduced dedicated generative AI performance reporting in Search Console in 2026, covering features such as AI Overviews and AI Mode. \nGoogle for Developers\n+1\n\nGoogle's guidance also cautions against assuming that third-party tools have access to its internal AI ranking signals.\n\n8. My recommended measurement framework\n\nIf I were building this programme from scratch for a B2B SaaS company, I'd prioritise the following:\n\nThe B2B SaaS AI Visibility Framework\n\nVisibility\n\nBrand mention rate\n\nCitation rate\n\nPrompt coverage\n\nCompetitive positioning\n\nAI share of voice\n\nRecommendation rate\n\nCompetitive displacement\n\nBrand quality and authority\n\nSentiment\n\nProduct positioning accuracy\n\nCitation source distribution\n\nCommercial performance\n\nAI referral traffic\n\nAI-generated leads\n\nAI-influenced opportunities\n\nAI-sourced pipeline and revenue\n\nA practical 90-day implementation\n\nPeriod\n\n\t\n\nWhat to accomplish\n\n\n\n\nDays 1–30\n\n\t\n\nEstablish baseline, select 40–60 prompts, identify competitors and measure platforms\n\n\n\n\nDays 31–60\n\n\t\n\nIdentify visibility gaps, analyse citation sources and improve priority content\n\n\n\n\nDays 61–90\n\n\t\n\nRe-measure, compare competitive performance and connect results to CRM pipeline\n\nDon't set arbitrary industry-wide targets before establishing your baseline. AI visibility varies considerably by category, platform, geography and prompt selection.\n\nThe most important strategic recommendation\n\nI'd make one distinction clear to the leadership team:\n\nAI search visibility is not the same as AI search performance.\n\nVisibility tells you whether AI engines recognise and recommend your brand.\n\nPerformance tells you whether that recognition creates commercial value.\n\nA B2B SaaS company should ultimately optimise for:\n\nBeing included in relevant AI-generated answers.\n\nBeing recommended during vendor evaluation.\n\nBeing accurately positioned against competitors.\n\nBeing cited by credible sources.\n\nGenerating measurable qualified pipeline and revenue.\n\nThe strongest north-star KPI would be AI recommendation share across high-intent buyer prompts, supported by AI-influenced pipeline as the commercial outcome.\n\nThat is a more meaningful measure of competitive advantage than simply counting how often ChatGPT mentions your company.\n\nFor further reading: 2026 State of AI Visibility for B2B SaaS and Google's guide to generative AI search optimisation.",
      "citations": [
        {
          "title": "Introducing Search Generative AI performance reports in Search Console | Google Search Central Blog |   · 六月 3, 2026",
          "host": "Google for Developers",
          "url": "https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports?hl=en&utm_source=chatgpt.com",
          "position": 1
        },
        {
          "title": "Google's Guide to Optimizing for Generative AI Features on Google Search | Google Search Central | Documentation |   · 七月 10, 2026",
          "host": "Google for Developers",
          "url": "https://developers.google.com/search/docs/fundamentals/ai-optimization-guide?autm_content=blog_emails&utm_source=chatgpt.com&hl=zh-cn",
          "position": 2
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 2 sources and 2 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-mention-vs-citation-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-mention-vs-citation-repeat-3",
      "published_at": "2026-11-15T13:45:00.000Z",
      "updated_at": "2026-11-15T13:45:00.000Z",
      "revision": 1,
      "slug": "chatgpt-mention-vs-citation-repeat-3",
      "panelId": "mention-vs-citation",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-mention-vs-citation",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "What is the difference between an AI mention and an AI citation?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:45 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b750-2b3c-83e9-9ffe-3fd3ea55f10f?temporary-chat=true",
      "mentions": "Not observed",
      "summary": "The difference between an AI mention and an AI citation is whether AI simply refers to your brand or website, or actually uses it as a source to support its answer.",
      "answerSnapshot": "The difference between an AI mention and an AI citation is whether AI simply refers to your brand or website, or actually uses it as a source to support its answer.\n\nBoth are important in Generative Engine Optimisation (GEO) and AI search visibility.\n\n1. AI mention\n\nAn AI mention happens when an AI tool mentions your brand, company, product or website in its response.\n\nExample:\n\nA user asks ChatGPT: \"What are the best project management tools?\"\n\nChatGPT responds:\n\n\"Popular project management tools include Asana, Trello, Monday.com and ClickUp.\"\n\nIf your brand is Asana, this is an AI mention.\n\nYour brand name appears in the answer.\n\nThe AI recognises or recommends your brand.\n\nYour website does not necessarily receive a link.\n\nThe AI may not have used your website as a source.\n\n2. AI citation\n\nAn AI citation happens when an AI tool references your website or content as a source for information in its answer.\n\nExample:\n\nA user asks: \"What are the main benefits of project management software?\"\n\nChatGPT responds:\n\n\"Project management software improves team collaboration, task visibility and workflow efficiency.\"\n\nIt then links to your website as the source of this information.\n\nYour website is used as a source.\n\nYour content supports the AI-generated answer.\n\nYou may receive a clickable link.\n\nIt demonstrates that your content is being recognised as a useful information source.\n\n3. Key differences\n\nFactor\n\n\t\n\nAI mention\n\n\t\n\nAI citation\n\n\n\n\nBrand name appears\n\n\t\n\nYes\n\n\t\n\nNot necessarily\n\n\n\n\nWebsite referenced\n\n\t\n\nNot necessarily\n\n\t\n\nYes\n\n\n\n\nClickable link\n\n\t\n\nNot required\n\n\t\n\nOften included\n\n\n\n\nBrand visibility\n\n\t\n\nYes\n\n\t\n\nYes\n\n\n\n\nContent used as a source\n\n\t\n\nNot necessarily\n\n\t\n\nYes\n\n\n\n\nPotential referral traffic\n\n\t\n\nIndirect\n\n\t\n\nDirect\n\n4. Which is more important for SEO and GEO?\n\nIdeally, you want both.\n\nAI mentions help build brand awareness, recognition and authority.\n\nAI citations help establish your website as a trusted information source and can generate referral traffic.\n\nFor a successful GEO strategy, you should monitor both metrics.\n\nA simple way to remember\n\nAI mention\n\n\"AI knows your brand.\"\n\nAI citation\n\n\"AI uses your content as a source.\"\n\nMy recommendation: When measuring GEO performance, track AI mentions, AI citations, citation share, brand sentiment and referral traffic separately. This gives you a much clearer picture of your brand's visibility across ChatGPT, Gemini, Perplexity and Google AI Overviews.",
      "citations": [],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 0 sources and 0 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-mention-vs-citation",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-mention-vs-citation",
      "published_at": "2026-11-15T13:38:00.000Z",
      "updated_at": "2026-11-15T13:38:00.000Z",
      "revision": 1,
      "slug": "chatgpt-mention-vs-citation",
      "panelId": "mention-vs-citation",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-mention-vs-citation",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "What is the difference between an AI mention and an AI citation?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:38 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b5c9-ae18-83e9-b00e-975c50dcaaab?temporary-chat=true",
      "mentions": "Not observed",
      "summary": "The difference between an AI mention and an AI citation is how an AI tool refers to a brand, business, website, or source in its answer.",
      "answerSnapshot": "The difference between an AI mention and an AI citation is how an AI tool refers to a brand, business, website, or source in its answer.\n\nBoth are important in AI visibility and Generative Engine Optimisation (GEO), but they measure different things.\n\n1. AI mention\n\nAn AI mention happens when an AI tool names your brand, business, product, or website in its response.\n\nIt doesn't necessarily link to your website or identify it as a source.\n\nExample:\n\nA user asks ChatGPT:\n\n\"Which are the best project management tools?\"\n\nChatGPT responds:\n\n\"Popular project management tools include Asana, Monday.com, Trello and ClickUp.\"\n\nIf your brand is Asana, this is an AI mention.\n\nYour brand is named.\n\nYour brand is visible in the AI-generated answer.\n\nThere may be no link to your website.\n\nYour website may not have been used as a source.\n\n2. AI citation\n\nAn AI citation happens when an AI tool references your website or content as a source supporting its answer.\n\nExample:\n\nA user asks:\n\n\"What are the benefits of project management software?\"\n\nChatGPT responds:\n\n\"Project management software helps teams improve collaboration, manage deadlines and track progress.\"\n\nIt then references an article on your website as a source, with a clickable link.\n\nYour website is referenced as a source.\n\nYour content helps support the AI-generated answer.\n\nUsers may click through to your website.\n\nIt indicates that your content has been used as a supporting source.\n\n3. Key differences\n\nMeasurement\n\n\t\n\nAI mention\n\n\t\n\nAI citation\n\n\n\n\nBrand name appears\n\n\t\n\nYes\n\n\t\n\nNot necessarily\n\n\n\n\nWebsite linked\n\n\t\n\nNot required\n\n\t\n\nUsually\n\n\n\n\nContent used as a source\n\n\t\n\nNot necessarily\n\n\t\n\nYes\n\n\n\n\nBrand awareness\n\n\t\n\nYes\n\n\t\n\nYes\n\n\n\n\nPotential referral traffic\n\n\t\n\nIndirect\n\n\t\n\nDirect\n\n\n\n\nIndicates source authority\n\n\t\n\nNot necessarily\n\n\t\n\nMore directly\n\n4. Which is more important for SEO and GEO?\n\nBoth are important, but they serve different purposes.\n\nAI mentions\n\nBuild brand awareness\n\nThey help your brand become recognised and included in AI-generated recommendations.\n\nAI citations\n\nBuild content visibility and credibility\n\nThey indicate that AI systems are referencing your content to support their answers.\n\n5. A practical GEO example\n\nImagine you run a digital marketing agency.\n\nA user asks ChatGPT: \"Who are the best digital marketing agencies in the UK?\"\n\nYou might see three scenarios:\n\nMention: ChatGPT names your agency in its list of recommended agencies.\n\nCitation: ChatGPT links to your website as a source when discussing digital marketing services.\n\nMention + citation: ChatGPT names your agency and links to your website.\n\nThe third scenario is particularly valuable because you gain both brand exposure and source visibility.\n\nMy recommendation: When measuring GEO performance, track both AI mentions and AI citations. Also monitor how frequently your brand appears in relevant answers, which pages are cited, and which AI platforms generate the visibility.\n\nThe ultimate goal is not simply to get mentioned or cited, but to make your brand a recognised and trusted answer in your industry.",
      "citations": [],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 0 sources and 0 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-mention-vs-citation-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-mention-vs-citation-repeat-2",
      "published_at": "2026-11-15T13:38:00.000Z",
      "updated_at": "2026-11-15T13:38:00.000Z",
      "revision": 1,
      "slug": "chatgpt-mention-vs-citation-repeat-2",
      "panelId": "mention-vs-citation",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-mention-vs-citation",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "What is the difference between an AI mention and an AI citation?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:38 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b5dc-a2c0-83ea-bf01-999da7f8bb26?temporary-chat=true",
      "mentions": "Not observed",
      "summary": "The difference between an AI mention and an AI citation is how an AI tool refers to a brand, business, website, or source when generating an answer.",
      "answerSnapshot": "The difference between an AI mention and an AI citation is how an AI tool refers to a brand, business, website, or source when generating an answer.\n\nBoth are important in AI visibility and Generative Engine Optimisation (GEO), but they measure different things.\n\n1. AI mention\n\nAn AI mention occurs when an AI tool names your brand, business, product, or website in its response.\n\nIt doesn't necessarily link to or reference your website.\n\nExample:\n\nA user asks ChatGPT:\n\n\"Which are the best CRM platforms for small businesses?\"\n\nChatGPT responds:\n\n\"Popular CRM platforms include HubSpot, Salesforce, Zoho and Pipedrive.\"\n\nIf your brand is HubSpot, this is an AI mention.\n\nYour brand name appears.\n\nThe AI recognises or recommends your brand.\n\nThere doesn't need to be a link to your website.\n\n2. AI citation\n\nAn AI citation occurs when an AI tool references your website or content as a source supporting its answer.\n\nExample:\n\nChatGPT responds:\n\n\"HubSpot offers CRM tools designed to help small businesses manage customer relationships.\"\n\nIt then references or links to a page on HubSpot's website as its source.\n\nYour website is used as a source.\n\nThe AI attributes information to your content.\n\nA clickable link or source reference may be displayed.\n\n3. Key differences\n\nMeasurement\n\n\t\n\nAI mention\n\n\t\n\nAI citation\n\n\n\n\nBrand name appears\n\n\t\n\nYes\n\n\t\n\nNot necessarily\n\n\n\n\nWebsite linked\n\n\t\n\nNot required\n\n\t\n\nUsually\n\n\n\n\nContent used as a source\n\n\t\n\nNot required\n\n\t\n\nYes\n\n\n\n\nBrand awareness\n\n\t\n\nStrong indicator\n\n\t\n\nCan contribute\n\n\n\n\nWebsite authority\n\n\t\n\nIndirect signal\n\n\t\n\nStronger source attribution\n\n\n\n\nPotential referral traffic\n\n\t\n\nNot necessarily\n\n\t\n\nPossible\n\n4. Can you have one without the other?\n\nYes. There are four possible scenarios:\n\nBoth\n\nBrand mentioned and website cited\n\nAI names your business and links to your website.\n\nMention only\n\nBrand mentioned without a citation\n\nAI recommends your business but doesn't link to your website.\n\nCitation only\n\nWebsite cited without a brand mention in the answer\n\nAI uses a page from your website as a source, but doesn't explicitly name your brand in its response.\n\nNeither\n\nNo visibility\n\nAI neither mentions your brand nor cites your website.\n\n5. Which is more important for GEO?\n\nIdeally, you want both.\n\nAI mentions help build brand awareness, recognition and consideration.\n\nAI citations demonstrate that AI systems are using your website as a source of information.\n\nFor a strong GEO strategy, I'd recommend tracking:\n\nAI mention frequency – how often your brand appears in AI answers.\n\nAI citation frequency – how often your website is referenced as a source.\n\nShare of AI voice – how often you appear compared with competitors.\n\nCitation quality – which pages are being cited and for which topics.\n\nSentiment – whether AI describes your brand positively, neutrally or negatively.\n\nThe simplest way to remember it:\n\nAn AI mention means AI talks about you.\n\nAn AI citation means AI uses your content as a source.\n\nThe strongest GEO strategy aims to make your brand both a recognised answer and a trusted source.",
      "citations": [],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 0 sources and 0 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-verify-ai-citation-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-verify-ai-citation-repeat-3",
      "published_at": "2026-11-15T13:38:00.000Z",
      "updated_at": "2026-11-15T13:38:00.000Z",
      "revision": 1,
      "slug": "chatgpt-verify-ai-citation-repeat-3",
      "panelId": "verify-ai-citation",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-verify-ai-citation",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "How can a publisher verify that an AI answer cited its source correctly?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:38 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b5a4-90fc-83ea-b1a5-4008d7a88372?temporary-chat=true",
      "mentions": "Not observed",
      "summary": "A publisher can verify whether an AI answer cited its source correctly by conducting a citation audit: checking whether the cited source actually supports the AI's statement, whether the information was represented accurately, and whether readers can trace the claim back to the original publication.",
      "answerSnapshot": "A publisher can verify whether an AI answer cited its source correctly by conducting a citation audit: checking whether the cited source actually supports the AI's statement, whether the information was represented accurately, and whether readers can trace the claim back to the original publication.\n\nImportantly, a citation being present does not necessarily mean it is correct. An AI system might cite a genuine article but misrepresent its findings, take a sentence out of context, or attribute information to the wrong source.\n\n1. A practical verification process\n\nStep\n\n\t\n\nWhat the publisher should check\n\n\n\n\n1. Locate the citation\n\n\t\n\nIdentify the URL, article, author and publication date.\n\n\n\n\n2. Find the original passage\n\n\t\n\nLocate the exact paragraph, data point or statement supporting the AI's answer.\n\n\n\n\n3. Check factual accuracy\n\n\t\n\nCompare the AI's claim with the original source.\n\n\n\n\n4. Check context\n\n\t\n\nEnsure the AI hasn't omitted qualifications, caveats or contradictory findings.\n\n\n\n\n5. Check attribution\n\n\t\n\nConfirm that the information is attributed to the correct author or publisher.\n\n\n\n\n6. Check the link\n\n\t\n\nEnsure the URL opens the intended article, rather than a homepage, unrelated page or broken link.\n\n\n\n\n7. Record the result\n\n\t\n\nDocument whether the citation is accurate, partially accurate or incorrect.\n\n2. Use a claim-to-source verification framework\n\nThe most reliable approach is to break an AI answer into individual factual claims and match each claim against its cited source.\n\nFor example:\n\nExample AI answer\n\n\"Research shows that 72% of consumers trust publishers more than AI-generated content.\"\n\nPublisher's verification\n\nDoes the article exist?\n\nYes\n\nDoes it contain the 72% figure?\n\nNo\n\nDoes it support the stated conclusion?\n\nUnverified\n\nAudit conclusion\n\nThe citation may point to a legitimate publication, but it does not substantiate the AI's specific claim.\n\nThis distinction is essential: a correct URL is not necessarily a correct citation.\n\n3. Classify citation quality\n\nPublishers can use a simple scoring framework:\n\nClassification\n\n\t\n\nMeaning\n\n\n\n\nAccurate\n\n\t\n\nSource directly supports the claim.\n\n\n\n\nPartial\n\n\t\n\nSource supports some, but not all, of the claim.\n\n\n\n\nMisleading\n\n\t\n\nSource is real but misrepresented or taken out of context.\n\n\n\n\nIncorrect\n\n\t\n\nSource does not support the claim.\n\n\n\n\nFabricated\n\n\t\n\nCitation or source cannot be verified.\n\nFor larger publishers, this can become a repeatable quality-assurance process.\n\n4. Automate the verification process\n\nPublishers can build or adopt systems that:\n\nExtract citations and URLs from AI-generated answers.\n\nMatch individual claims to passages in the original articles.\n\nCompare dates, figures, names and quotations.\n\nDetect citations pointing to irrelevant content.\n\nFlag unsupported claims for human review.\n\nTrack how frequently their content is cited accurately by different AI platforms.\n\nSome useful technologies and approaches include:\n\nRetrieval-Augmented Generation (RAG): Helps identify which source passages were retrieved to support an answer.\n\nNatural Language Inference (NLI): Evaluates whether a source passage supports, contradicts or is unrelated to a claim.\n\nSemantic search: Finds relevant passages even when the AI paraphrases the original text.\n\nHuman editorial review: Provides the final judgement where context or interpretation matters.\n\n5. Measure citation accuracy across AI platforms\n\nA publisher could regularly test its content against platforms such as ChatGPT, Google Gemini, and Perplexity.\n\nUseful performance indicators include:\n\nMetric\n\n\t\n\nWhat it measures\n\n\n\n\nCitation accuracy\n\n\t\n\nPercentage of citations that genuinely support their claims.\n\n\n\n\nCitation coverage\n\n\t\n\nPercentage of factual claims accompanied by sources.\n\n\n\n\nAttribution accuracy\n\n\t\n\nWhether the correct publisher and author receive credit.\n\n\n\n\nLink accuracy\n\n\t\n\nWhether citations lead to the correct original content.\n\n\n\n\nContext accuracy\n\n\t\n\nWhether the AI preserves the original meaning.\n\n\n\n\nCitation visibility\n\n\t\n\nHow frequently the publisher is cited in relevant AI answers.\n\n6. The bigger opportunity for publishers\n\nThere is an important distinction between two questions:\n\nIs the AI citing my content correctly?\n\nIs the AI giving my publication appropriate recognition and attribution?\n\nThese are related but different issues.\n\nA publisher may find that its research is being used extensively by AI systems, but:\n\nIts publication is not credited.\n\nIts journalists are not named.\n\nIts findings are attributed to another organisation.\n\nIts content is paraphrased in a way that changes the original meaning.\n\nThis creates an opportunity for publishers to develop an ongoing AI citation monitoring and verification framework.\n\nMy recommendation: Build a publisher-facing AI citation audit that captures the AI platform, answer, cited URL, original source passage, claim-level accuracy, attribution accuracy and verification outcome. This would give publishers measurable evidence of how AI systems use and represent their journalism.",
      "citations": [],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 0 sources and 0 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-verify-ai-citation-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-verify-ai-citation-repeat-2",
      "published_at": "2026-11-15T13:37:00.000Z",
      "updated_at": "2026-11-15T13:37:00.000Z",
      "revision": 1,
      "slug": "chatgpt-verify-ai-citation-repeat-2",
      "panelId": "verify-ai-citation",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-verify-ai-citation",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "How can a publisher verify that an AI answer cited its source correctly?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:37 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b578-f0f4-83ea-b425-de828e7ea3dd?temporary-chat=true",
      "mentions": "Cited",
      "summary": "A publisher can verify whether an AI answer cited its source correctly by conducting a citation audit: checking whether the cited page actually supports the AI's claims, whether the information has been represented accurately, and whether the citation leads readers to the original material.",
      "answerSnapshot": "A publisher can verify whether an AI answer cited its source correctly by conducting a citation audit: checking whether the cited page actually supports the AI's claims, whether the information has been represented accurately, and whether the citation leads readers to the original material.\n\nThe key distinction is that a citation can be technically valid (the link works and points to the right page) but factually misleading (the page does not support the statement attributed to it).\n\n1. A practical five-step verification process\n\n1. Break the AI answer into individual claims.\nDon't assess an entire paragraph as one statement. Identify each factual claim and the citation attached to it.\n\n2. Open the original source.\nCheck that the link works and leads to the correct article, report or research paper. Verify:\n\nArticle title and publisher.\n\nAuthor, where available.\n\nPublication and update dates.\n\nWhether the cited page is the original source or a secondary reference.\n\n3. Match the claim to the evidence.\nFind the exact passage that supports the AI's statement. Check whether the AI has:\n\nQuoted the source accurately.\n\nPreserved the original meaning.\n\nIncluded relevant qualifications and context.\n\nAvoided presenting an opinion or inference as an established fact.\n\n4. Check completeness and accuracy.\nDetermine whether the AI has omitted important information, misunderstood statistics, or used an outdated version of the article.\n\n5. Cross-check important claims.\nCompare the statement with independent sources, preferably original research, official datasets or primary documents.\n\nThese checks are consistent with guidance from the University of Oxford's Bodleian Libraries on evaluating AI-generated references. \nlibguides.bodleian.ox.ac.uk\n+1\n\n2. Use a citation verification scorecard\n\nA publisher can introduce a standard audit template:\n\nVerification criterion\n\n\t\n\nResult\n\n\n\n\nDoes the URL work?\n\n\t\n\nPass / Fail\n\n\n\n\nIs it the correct article?\n\n\t\n\nPass / Fail\n\n\n\n\nIs the publisher correctly identified?\n\n\t\n\nPass / Fail\n\n\n\n\nDoes the source support the claim?\n\n\t\n\nFull / Partial / No\n\n\n\n\nIs the information represented accurately?\n\n\t\n\nYes / No\n\n\n\n\nIs the information current?\n\n\t\n\nYes / No\n\n\n\n\nIs sufficient context included?\n\n\t\n\nYes / No\n\nFor an overall assessment, classify each citation:\n\nVerified The source directly supports the claim.\n\nPartially supported The source supports some, but not all, of the claim.\n\nMisattributed The source does not support the claim.\n\nUnverifiable The source cannot be accessed or confirmed.\n\n3. Test the AI answer using controlled prompts\n\nPublishers should create a repeatable test programme.\n\nFor example:\n\nAsk an AI system a question about a specific article.\n\nRecord the complete answer and its citations.\n\nRepeat the same question across different AI platforms.\n\nRepeat the test at different times.\n\nCompare the citations against the original article.\n\nKeep records of the prompts, dates, answers, URLs and verification results.\n\nThis helps distinguish occasional citation errors from recurring attribution problems.\n\n4. Monitor how AI platforms attribute publisher content\n\nPublishers should also measure:\n\nHow frequently their articles are cited.\n\nWhether citations link to the correct pages.\n\nWhether the publisher is credited visibly.\n\nWhether links generate referral traffic.\n\nWhether AI summaries accurately represent the original reporting.\n\nIn the UK, this is becoming particularly relevant. The Competition and Markets Authority's Google Publisher Conduct Requirement addresses accurate attribution, access to original publisher content and greater transparency around AI search features. \n政府资产出版服务\n+2\n\n5. The most important distinction\n\nThree separate questions must be answered:\n\nCitation validity\n\nDoes the citation lead to a genuine, correct source?\n\nCitation accuracy\n\nDoes that source actually support the AI's statement?\n\nAttribution fairness\n\nHas the publisher received appropriate credit and a usable link to its content?\n\nMy recommendation: Publishers should prioritise claim-level verification rather than simply counting citations or measuring how often their domains appear in AI answers. A hundred citations do not necessarily mean a hundred correctly attributed claims.\n\nThe most meaningful measure is the percentage of AI-generated claims for which the cited original source provides accurate, directly verifiable evidence.",
      "citations": [
        {
          "title": "Oxford LibGuides: AI for literature searching: Evaluating AI-generated references and information",
          "host": "libguides.bodleian.ox.ac.uk",
          "url": "https://libguides.bodleian.ox.ac.uk/ai-for-literature-searching/evaluating-ai-references?utm_source=chatgpt.com",
          "position": 1
        },
        {
          "title": "Google Search's Guidance on Generative AI Content on Your Website | Google Search Central | Documentation |",
          "host": "Google for Developers",
          "url": "https://developers.google.com/search/docs/fundamentals/using-gen-ai-content?utm_source=chatgpt.com&hl=zh-cn",
          "position": 2
        },
        {
          "title": "Interpretative notes 六月 2, 2026",
          "host": "政府资产出版服务",
          "url": "https://assets.publishing.service.gov.uk/media/6a1ec397050971fbebf3bd42/_Publisher_CR_Interpretative_Notes_.pdf?utm_source=chatgpt.com",
          "position": 3
        },
        {
          "title": "Publisher CR 4 August 2026 八月 4, 2026",
          "host": "政府资产出版服务",
          "url": "https://assets.publishing.service.gov.uk/media/6a7195d7aec8358a34958bfa/Publisher_CR_4_Aug_2026.pdf?utm_source=chatgpt.com",
          "position": 4
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 4 sources and 4 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-source-freshness-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-source-freshness-repeat-3",
      "published_at": "2026-11-15T13:36:00.000Z",
      "updated_at": "2026-11-15T13:36:00.000Z",
      "revision": 1,
      "slug": "chatgpt-source-freshness-repeat-3",
      "panelId": "source-freshness",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-source-freshness",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "How do AI search engines decide whether information is fresh enough?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:36 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b534-65d0-83ea-9785-209ab9c87f96?temporary-chat=true",
      "mentions": "Cited",
      "summary": "AI search engines decide whether information is fresh enough by estimating how recently it was published or updated, how quickly the topic changes, and whether newer information is available.",
      "answerSnapshot": "AI search engines decide whether information is fresh enough by estimating how recently it was published or updated, how quickly the topic changes, and whether newer information is available.\n\nImportantly, there isn't one universal freshness rule. A page published yesterday might be outdated, while a five-year-old page might still be perfectly accurate.\n\nMost AI search engines use a combination of signals.\n\n1. The main freshness signals\n\nSignal\n\n\t\n\nWhat the AI looks for\n\n\n\n\nPublication date\n\n\t\n\nWhen the content first appeared\n\n\n\n\nLast updated date\n\n\t\n\nWhen the page was most recently revised\n\n\n\n\nSource activity\n\n\t\n\nWhether the website regularly publishes new information\n\n\n\n\nRecent developments\n\n\t\n\nWhether newer events contradict the information\n\n\n\n\nSearch engine indexing\n\n\t\n\nWhen search crawlers last discovered or revisited the page\n\n\n\n\nContent changes\n\n\t\n\nWhether meaningful information has changed, rather than just the page timestamp\n\nFor example, changing a page's footer from 2025 to 2026 doesn't necessarily make its information fresh.\n\n2. Freshness depends on the topic\n\nAI search engines effectively need to understand the expected lifespan of information.\n\nFor example:\n\nBreaking news\n\nMinutes to hours\nA major announcement or developing event can become outdated within minutes.\n\nTechnology and AI\n\nDays to months\nSoftware releases, AI models and technology prices change frequently.\n\nBusiness information\n\nWeeks to months\nCompany leadership, financial results and product offerings need periodic updates.\n\nHistorical facts\n\nYears or decades\nHistorical events and established mathematical principles rarely need updating.\n\nThese are illustrative timeframes, not fixed industry rules.\n\nGoogle explicitly describes this concept through its Query Deserves Freshness systems: searches about recent earthquakes or newly released films should favour newer information. \nGoogle for Developers\n\n3. How AI search engines calculate freshness\n\nMany AI search systems use a process similar to this:\n\nIdentify query intent: Is the user asking about something current or something timeless?\n\nFind relevant documents: Retrieve potentially useful pages from search indexes or connected databases.\n\nCheck timestamps: Examine publication dates, modification dates and available indexing information.\n\nEvaluate content changes: Determine whether the page contains genuinely new information.\n\nApply a recency score: Give newer information more weight when appropriate.\n\nBalance freshness with reliability: Avoid choosing a recent but unreliable source over an older, authoritative one.\n\nOne technical approach is exponential time decay:\n\n𝐹\n(\n𝑡\n)\n=\n𝑒\n−\n𝜆\n𝑡\nF(t)=e\n−λt\n\nWhere:\n\n𝐹\n(\n𝑡\n)\nF(t) = freshness score.\n\n𝑡\nt = age of the information.\n\n𝜆\nλ = how quickly freshness decreases.\n\nA larger \n𝜆\nλ means information becomes outdated more quickly.\n\nFor example, a system might use a short freshness period for breaking news and a much longer one for historical research.\n\nMicrosoft documents a similar approach in Azure AI Search, where freshness influences ranking without automatically excluding older documents. \nMicrosoft Learn\n\n4. Why crawling and indexing matter\n\nThere is another important distinction:\n\nRecently updated does not necessarily mean recently discovered.\n\nAn AI search engine can only retrieve updated information if its underlying search infrastructure has discovered and indexed that information.\n\nThe process generally looks like this:\n\nWebsite publishes updated information\n\nSearch crawler discovers the update\n\nSearch engine updates its index\n\nAI retrieves the updated information\n\nAI generates an answer using its sources\n\nGoogle explains this crawling, indexing and serving process in its official documentation. \nGoogle for Developers\n\n5. The biggest challenge: distinguishing genuine freshness from fake freshness\n\nConsider two websites:\n\nWebsite A publishes an article today but uses statistics from three years ago.\n\nWebsite B publishes an article two months ago with genuinely updated research and data.\n\nWebsite A is technically newer, but Website B may contain fresher information.\n\nThis is why sophisticated retrieval systems should consider more than publication dates.\n\nThey need to consider:\n\nWhether the underlying facts have changed.\n\nWhether the information is supported by reliable sources.\n\nWhether newer evidence contradicts older claims.\n\nWhether the page has been substantially updated.\n\n6. The key takeaway\n\nAI search engines don't simply ask, \"How old is this webpage?\"\n\nThey effectively ask:\n\n\"Given what the user is searching for, how current does this information need to be, and can we trust that it is still accurate?\"\n\nFreshness is therefore a combination of time, relevance, reliability and indexing.\n\nFor further reading:\n\nGoogle – Search ranking systems\n\nMicrosoft Bing – How search results are delivered\n\nMicrosoft – Freshness-aware AI retrieval",
      "citations": [
        {
          "title": "A Guide to Google Search Ranking Systems | Google Search Central | Documentation |",
          "host": "Google for Developers",
          "url": "https://developers.google.com/search/docs/appearance/ranking-systems-guide?ashby_jid=af2e43fb-f98a-4db3-9ad1-ae223933c50f&utm_source=chatgpt.com&hl=zh-cn",
          "position": 1
        },
        {
          "title": "Configure Freshness-Aware Retrieval - Azure AI Search Docs · 九月 17, 2026",
          "host": "Microsoft Learn",
          "url": "https://learn.microsoft.com/en-us/azure/search/agentic-retrieval-how-to-configure-freshness?utm_source=chatgpt.com",
          "position": 2
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 2 sources and 2 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-verify-ai-citation",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-verify-ai-citation",
      "published_at": "2026-11-15T13:36:00.000Z",
      "updated_at": "2026-11-15T13:36:00.000Z",
      "revision": 1,
      "slug": "chatgpt-verify-ai-citation",
      "panelId": "verify-ai-citation",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-verify-ai-citation",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "How can a publisher verify that an AI answer cited its source correctly?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:36 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b559-d994-83ea-8018-612bcb56f290?temporary-chat=true",
      "mentions": "Not observed",
      "summary": "A publisher can verify whether an AI answer cited its source correctly by performing a citation audit: checking whether the cited page actually supports the AI's claims, whether the information is represented accurately, and whether readers can trace the answer back to the original publication.",
      "answerSnapshot": "A publisher can verify whether an AI answer cited its source correctly by performing a citation audit: checking whether the cited page actually supports the AI's claims, whether the information is represented accurately, and whether readers can trace the answer back to the original publication.\n\nImportantly, a citation being present does not necessarily mean it is correct. An AI might cite a genuine article but misrepresent its findings, attribute a claim to the wrong author, or use information outside the article's original context.\n\n1. The five checks publishers should perform\n\nCheck\n\n\t\n\nWhat to verify\n\n\n\n\nSource accuracy\n\n\t\n\nDoes the cited URL lead to the actual article or document?\n\n\n\n\nClaim accuracy\n\n\t\n\nDoes the source explicitly support the AI's statement?\n\n\n\n\nContext accuracy\n\n\t\n\nHas the AI preserved the original meaning and qualifications?\n\n\n\n\nAttribution accuracy\n\n\t\n\nAre the author, publisher, publication date and title correct?\n\n\n\n\nCitation completeness\n\n\t\n\nCan readers access the source and locate the relevant passage?\n\n2. A practical verification workflow\n\nSuppose an AI answer says:\n\n\"Research by Publisher X found that 68% of consumers prefer independent news sources.\"\n\nThe publisher should:\n\nOpen the cited URL. Confirm that it points to the correct article, not a homepage, unrelated page or broken link.\n\nFind the original claim. Search the article for the statistic or statement.\n\nCompare the wording. Check whether the AI has reproduced the finding accurately.\n\nCheck the surrounding context. Was the statistic based on a particular country, age group, survey or time period?\n\nCheck attribution. Confirm that the research, quotation or conclusion belongs to the cited publisher.\n\nRecord the result. Mark the citation as accurate, partially accurate, misleading or incorrect.\n\n3. Build an AI citation audit spreadsheet\n\nPublishers can use a simple tracking system:\n\nField\n\n\t\n\nExample\n\n\n\n\nAI platform\n\n\t\n\nChatGPT\n\n\n\n\nAI answer\n\n\t\n\nOriginal response\n\n\n\n\nCited URL\n\n\t\n\nPublisher article\n\n\n\n\nClaim extracted\n\n\t\n\n68% prefer independent news\n\n\n\n\nSupporting passage\n\n\t\n\nParagraph 6\n\n\n\n\nAccuracy\n\n\t\n\nAccurate\n\n\n\n\nContext preserved\n\n\t\n\nYes\n\n\n\n\nAudit date\n\n\t\n\n15 November 2026\n\nThis makes it possible to monitor citation quality over time and identify recurring problems.\n\n4. Automate citation verification\n\nFor publishers monitoring hundreds or thousands of AI answers, the process can be partially automated.\n\nA verification system could:\n\nCollect AI answers and their cited URLs.\n\nExtract individual factual claims.\n\nRetrieve the original source pages.\n\nMatch claims against relevant passages.\n\nFlag unsupported or contradictory statements.\n\nSend uncertain cases for human review.\n\nAn important distinction: automated text similarity is not enough. A source can contain similar words without actually supporting the claim.\n\n5. Measure citation quality\n\nPublishers can introduce a few useful performance indicators:\n\nMetric\n\n\t\n\nWhat it measures\n\n\n\n\nCitation accuracy\n\n\t\n\nPercentage of citations that genuinely support their claims\n\n\n\n\nAttribution accuracy\n\n\t\n\nPercentage with correct author and publisher attribution\n\n\n\n\nContext preservation\n\n\t\n\nPercentage retaining the original meaning\n\n\n\n\nBroken citation rate\n\n\t\n\nPercentage of inaccessible or incorrect links\n\n\n\n\nCitation coverage\n\n\t\n\nPercentage of significant factual claims with supporting sources\n\n6. What publishers should do when they find an error\n\nIf an AI answer misrepresents their content, publishers should:\n\nPreserve a screenshot or record of the AI answer.\n\nRecord the cited URL and the original supporting passage.\n\nDocument precisely what is inaccurate.\n\nReport the issue through the AI platform's feedback or reporting mechanism.\n\nMonitor whether the error recurs.\n\nPublishers should also maintain clear article metadata, canonical URLs, author information, publication dates and accessible source material. These help AI systems identify and attribute content correctly, although they cannot guarantee accurate citations.\n\nThe bigger opportunity: independent AI citation monitoring\n\nThere is an important distinction between two questions:\n\nIs the AI citing the publisher?\n\nIs the AI citing the publisher correctly?\n\nThe first measures visibility. The second measures attribution quality and editorial integrity.\n\nA publisher could therefore develop an independent monitoring system that regularly tests AI platforms using representative questions, captures their answers, checks cited URLs and evaluates whether the cited material supports the claims.\n\nMy recommendation: Treat AI citation verification as a new form of editorial fact-checking, rather than simply an SEO monitoring exercise. The objective is not just to get cited more often, but to ensure that when an AI uses a publisher's journalism, it represents that journalism faithfully.",
      "citations": [],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 0 sources and 0 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-source-freshness",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-source-freshness",
      "published_at": "2026-11-15T13:35:00.000Z",
      "updated_at": "2026-11-15T13:35:00.000Z",
      "revision": 1,
      "slug": "chatgpt-source-freshness",
      "panelId": "source-freshness",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-source-freshness",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "How do AI search engines decide whether information is fresh enough?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:35 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b4f3-3108-83ea-bc15-13a8def2c9f6?temporary-chat=true",
      "mentions": "Cited",
      "summary": "AI search engines don't use one universal definition of \"fresh enough.\" Instead, they estimate whether information is still relevant, accurate, and up to date for the particular question being asked.",
      "answerSnapshot": "AI search engines don't use one universal definition of \"fresh enough.\" Instead, they estimate whether information is still relevant, accurate, and up to date for the particular question being asked.\n\nThe key idea is that freshness is relative to the topic and the user's intent.\n\nFor example:\n\nWeather forecasts: information from yesterday may already be outdated.\n\nTechnology news: information from last week might be useful, but breaking news needs something much newer.\n\nHistorical events: information from ten years ago can still be perfectly fresh.\n\nMedical or scientific research: newer evidence may matter, but a recent article isn't automatically more reliable.\n\n1. The main signals AI search engines use\n1. Publication and update dates\n\nSearch engines examine signals such as:\n\nOriginal publication date.\n\nLast updated date.\n\nStructured metadata, such as datePublished and dateModified.\n\nSitemap update information.\n\nHTTP last-modified headers.\n\nHowever, dates aren't always trustworthy. A website could change its displayed date without actually updating its information.\n\n2. How frequently the information changes\n\nDifferent subjects have different freshness requirements.\n\nFor example:\n\nTopic\n\n\t\n\nTypical freshness requirement\n\n\n\n\nBreaking news\n\n\t\n\nMinutes to hours\n\n\n\n\nStock market information\n\n\t\n\nMinutes\n\n\n\n\nSoftware documentation\n\n\t\n\nWhen versions change\n\n\n\n\nProduct pricing\n\n\t\n\nDays or hours\n\n\n\n\nScientific research\n\n\t\n\nWhen significant evidence emerges\n\n\n\n\nHistorical facts\n\n\t\n\nOften years or decades\n\nThese are illustrative expectations, not fixed search-engine rules.\n\n3. How recently the page was crawled\n\nSearch engines track when they last discovered or revisited a webpage.\n\nThey may consider:\n\nWhen the page was last crawled.\n\nWhether its content has changed since the previous crawl.\n\nWhether the website regularly publishes new information.\n\nWhether the page is likely to change frequently.\n\nAn important distinction: a recently crawled page isn't necessarily recently updated.\n\n4. Whether the actual information has changed\n\nAI search systems need to distinguish between genuine updates and cosmetic changes.\n\nFor example:\n\nArticle A was published in 2024 and contains 2024 statistics.\n\nArticle B was published in 2024 but updated in 2026 with new statistics and research.\n\nArticle C was published in 2026 but simply republishes old information.\n\nArticle B might be the most useful, even though Article C has the newest publication date.\n\n5. The nature of the search query\n\nThis is particularly important.\n\nGoogle describes a concept called Query Deserves Freshness (QDF).\n\nIts ranking systems recognize that certain searches have a stronger need for recent information.\n\nFor example:\n\nHigh freshness requirement\n\n\"Who won the election yesterday?\"\n\nRecent, verified reporting is essential.\n\nMedium freshness requirement\n\n\"What are the best programming languages?\"\n\nRecent developments matter, but foundational information remains useful.\n\nLow freshness requirement\n\n\"What is the Pythagorean theorem?\"\n\nAn older, authoritative explanation is just as useful as a new one.\n\nGoogle for Developers\n+1\n2. How AI search engines combine these signals\n\nA simplified model looks like this:\n\nUser's search query\n\nWhat information is being requested?\n\nFreshness requirements\n\nHow quickly does this topic become outdated?\n\nSource evaluation\n\nDates, crawl history, content changes and credibility\n\nRanked information\n\nRelevant, credible and sufficiently current sources\n\nThe exact weighting differs between search engines, and companies generally don't publish their complete ranking formulas.\n\n3. How this applies to AI answer engines\n\nDifferent AI search products expose different freshness signals.\n\nPlatform\n\n\t\n\nRelevant approach\n\n\n\n\nGoogle Search / AI Overviews\n\n\t\n\nUses freshness systems for queries that deserve recent information\n\n\n\n\nPerplexity\n\n\t\n\nProvides publication and update-date information in its search results and API\n\n\n\n\nChatGPT Search\n\n\t\n\nRetrieves web information to help answer questions about current events and changing facts\n\n\n\n\nClaude with web search\n\n\t\n\nCan use web search results that include page-age information\n\nGoogle for Developers\n+1\n4. The biggest takeaway\n\nThere are three different concepts that are easy to confuse:\n\nRecency: How recently was the information published?\n\nFreshness: How accurately does it reflect the current situation?\n\nReliability: How much confidence should we place in the information?\n\nThe best AI search systems try to balance all three.\n\nA recently published article can be inaccurate. An older article can remain completely correct.\n\nUltimately, the goal isn't to find the newest information. It's to find the most current information that is relevant and trustworthy for the question being asked.\n\nFor further reading: Google's guide to search ranking systems.",
      "citations": [
        {
          "title": "A Guide to Google Search Ranking Systems | Google Search Central | Documentation |",
          "host": "Google for Developers",
          "url": "https://developers.google.com/search/docs/appearance/ranking-systems-guide?%20=undefined&utm_source=chatgpt.com&hl=zh-cn",
          "position": 1
        },
        {
          "title": "How Does Google Determine Ranking Results How Search Works",
          "host": "Google Search",
          "url": "https://www.google.com/search/howsearchworks/how-search-works/ranking-results/?device=c&utm_source=chatgpt.com",
          "position": 2
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 2 sources and 2 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-source-freshness-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-source-freshness-repeat-2",
      "published_at": "2026-11-15T13:35:00.000Z",
      "updated_at": "2026-11-15T13:35:00.000Z",
      "revision": 1,
      "slug": "chatgpt-source-freshness-repeat-2",
      "panelId": "source-freshness",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-source-freshness",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "How do AI search engines decide whether information is fresh enough?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:35 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b510-24e4-83ea-a9a1-f47a2cbc6313?temporary-chat=true",
      "mentions": "Cited",
      "summary": "AI search engines don't usually have one universal rule for deciding whether information is \"fresh enough.\" Instead, they estimate freshness using several signals, depending on the question being asked.",
      "answerSnapshot": "AI search engines don't usually have one universal rule for deciding whether information is \"fresh enough.\" Instead, they estimate freshness using several signals, depending on the question being asked.\n\nFor example, information published yesterday might be essential for a question about today's weather, but completely unnecessary for a question about how gravity works.\n\nThe key idea is freshness is relative to the topic and the user's intent.\n\n1. The main signals AI search engines use\n\nSignal\n\n\t\n\nWhat the AI looks at\n\n\n\n\nPublication date\n\n\t\n\nWhen the article or page was published\n\n\n\n\nLast updated date\n\n\t\n\nWhen the content was most recently revised\n\n\n\n\nCrawl date\n\n\t\n\nWhen the search engine last discovered or indexed the page\n\n\n\n\nEvent date\n\n\t\n\nWhen the information actually happened\n\n\n\n\nSource authority\n\n\t\n\nWhether the source is reliable and regularly maintained\n\n\n\n\nConsistency\n\n\t\n\nWhether other independent sources confirm the information\n\n\n\n\nHistorical relevance\n\n\t\n\nWhether older information is still valid\n\nAn important distinction: a page updated today isn't necessarily reporting something that happened today. AI systems ideally distinguish between publication time, update time, and the actual date of an event.\n\n2. Query intent determines the freshness requirement\n\nAI search engines first try to understand whether the question needs recent information.\n\nFor example:\n\nBreaking news\n\nVery high freshness\n\n\"What happened in the UK today?\"\n\nInformation from the last few hours may be most useful.\n\nTechnology\n\nHigh freshness\n\n\"What are the latest AI models?\"\n\nInformation from recent weeks or months matters.\n\nEducation\n\nModerate freshness\n\n\"What is photosynthesis?\"\n\nOlder, well-established explanations remain useful.\n\nHistorical information\n\nLow freshness\n\n\"What caused the First World War?\"\n\nOriginal historical evidence and accuracy matter more than publication date.\n\nGoogle describes a related concept called Query Deserves Freshness (QDF). Its systems identify searches where users are likely to expect recent information and adjust results accordingly. \nGoogle for Developers\n+1\n\n3. How freshness is calculated\n\nMany search systems use a time-decay model.\n\nThe basic principle is that newer information receives a higher freshness score, while older information gradually loses that advantage.\n\nOne simplified mathematical example is:\n\n𝐹\n(\n𝑡\n)\n=\n𝑒\n−\n𝜆\n𝑡\nF(t)=e\n−λt\n\nWhere:\n\n𝐹\n(\n𝑡\n)\nF(t) = freshness score.\n\n𝑡\nt = time elapsed since publication or update.\n\n𝜆\nλ = decay rate.\n\nA high decay rate means information becomes outdated quickly. A low decay rate means information stays useful for longer.\n\nThis is a simplified illustration, not Google's actual ranking formula.\n\nMicrosoft Azure AI Search, for example, offers configurable freshness policies that boost newer documents without automatically excluding older, relevant documents. \nMicrosoft Learn\n\n4. How AI search engines avoid misleading freshness signals\n\nThere are several challenges:\n\nFake updates: A website changes its publication date without meaningfully updating its content.\n\nOld information republished: An old article appears with a recent date.\n\nDelayed indexing: A search engine hasn't yet discovered a newly published article.\n\nConflicting reports: Different sources publish different versions of an unfolding story.\n\nEvergreen information: A reliable older source may still be more accurate than a recent article.\n\nGoogle, for example, uses multiple signals to estimate publication and modification dates rather than trusting a single date field. \nGoogle for Developers\n+1\n\n5. The final decision: freshness versus relevance\n\nA useful way to think about AI search ranking is:\n\nUser's search query\n\nUnderstand search intent\n\nDoes the question require current information?\n\nRetrieve relevant sources\n\nAssess dates, reliability and relevance\n\nRank and cross-check\n\nBalance freshness with accuracy\n\nGenerate an answer\n\nUse supporting sources and communicate uncertainty\n\nThe most important takeaway: AI search engines shouldn't simply select the newest information. They should select the most relevant and reliable information that is sufficiently current for the question.\n\nFor further reading:\n\nGoogle's explanation of freshness in search rankings.\n\nHow Google determines publication dates.\n\nMicrosoft's freshness-aware AI retrieval system.",
      "citations": [
        {
          "title": "A Guide to Google Search Ranking Systems | Google Search Central | Documentation |",
          "host": "Google for Developers",
          "url": "https://developers.google.com/search/docs/appearance/ranking-systems-guide?content_language=English&utm_source=chatgpt.com&hl=zh-cn",
          "position": 1
        },
        {
          "title": "Query Deserves Freshness: what it is and how it works  · 十一月 28, 2025",
          "host": "Search Engine Land",
          "url": "https://searchengineland.com/guide/query-deserves-freshness-qdf?utm_source=chatgpt.com",
          "position": 2
        },
        {
          "title": "Configure Freshness-Aware Retrieval - Azure AI Search Docs · 九月 17, 2026",
          "host": "Microsoft Learn",
          "url": "https://learn.microsoft.com/en-us/azure/search/agentic-retrieval-how-to-configure-freshness?utm_source=chatgpt.com",
          "position": 3
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 3 sources and 3 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-official-documentation-preference-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-official-documentation-preference-repeat-2",
      "published_at": "2026-11-15T13:34:00.000Z",
      "updated_at": "2026-11-15T13:34:00.000Z",
      "revision": 1,
      "slug": "chatgpt-official-documentation-preference-repeat-2",
      "panelId": "official-documentation-preference",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-official-documentation-preference",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "Do AI search engines prefer official documentation?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:34 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b4cc-5384-83ea-933e-b84e4deb032d?temporary-chat=true",
      "mentions": "Not observed",
      "summary": "Yes. AI search engines generally favour official documentation as a source, particularly for technical questions, but it depends on the search engine, the query and the quality of the documentation.",
      "answerSnapshot": "Yes. AI search engines generally favour official documentation as a source, particularly for technical questions, but it depends on the search engine, the query and the quality of the documentation.\n\nWhy AI search engines favour official documentation\n\nFirst-party authority: Official documentation comes directly from the organisation responsible for a product, API or service.\n\nTechnical accuracy: It is usually the most reliable source for API parameters, configuration options, product features and supported functionality.\n\nStructured information: Documentation often uses clear headings, tables, code examples and consistent terminology, making it easier for AI systems to retrieve relevant information.\n\nReduced ambiguity: Official documentation is less likely to contain outdated community interpretations or conflicting advice.\n\nDirect answers: It often provides precise answers to specific questions.\n\nHow different AI search engines approach sources\n\nAI search engine\n\n\t\n\nGeneral source preference\n\n\n\n\nChatGPT Search\n\n\t\n\nRelevant authoritative sources, including official documentation\n\n\n\n\nGoogle AI Overviews\n\n\t\n\nHelpful, relevant and authoritative web content\n\n\n\n\nPerplexity\n\n\t\n\nRelevant sources with an emphasis on citations and source quality\n\n\n\n\nMicrosoft Copilot\n\n\t\n\nRelevant web sources, including first-party material\n\nThese are general tendencies, not guaranteed ranking rules.\n\nDoes official documentation always win?\n\nNo. AI search engines may prefer other sources when they provide better answers.\n\nFor example:\n\nSearch query\n\n\t\n\nPotentially preferred source\n\n\n\n\nHow to use an API\n\n\t\n\nOfficial API documentation\n\n\n\n\nLatest product features\n\n\t\n\nOfficial release notes\n\n\n\n\nHow to fix a common error\n\n\t\n\nStack Overflow or GitHub discussions\n\n\n\n\nWhich product is better?\n\n\t\n\nIndependent reviews and comparisons\n\n\n\n\nReal-world user experiences\n\n\t\n\nReddit and community forums\n\n\n\n\nWhat does a company offer?\n\n\t\n\nOfficial website\n\nWhat this means for SEO and GEO\n\nIf you're optimising a website to appear in AI-generated answers (Generative Engine Optimisation), official documentation can be particularly valuable.\n\nI'd recommend:\n\nPublishing comprehensive, first-party documentation.\n\nUsing descriptive headings and clear question-based sections.\n\nProviding accurate examples and practical explanations.\n\nKeeping documentation updated.\n\nIncluding structured data where appropriate.\n\nBuilding corroborating authority through reputable third-party websites.\n\nOne important distinction: AI search engines don't necessarily give official documentation a universal ranking advantage. They aim to retrieve sources that best answer the user's question. Official documentation has a strong advantage for factual, product-specific and technical queries, but not necessarily for opinion-based or experience-based queries.",
      "citations": [],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 0 sources and 0 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-official-documentation-preference-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-official-documentation-preference-repeat-3",
      "published_at": "2026-11-15T13:34:00.000Z",
      "updated_at": "2026-11-15T13:34:00.000Z",
      "revision": 1,
      "slug": "chatgpt-official-documentation-preference-repeat-3",
      "panelId": "official-documentation-preference",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-official-documentation-preference",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "Do AI search engines prefer official documentation?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:34 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b4dd-10ac-83e9-9d22-b03111bafff7?temporary-chat=true",
      "mentions": "Not observed",
      "summary": "Yes. AI search engines generally give significant weight to official documentation, particularly when answering technical, factual, or product-specific questions. However, they don't exclusively prefer official sources.",
      "answerSnapshot": "Yes. AI search engines generally give significant weight to official documentation, particularly when answering technical, factual, or product-specific questions. However, they don't exclusively prefer official sources.\n\nWhy AI search engines favour official documentation\n\nAuthority: Official documentation comes directly from the organisation responsible for a product, service, or technology.\n\nAccuracy: It is more likely to contain correct specifications, supported features, pricing, and implementation details.\n\nFirst-party information: It provides information that third-party websites might only partially understand or reproduce.\n\nTrustworthiness: AI systems can use official sources to verify claims made elsewhere.\n\nFreshness: Well-maintained documentation can provide more current information than older blog posts or tutorials.\n\nHow different AI search engines treat sources\n\nAI search engine\n\n\t\n\nTypical source preferences\n\n\n\n\nChatGPT Search\n\n\t\n\nOfficial websites, authoritative publications, specialist sources\n\n\n\n\nGoogle AI Overviews\n\n\t\n\nRelevant, authoritative, well-indexed web content\n\n\n\n\nPerplexity\n\n\t\n\nPrimary sources, specialist websites, reputable publications\n\n\n\n\nMicrosoft Copilot\n\n\t\n\nOfficial sources, established publishers, relevant web pages\n\n\n\n\nClaude (web search)\n\n\t\n\nRelevant primary sources and credible secondary sources\n\nThese are general tendencies, not confirmed ranking formulas.\n\nDoes official documentation always win?\n\nNo. AI search engines also need to understand the context of a question.\n\nFor example, if someone asks:\n\n\"How do I configure the OpenAI API?\" — Official OpenAI documentation is particularly valuable.\n\n\"What are the best alternatives to OpenAI?\" — Independent comparisons and user experiences are important.\n\n\"Is Product X reliable?\" — Independent reviews, customer experiences, and technical evaluations may be more useful.\n\n\"What changed in Product X's latest release?\" — Official release notes are particularly valuable.\n\nWhat this means for GEO (Generative Engine Optimisation)\n\nIf you're trying to make a website more visible in AI-generated answers, I'd recommend:\n\nPublish authoritative documentation explaining your products, services, and expertise.\n\nCreate clearly structured content with descriptive headings and direct answers.\n\nMaintain factual consistency across your website and other reputable sources.\n\nPublish original research that other websites might reference.\n\nEarn independent mentions from reputable publications, industry websites, and specialist communities.\n\nKeep information updated, particularly specifications, pricing, and product capabilities.\n\nMy key recommendation\n\nDon't focus exclusively on getting AI search engines to read your official documentation.\n\nFocus on creating an ecosystem of information:\n\nOfficial documentation establishes authority.\n\nIndependent publications establish credibility.\n\nExpert articles establish topical expertise.\n\nCommunity discussions establish real-world relevance.\n\nStructured product information establishes clarity.\n\nThe strongest GEO strategy combines first-party authority with third-party validation.\n\nOfficial documentation tells AI engines what you claim. Independent sources help establish whether those claims are credible.",
      "citations": [],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 0 sources and 0 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-google-ai-mode-source-selection-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-google-ai-mode-source-selection-repeat-3",
      "published_at": "2026-11-15T13:33:00.000Z",
      "updated_at": "2026-11-15T13:33:00.000Z",
      "revision": 1,
      "slug": "chatgpt-google-ai-mode-source-selection-repeat-3",
      "panelId": "google-ai-mode-source-selection",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-google-ai-mode-source-selection",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "How does Google AI Mode choose the sources it cites?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:33 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b475-6a80-83e9-81bc-d1a320bd27a5?temporary-chat=true",
      "mentions": "Cited",
      "summary": "Google AI Mode doesn't simply take the first few Google search results and turn them into citations. It uses a combination of Google's traditional search-ranking systems, AI-driven query expansion, content relevance and model-based source selection.",
      "answerSnapshot": "Google AI Mode doesn't simply take the first few Google search results and turn them into citations. It uses a combination of Google's traditional search-ranking systems, AI-driven query expansion, content relevance and model-based source selection.\n\nGoogle hasn't published the complete algorithm or exact weighting of these factors, but its documentation gives us a fairly good picture of how it works.\n\nHow Google AI Mode selects sources\n\n1. Query fan-out\n\nAI Mode breaks your question into several smaller, related searches and runs them simultaneously.\n\nFor example, a question about the best ways to improve website visibility might generate searches about SEO, technical optimisation, content quality and search ranking.\n\n2. Google's search-ranking systems\n\nIt retrieves relevant pages from Google's Search index using existing ranking and quality systems.\n\nThis means traditional SEO remains important. Pages need to be discoverable, relevant and useful.\n\n3. Content evaluation\n\nGoogle's systems examine information from retrieved pages to identify material that helps answer the question.\n\nRelevance, information quality, freshness and the usefulness of specific information all matter.\n\n4. AI synthesis and citation selection\n\nThe model combines information from its retrieved sources to construct an answer and identifies relevant links to support or expand on the information presented.\n\nGoogle for Developers\n+2\nWhat makes one website more likely to be cited?\n\nBased on Google's published guidance, these are important considerations:\n\nFactor\n\n\t\n\nWhy it matters\n\n\n\n\nTopical relevance\n\n\t\n\nThe content directly addresses the question.\n\n\n\n\nContent quality\n\n\t\n\nThe information is useful, reliable and well presented.\n\n\n\n\nOriginal information\n\n\t\n\nFirst-hand insights and unique information provide additional value.\n\n\n\n\nSearch visibility\n\n\t\n\nGoogle needs to discover and retrieve the content.\n\n\n\n\nFreshness\n\n\t\n\nUp-to-date information can be important for changing subjects.\n\n\n\n\nSupporting evidence\n\n\t\n\nSpecific claims should be supported by useful information.\n\n\n\n\nContent accessibility\n\n\t\n\nGoogle needs to be able to crawl and understand the page.\n\nImportantly, Google says its AI search features use its existing Search ranking and quality systems, rather than a completely separate ranking system. \nGoogle for Developers\n+1\n\nDoes the highest-ranking website automatically get cited?\n\nNo. This is one of the most important distinctions.\n\nImagine these search results:\n\nWebsite A ranks first but provides a relatively general overview.\n\nWebsite B ranks fifth but contains original research and detailed statistics.\n\nWebsite C ranks eighth but provides a particularly useful explanation of one specific aspect.\n\nAI Mode might cite B and C because their content supports particular parts of the generated answer.\n\nTraditional organic ranking and AI citation selection are related, but they aren't identical.\n\nDoes Google reveal the exact citation algorithm?\n\nNo. Google has not publicly disclosed:\n\nThe precise weighting of individual ranking signals.\n\nHow it scores individual passages for citation.\n\nHow it chooses between competing sources.\n\nThe exact threshold for including a website as a citation.\n\nWhether a particular citation was selected because of authority, relevance or another signal.\n\nGoogle also explains that AI Mode can sometimes misunderstand web content or miss context. \nGoogle Search Help\n+1\n\nAn important development: Preferred Sources\n\nIn May 2026, Google announced that its Preferred Sources feature was coming to AI Overviews and AI Mode.\n\nThis allows users to identify websites they prefer, making those sources more visible in AI-generated search experiences. \nblog.google\n\nThe key takeaway\n\nGoogle AI Mode is best understood as a combination of:\n\nSearch retrieval – finding potentially relevant webpages.\n\nQuery fan-out – expanding the question into related searches.\n\nContent assessment – identifying useful information.\n\nAI synthesis – generating an answer from retrieved information.\n\nCitation selection – linking to relevant supporting webpages.\n\nThe interesting part is that being ranked highly in Google does not guarantee being cited by Google AI Mode, and ranking lower doesn't automatically exclude a website.\n\nFor a deeper technical explanation, Google's own documentation is particularly useful:\n\nGoogle's AI Search optimisation guide\n\nGoogle's explanation of AI Mode architecture\n\nGoogle AI Mode Help Centre",
      "citations": [
        {
          "title": "Google's Guide to Optimizing for Generative AI Features on Google Search | Google Search Central | Documentation |",
          "host": "Google for Developers",
          "url": "https://developers.google.com/search/docs/fundamentals/ai-optimization-guide?utm_source=chatgpt.com&hl=zh-cn",
          "position": 1
        },
        {
          "title": "Get AI-powered responses with AI Mode in Google Search - Computer",
          "host": "Google Search Help",
          "url": "https://support.google.com/websearch/answer/16011537?co=GENIE.Platform=Desktop&hl=eN&utm_source=chatgpt.com",
          "position": 2
        },
        {
          "title": "AI Mode in Search 五月 20, 2025",
          "host": "Google Search - A new kind of help",
          "url": "https://search.google/pdf/google-about-AI-overviews-AI-Mode.pdf?utm_source=chatgpt.com",
          "position": 3
        },
        {
          "title": "How Google Search helps you find original, quality content Google · 五月 27, 2026",
          "host": "blog.google",
          "url": "https://blog.google/products-and-platforms/products/search/original-high-quality-content-search/?utm_source=chatgpt.com",
          "position": 4
        },
        {
          "title": "Get AI-powered responses with AI Mode in Google Search - Computer",
          "host": "Google Search Help",
          "url": "https://support.google.com/websearch/answer/16011537?co=GENIE.Platform=Desktop&hl=en-GB3Accessibility&utm_source=chatgpt.com",
          "position": 5
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 5 sources and 5 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-official-documentation-preference",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-official-documentation-preference",
      "published_at": "2026-11-15T13:33:00.000Z",
      "updated_at": "2026-11-15T13:33:00.000Z",
      "revision": 1,
      "slug": "chatgpt-official-documentation-preference",
      "panelId": "official-documentation-preference",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-official-documentation-preference",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "Do AI search engines prefer official documentation?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:33 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b4b6-e7b4-83e9-892c-741c17f27366?temporary-chat=true",
      "mentions": "Not observed",
      "summary": "Yes. AI search engines generally tend to favour official documentation for technical queries, but the extent varies by search engine and the type of question.",
      "answerSnapshot": "Yes. AI search engines generally tend to favour official documentation for technical queries, but the extent varies by search engine and the type of question.\n\nWhy AI search engines favour official documentation\n\nAuthority and credibility – Official documentation comes directly from the organisation responsible for the product, API, or technology.\n\nAccuracy – It is more likely to contain authoritative information about features, configuration, supported parameters, and limitations.\n\nFirst-party information – AI systems can use documentation to avoid relying on outdated third-party explanations.\n\nStructured content – Well-organised documentation, with headings, examples, code snippets, and clear URLs, is easier for search and retrieval systems to interpret.\n\nReduced ambiguity – Official sources are often the most reliable reference for resolving conflicting technical information.\n\nHow different AI search engines approach it\n\nAI search engine\n\n\t\n\nGeneral source preference\n\n\n\n\nChatGPT Search\n\n\t\n\nRelevant authoritative sources, including official documentation\n\n\n\n\nGoogle AI Overviews\n\n\t\n\nRelevant, indexed, trustworthy web content\n\n\n\n\nPerplexity\n\n\t\n\nMultiple relevant sources, often including official websites\n\n\n\n\nMicrosoft Copilot\n\n\t\n\nRelevant web sources, including first-party documentation\n\n\n\n\nClaude (web search)\n\n\t\n\nRelevant authoritative sources, including primary sources\n\nDoes official documentation always win?\n\nNo. AI search engines don't necessarily rank official documentation above everything else.\n\nFor example, consider someone searching:\n\n\"How do I configure the Stripe API?\" → Official Stripe documentation is likely highly relevant.\n\n\"Is Stripe better than Adyen?\" → Independent comparisons and user experiences may be more useful.\n\n\"Why does this Stripe API integration fail?\" → Official documentation, GitHub issues, Stack Overflow, and technical discussions may all be relevant.\n\nHow to optimise documentation for AI search engines\n\nIf you're publishing technical documentation and want AI engines to reference it, prioritise:\n\nOptimisation\n\n\t\n\nRecommendation\n\n\n\n\nContent\n\n\t\n\nProvide comprehensive, accurate first-party information\n\n\n\n\nStructure\n\n\t\n\nUse descriptive H1/H2/H3 headings\n\n\n\n\nAnswers\n\n\t\n\nAnswer specific questions directly\n\n\n\n\nExamples\n\n\t\n\nInclude working code examples\n\n\n\n\nFreshness\n\n\t\n\nKeep documentation updated\n\n\n\n\nAccessibility\n\n\t\n\nAllow search crawlers to access important pages\n\n\n\n\nInternal links\n\n\t\n\nConnect related documentation pages\n\n\n\n\nMetadata\n\n\t\n\nUse descriptive page titles and meta descriptions\n\nOne important distinction: Being an official source improves credibility and relevance, but it does not guarantee citation or visibility in AI-generated answers.\n\nThe strongest strategy is to make official documentation the most useful, comprehensive and accessible source for the questions your audience is asking.",
      "citations": [],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 0 sources and 0 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-google-ai-mode-source-selection-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-google-ai-mode-source-selection-repeat-2",
      "published_at": "2026-11-15T13:32:00.000Z",
      "updated_at": "2026-11-15T13:32:00.000Z",
      "revision": 1,
      "slug": "chatgpt-google-ai-mode-source-selection-repeat-2",
      "panelId": "google-ai-mode-source-selection",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-google-ai-mode-source-selection",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "How does Google AI Mode choose the sources it cites?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:32 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b450-213c-83ea-81e7-bbe0f400be14?temporary-chat=true",
      "mentions": "Cited",
      "summary": "Google AI Mode chooses its citations by combining Google Search results, multiple related searches, and AI-based selection of pages that support its answer.",
      "answerSnapshot": "Google AI Mode chooses its citations by combining Google Search results, multiple related searches, and AI-based selection of pages that support its answer.\n\nThe important distinction is that Google has explained the overall process, but hasn't publicly disclosed the exact formula it uses to decide which individual pages get cited.\n\nHere's how it works.\n\n1. It breaks your question into smaller searches\n\nGoogle calls this query fan-out.\n\nRather than searching only for the exact words you typed, AI Mode divides your question into related subtopics and searches for them simultaneously.\n\nFor example, if you ask:\n\n\"What is the best electric car for a family?\"\n\nIt might explore several related areas:\n\nBest electric cars for families.\n\nElectric car safety ratings.\n\nBattery range and charging.\n\nInterior space and practicality.\n\nReliability and ownership costs.\n\nIt then combines the information it finds into one response. \nGoogle Search Help\n+1\n\n2. It looks for supporting webpages\n\nAI Mode uses Google's Search infrastructure and information sources, including:\n\nGoogle's web index.\n\nRelevant websites and articles.\n\nGoogle's Knowledge Graph.\n\nReal-time information and other specialised data sources.\n\nIt identifies webpages that can support different parts of its response.\n\nThis means a page doesn't necessarily need to rank first for your original search to potentially be cited. It might be relevant to one of the smaller searches instead. \nblog.google\n+1\n\n3. It selects sources to support the answer\n\nThe AI combines the retrieved information and identifies relevant supporting pages.\n\nIn practice, useful characteristics of a source include:\n\nRelevance: Does it answer the particular question?\n\nAccuracy: Does it provide reliable information?\n\nSpecificity: Does it contain evidence for the particular claim?\n\nCoverage: Does it address an important part of the question?\n\nCredibility: Is the information trustworthy?\n\nThese are sensible ways to evaluate sources, but Google hasn't published a precise AI Mode citation scoring formula or confirmed that these are separate weighted selection factors.\n\n4. It produces citations alongside its response\n\nThe final response includes links to webpages that help support the information presented.\n\nAn interesting detail: Google says AI Mode can show a wider and more diverse set of links than traditional Search.\n\nAlso, AI Mode and AI Overviews can produce different citations because they may use different models and techniques. \nGoogle Search Help\n+1\n\n5. What determines whether a website can be cited?\n\nGoogle's published guidance for its AI search features says that a page must:\n\nBe indexed by Google.\n\nBe eligible to appear in Google Search with a snippet.\n\nComply with Google's normal Search technical requirements.\n\nGoogle says there are no additional technical requirements specifically for appearing as a supporting link in its AI search features.\n\nGoogle Search Help\n+1\nWhat Google hasn't revealed\n\nThere are some important unanswered questions:\n\nQuestion\n\n\t\n\nWhat we know\n\n\n\n\nDoes ranking position matter?\n\n\t\n\nGoogle uses its existing Search systems, but hasn't disclosed the exact citation weighting.\n\n\n\n\nDoes domain authority determine citations?\n\n\t\n\nNo published AI Mode formula confirms this.\n\n\n\n\nDoes AI Mode prefer recent articles?\n\n\t\n\nNo universal recency rule has been disclosed.\n\n\n\n\nDoes it favour certain websites?\n\n\t\n\nNo official preference list has been published.\n\n\n\n\nHow many pages does it evaluate?\n\n\t\n\nThe exact number isn't publicly specified.\n\n\n\n\nWhy does one page get cited over another?\n\n\t\n\nThe precise selection mechanism remains undisclosed.\n\nThe key takeaway\nHow Google AI Mode selects citations\n\n1. Understand the question\n\nIdentify the different topics and information needed.\n\n2. Query fan-out\n\nRun multiple related searches simultaneously.\n\n3. Retrieve relevant sources\n\nFind supporting information across Google's data sources.\n\n4. Generate the answer\n\nCombine information and identify supporting webpages.\n\n5. Display citations\n\nShow links alongside the generated response.\n\nMy main conclusion: Getting cited by Google AI Mode isn't simply about ranking number one on Google. It's about having relevant, useful, accessible content that can support the different aspects of a user's question.\n\nGoogle's own explanation is available here: How AI Mode works – Google Search Help.",
      "citations": [
        {
          "title": "Get AI-powered responses with AI Mode in Google Search - Computer",
          "host": "Google Search Help",
          "url": "https://support.google.com/websearch/answer/16011537?co=GENIE.Platform=Desktop&hl=en-GB3Accessibility&utm_source=chatgpt.com",
          "position": 1
        },
        {
          "title": "Expanding AI Overviews and introducing AI Mode Google · 三月 5, 2025",
          "host": "blog.google",
          "url": "https://blog.google/products-and-platforms/products/search/ai-mode-search/?utm_source=chatgpt.com",
          "position": 2
        },
        {
          "title": "May 2025 AI Overviews and AI Mode in Search 五月 20, 2025",
          "host": "Google Search - A new kind of help",
          "url": "https://search.google/pdf/google-about-AI-overviews-AI-Mode.pdf?_hsmi=351072060&utm_source=chatgpt.com",
          "position": 3
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 3 sources and 3 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-chatgpt-search-source-selection-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-chatgpt-search-source-selection-repeat-2",
      "published_at": "2026-11-15T13:31:00.000Z",
      "updated_at": "2026-11-15T13:31:00.000Z",
      "revision": 1,
      "slug": "chatgpt-chatgpt-search-source-selection-repeat-2",
      "panelId": "chatgpt-search-source-selection",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-chatgpt-search-source-selection",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "How does ChatGPT Search choose the sources it cites?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:31 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b3fb-d398-83e9-ab3d-232056904cf5?temporary-chat=true",
      "mentions": "Cited",
      "summary": "ChatGPT Search selects sources based on relevance, reliability and usefulness to the question. It doesn't simply copy the first few results from a traditional search engine.",
      "answerSnapshot": "ChatGPT Search selects sources based on relevance, reliability and usefulness to the question. It doesn't simply copy the first few results from a traditional search engine.\n\nOpenAI confirms that its search system uses multiple factors to identify relevant and reliable sources, although it hasn't publicly disclosed the exact ranking formula or the weight given to each factor. \nOpenAI Help Center\n+1\n\nHere's how the process works.\n\n1. Understanding the question\n\nChatGPT first interprets what you're asking and identifies the information needed.\n\nIt may rewrite your question into several more targeted search queries.\n\nFor example, if you ask:\n\n\"What are the benefits of solar energy in the UK?\"\n\nIt might search for:\n\nSolar energy benefits.\n\nUK solar energy statistics.\n\nGovernment research on renewable energy.\n\nRecent studies on solar energy costs.\n\nOpenAI Help Center\n2. Finding relevant sources\n\nChatGPT can use search providers and information supplied directly by publishing partners to find relevant webpages.\n\nIt looks for material that helps answer the specific question.\n\nPotentially useful sources include:\n\nGovernment websites.\n\nAcademic research and scientific publications.\n\nEstablished news organisations.\n\nSpecialist industry publications.\n\nRelevant company websites and technical documentation.\n\nNo particular category is automatically guaranteed to appear.\n\n3. Evaluating the information\n\nSearch results are ranked using multiple factors intended to identify relevant and reliable information.\n\nSome important practical considerations when assessing sources are:\n\nFactor\n\n\t\n\nWhat it means\n\n\n\n\nRelevance\n\n\t\n\nDoes the page directly answer the question?\n\n\n\n\nReliability\n\n\t\n\nIs the information supported by credible evidence?\n\n\n\n\nFreshness\n\n\t\n\nIs the information current enough?\n\n\n\n\nSpecificity\n\n\t\n\nDoes the source provide useful detail?\n\n\n\n\nAccessibility\n\n\t\n\nCan the search system retrieve the content?\n\nOpenAI hasn't confirmed that these are individual ranking signals with specific weights. They are useful ways to understand what makes a source valuable, rather than a published ranking formula.\n\n4. Attaching citations\n\nOnce ChatGPT has gathered information, it generates an answer and attaches citations to relevant sources.\n\nIdeally:\n\nEach citation supports the statement beside it.\n\nDifferent sources support different parts of the answer.\n\nReaders can open the original source and verify the information.\n\nA citation doesn't necessarily mean that OpenAI endorses the publisher or that every claim on that website is correct.\n\n5. Can websites influence whether ChatGPT cites them?\n\nYes, to some extent.\n\nOpenAI provides a crawler called OAI-SearchBot, which helps websites become discoverable in ChatGPT Search.\n\nWebsite owners can:\n\nAllow OAI-SearchBot to access their content.\n\nPublish clear, useful and relevant information.\n\nKeep time-sensitive information updated.\n\nMake important information accessible to search systems.\n\nHowever, allowing the crawler doesn't guarantee that a website will be cited. \nOpenAI Help Center\n\nWhat is the most important thing to understand?\n\nChatGPT Search is not a simple popularity contest.\n\nA website isn't guaranteed to be cited because it:\n\nHas the most visitors.\n\nHas the highest Google ranking.\n\nBelongs to the biggest company.\n\nPublishes the most content.\n\nWhat matters is whether the retrieved information is useful for answering the particular question.\n\nOne important limitation: OpenAI acknowledges that search results and citations can sometimes be incomplete, outdated or incorrect. It's worth opening the original sources when accuracy matters. \nOpenAI Help Center\n+1\n\nOfficial reference: OpenAI – Searching the web with ChatGPT.",
      "citations": [
        {
          "title": "Searching the web with ChatGPT",
          "host": "OpenAI Help Center",
          "url": "https://help.openai.com/en/articles/9237897-searching-the-web-with-chatgpt?utm_source=chatgpt.com",
          "position": 1
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 1 sources and 1 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-chatgpt-search-source-selection-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-chatgpt-search-source-selection-repeat-3",
      "published_at": "2026-11-15T13:31:00.000Z",
      "updated_at": "2026-11-15T13:31:00.000Z",
      "revision": 1,
      "slug": "chatgpt-chatgpt-search-source-selection-repeat-3",
      "panelId": "chatgpt-search-source-selection",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-chatgpt-search-source-selection",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "How does ChatGPT Search choose the sources it cites?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:31 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b41d-90a8-83e9-a56f-192938a4c8fd?temporary-chat=true",
      "mentions": "Cited",
      "summary": "ChatGPT Search selects sources based on relevance, usefulness and reliability. It doesn't simply copy the first few results from Google or automatically cite the most popular websites.",
      "answerSnapshot": "ChatGPT Search selects sources based on relevance, usefulness and reliability. It doesn't simply copy the first few results from Google or automatically cite the most popular websites.\n\nOpenAI hasn't published the complete source-ranking algorithm, so the exact weighting of different factors isn't publicly known.\n\nHere's how the process works.\n\n1. It interprets your question\n\nChatGPT first analyses what you're asking and identifies the information needed.\n\nIt may rewrite your question into several more targeted search queries.\n\nFor example, if you ask:\n\n\"What are the latest developments in renewable energy?\"\n\nIt might search separately for:\n\nRecent renewable energy research.\n\nGovernment energy policies.\n\nNew solar and wind technologies.\n\nIndustry developments.\n\nThis helps it find more relevant information than a single broad search. \nOpenAI Help Center\n+1\n\n2. It retrieves potential sources\n\nChatGPT Search can use:\n\nThird-party search providers.\n\nWebsites and articles available on the public web.\n\nContent supplied directly by publishing partners.\n\nIt gathers potentially useful pages and information before constructing an answer. \nOpenAI\n+1\n\n3. It evaluates which sources are useful\n\nOpenAI says its search system ranks results using multiple factors intended to identify relevant and reliable information.\n\nSome practical qualities that make a source useful include:\n\nFactor\n\n\t\n\nWhat it means\n\n\n\n\nRelevance\n\n\t\n\nDoes the source directly answer the question?\n\n\n\n\nReliability\n\n\t\n\nIs the publisher trustworthy?\n\n\n\n\nEvidence\n\n\t\n\nDoes it provide supporting facts or research?\n\n\n\n\nFreshness\n\n\t\n\nIs the information current?\n\n\n\n\nSpecificity\n\n\t\n\nDoes it address the particular subject?\n\nThese are useful ways to understand source quality, not a published list of individual ranking signals or their weights. \nOpenAI Help Center\n+1\n\n4. It decides which sources to cite\n\nThis is an important distinction.\n\nA source being found doesn't automatically mean it will be cited.\n\nChatGPT uses information from retrieved sources to construct its response, then provides citations linking relevant claims to supporting webpages.\n\nFor example:\n\nA scientific paper might support a scientific claim.\n\nAn official government website might support a policy or legal claim.\n\nA reputable news organisation might support a recent event.\n\nA company's website might support information about its own products.\n\nThe aim is to make it possible to check where information came from.\n\n5. Can websites influence whether ChatGPT cites them?\n\nTo an extent.\n\nOpenAI provides a search crawler called OAI-SearchBot. Website owners can allow it to crawl their websites, making their content eligible to appear in ChatGPT Search.\n\nHowever, being eligible doesn't guarantee being cited or appearing prominently. \nOpenAI Help Center\n\nWhat ChatGPT doesn't publicly disclose\n\nThere are several important unanswered questions:\n\nThe exact mathematical ranking formula.\n\nHow much authority, relevance and freshness contribute individually.\n\nWhether one search provider has more influence than another.\n\nThe precise process for deciding which retrieved pages receive citations.\n\nHow to check the sources yourself\nsemrush.com\n\nWhen ChatGPT gives you a search-based answer:\n\nSelect an inline citation to open its source.\n\nSelect Sources to see cited sources and other relevant links.\n\nCheck the publication date and whether the source actually supports the statement.\n\nCompare important claims against independent sources.\n\nOpenAI also acknowledges that search results and citations can sometimes be incomplete, outdated or incorrect. \nOpenAI Help Center\n+1\n\nThe key takeaway: ChatGPT Search aims to cite sources that are relevant and reliable, rather than simply the most popular websites. But its complete ranking algorithm remains proprietary, and a citation should never be treated as an automatic guarantee of accuracy.",
      "citations": [
        {
          "title": "Searching the web with ChatGPT",
          "host": "OpenAI Help Center",
          "url": "https://help-lb.openai.com/en/articles/9237897-searching-the-web-with-chatgpt?utm_source=chatgpt.com",
          "position": 1
        },
        {
          "title": "Searching the web with ChatGPT",
          "host": "OpenAI Help Center",
          "url": "https://help.openai.com/en/articles/9237897-searching-the-web-with-chatgpt?utm_source=chatgpt.com",
          "position": 2
        },
        {
          "title": "Introducing ChatGPT search  · 十月 31, 2024",
          "host": "OpenAI",
          "url": "https://openai.com/zh-Hans-CN/index/introducing-chatgpt-search/?utm_source=chatgpt.com",
          "position": 3
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 3 sources and 3 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-google-ai-mode-source-selection",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-google-ai-mode-source-selection",
      "published_at": "2026-11-15T13:30:00.000Z",
      "updated_at": "2026-11-15T13:30:00.000Z",
      "revision": 1,
      "slug": "chatgpt-google-ai-mode-source-selection",
      "panelId": "google-ai-mode-source-selection",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-google-ai-mode-source-selection",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "How does Google AI Mode choose the sources it cites?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:30 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b3a1-8b74-83ea-8c61-814b71a55ff2?temporary-chat=true",
      "mentions": "Cited",
      "summary": "Google AI Mode chooses cited sources through a combination of Google Search ranking, query expansion, relevance assessment, and AI-generated synthesis.",
      "answerSnapshot": "Google AI Mode chooses cited sources through a combination of Google Search ranking, query expansion, relevance assessment, and AI-generated synthesis.\n\nThe important distinction is that it doesn't simply find one webpage, read it, and cite it. It can search across multiple related queries, gather information from different websites, and use those pages to construct a single answer.\n\nHere's how the process works.\n\n1. Query fan-out: it searches beyond your exact question\n\nThis is one of the most important mechanisms.\n\nGoogle's AI Mode uses a technique called query fan-out.\n\nFor example, if you ask:\n\n“What are the best ways for small businesses to improve customer retention?”\n\nIt might break that into related searches covering:\n\nCustomer retention strategies.\n\nCustomer loyalty programmes.\n\nCustomer experience research.\n\nCustomer churn statistics.\n\nExamples of successful retention campaigns.\n\nThese searches happen concurrently, allowing AI Mode to find relevant information that a single search might miss.\n\nGoogle explicitly documents this process. \nGoogle Search Help\n+1\n\n2. Google Search ranking: it finds candidate sources\n\nAI Mode builds on Google's existing Search infrastructure.\n\nIts source discovery process draws on:\n\nGoogle's search index.\n\nExisting relevance and quality-ranking systems.\n\nInformation freshness.\n\nRelevant information from Google's Knowledge Graph and other data sources.\n\nGoogle describes this as retrieval-augmented generation (RAG), also known as grounding.\n\nIn simple terms:\n\nGoogle retrieves potentially relevant webpages.\n\nThe AI examines information from those pages.\n\nIt uses relevant information to construct an answer.\n\nIt provides links to supporting webpages.\n\nGoogle for Developers\n+1\n3. Relevance to individual statements\n\nNot every source is necessarily used to support the entire answer.\n\nFor example, an answer about electric vehicles might contain:\n\nAnswer section\n\n\t\n\nRelevant source type\n\n\n\n\nBattery technology\n\n\t\n\nTechnical research\n\n\n\n\nVehicle prices\n\n\t\n\nManufacturer or pricing pages\n\n\n\n\nEnvironmental impact\n\n\t\n\nScientific studies\n\n\n\n\nGovernment incentives\n\n\t\n\nOfficial government websites\n\nThe aim is to connect relevant sources to the information they help support.\n\nGoogle doesn't publicly disclose the precise algorithm or weighting used to decide which retrieved pages receive citations.\n\n4. Quality and usefulness matter\n\nGoogle says its AI Search experiences rely on high-quality web content.\n\nFactors worth understanding include:\n\nRelevance: Does the page answer the question?\n\nInformation quality: Is the information useful and reliable?\n\nOriginality: Does it offer firsthand information or unique insights?\n\nFreshness: Is the information current where that matters?\n\nSpecificity: Does it address the particular subtopic?\n\nThese are useful ways to understand Google's approach, rather than a published formula for citation selection.\n\nGoogle for Developers\n+2\n5. Preferred Sources can influence what users see\n\nGoogle introduced Preferred Sources into AI Mode in 2026.\n\nUsers can select websites they particularly value, and Google can highlight those websites when they appear in AI responses.\n\nThis is worth distinguishing from ordinary source selection: a preferred source isn't automatically guaranteed a citation.\n\nblog.google\n+1\nThe complete process\n\n1. User question\n\nAI interprets the intent and context\n\n2. Query fan-out\n\nGenerates multiple related searches\n\n3. Search and retrieval\n\nFinds relevant webpages and other information\n\n4. AI synthesis and grounding\n\nCombines retrieved information into an answer\n\n5. Answer and citations\n\nDisplays the response with links to relevant sources\n\nWhat does this mean for website owners?\n\nThe key takeaway is that ranking in traditional Google Search and being cited in AI Mode are related, but not identical.\n\nA website can rank well organically without necessarily being cited in a particular AI answer.\n\nEqually, a page that provides particularly useful information for one of the subtopics generated by query fan-out may be relevant to an AI response.\n\nGoogle's official guidance is that traditional SEO fundamentals remain important for AI Search. \nGoogle for Developers\n\nOne important limitation: Google has not published a complete, deterministic explanation of how it selects one specific URL over another for an individual AI Mode citation.\n\nFor further reading: Google's official AI Search optimisation guide.",
      "citations": [
        {
          "title": "Get AI-powered responses with AI Mode in Google Search - Android phone & tablet",
          "host": "Google Search Help",
          "url": "https://support.google.com/websearch/answer/16011537?co=GENIE.Platform=Android&hl=en&utm_source=chatgpt.com",
          "position": 1
        },
        {
          "title": "Google's Guide to Optimizing for Generative AI Features on Google Search | Google Search Central | Documentation |",
          "host": "Google for Developers",
          "url": "https://developers.google.com/search/docs/fundamentals/ai-optimization-guide?error=error_code&utm_source=chatgpt.com&hl=zh-cn",
          "position": 2
        },
        {
          "title": "AI Mode in Search 五月 20, 2025",
          "host": "Google Search - A new kind of help",
          "url": "https://search.google/pdf/google-about-AI-overviews-AI-Mode.pdf?stream=top&utm_source=chatgpt.com",
          "position": 3
        },
        {
          "title": "How Google Search helps you find original, quality content Google · 五月 27, 2026",
          "host": "blog.google",
          "url": "https://blog.google/products-and-platforms/products/search/original-high-quality-content-search/?utm_source=chatgpt.com",
          "position": 4
        },
        {
          "title": "How AI Mode and AI Overviews help you explore the web Google · 五月 6, 2026",
          "host": "blog.google",
          "url": "https://blog.google/products-and-platforms/products/search/explore-web-generative-ai-search/?trk=article-ssr-frontend-pulse_little-text-block&utm_source=chatgpt.com",
          "position": 5
        },
        {
          "title": "Google AI Mode & Overviews will highlight your Preferred Sources  · 五月 27, 2026",
          "host": "9to5Google",
          "url": "https://9to5google.com/2026/05/27/google-ai-mode-preferred-sources/?extended-comments=1&utm_source=chatgpt.com",
          "position": 6
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 6 sources and 6 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-chatgpt-search-source-selection",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-chatgpt-search-source-selection",
      "published_at": "2026-11-15T13:28:00.000Z",
      "updated_at": "2026-11-15T13:28:00.000Z",
      "revision": 1,
      "slug": "chatgpt-chatgpt-search-source-selection",
      "panelId": "chatgpt-search-source-selection",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-chatgpt-search-source-selection",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "How does ChatGPT Search choose the sources it cites?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:28 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b344-1ab0-83ea-84e4-c77b2571aa77?temporary-chat=true",
      "mentions": "Cited",
      "summary": "ChatGPT Search selects sources based on relevance, usefulness and reliability. It doesn't simply copy the top results from Google or cite the most popular websites.",
      "answerSnapshot": "ChatGPT Search selects sources based on relevance, usefulness and reliability. It doesn't simply copy the top results from Google or cite the most popular websites.\n\nThe exact ranking formula isn't publicly disclosed by OpenAI, but we can explain the general process.\n\nHow ChatGPT Search selects sources\n\n1. It interprets your question\n\nChatGPT analyses what you're asking and may rewrite your question into several more targeted search queries.\n\n2. It retrieves relevant web pages\n\nIt uses search providers to find potentially useful sources. These might include news organisations, research papers, official websites, specialist publications and other websites.\n\n3. It evaluates the results\n\nSearch ranking uses multiple factors intended to identify relevant and reliable information. ChatGPT may also perform additional searches to investigate specific aspects of your question.\n\n4. It uses sources to construct an answer\n\nChatGPT draws on the retrieved information and attaches citations to sources that support relevant statements.\n\n5. It displays citations\n\nYou can click citations within the answer or select the Sources section to see cited sources and other relevant links.\n\nSource: OpenAI – Searching the web with ChatGPT.\n\nWhat makes a source more likely to be cited?\n\nAlthough OpenAI doesn't publish the precise weighting of its ranking factors, these are useful qualities to consider:\n\nFactor\n\n\t\n\nWhy it matters\n\n\n\n\nRelevance\n\n\t\n\nDirectly addresses the question\n\n\n\n\nReliability\n\n\t\n\nProvides trustworthy information\n\n\n\n\nAccuracy\n\n\t\n\nSupports claims with evidence\n\n\n\n\nFreshness\n\n\t\n\nContains up-to-date information when needed\n\n\n\n\nSpecificity\n\n\t\n\nProvides detailed answers rather than vague statements\n\n\n\n\nAccessibility\n\n\t\n\nCan be retrieved and understood by Search\n\nThese are sensible qualities for evaluating sources, not a published list of individually weighted ranking signals.\n\nDoes ChatGPT favour certain websites?\n\nNot necessarily.\n\nFor example, depending on the question, ChatGPT might cite:\n\nGovernment websites for official statistics.\n\nUniversities for academic research.\n\nScientific journals for scientific findings.\n\nNews organisations for current events.\n\nSpecialist websites for technical explanations.\n\nA smaller specialist website can be more useful than a major news organisation if it contains the most relevant evidence.\n\nCan a website guarantee that ChatGPT will cite it?\n\nNo.\n\nOpenAI doesn't offer a guaranteed way to obtain citations or a published formula that allows website owners to control citation placement.\n\nAlso, an important distinction:\n\nBeing discovered: A website appears among the search results.\n\nBeing cited: ChatGPT actually uses that website as a source in its answer.\n\nThese aren't the same thing.\n\nHow can you check the sources yourself?\nsemrush.com\n\nClick a citation in a ChatGPT answer.\n\nOpen the original webpage.\n\nCheck whether it actually supports the statement.\n\nCheck its publication date and credibility.\n\nCompare it with other sources.\n\nOpenAI also acknowledges that citations can sometimes be incomplete, outdated or incorrect. \nOpenAI Help Center\n+1\n\nThe key takeaway: ChatGPT Search aims to cite sources that help answer your question accurately and relevantly. However, its exact source-selection algorithm remains proprietary.",
      "citations": [
        {
          "title": "Searching the web with ChatGPT",
          "host": "OpenAI Help Center",
          "url": "https://help-lb.openai.com/en/articles/9237897-searching-the-web-with-chatgpt?utm_source=chatgpt.com",
          "position": 1
        },
        {
          "title": "ChatGPT search for Enterprise and Edu",
          "host": "OpenAI Help Center",
          "url": "https://help.openai.com/en/articles/10093903-chatgpt-search-for-enterprise-and-edu?utm_source=chatgpt.com",
          "position": 2
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 2 sources and 2 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-get-cited-perplexity-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-get-cited-perplexity-repeat-3",
      "published_at": "2026-11-15T13:25:00.000Z",
      "updated_at": "2026-11-15T13:25:00.000Z",
      "revision": 1,
      "slug": "chatgpt-get-cited-perplexity-repeat-3",
      "panelId": "get-cited-perplexity",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-get-cited-perplexity",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "How can a new website get cited by Perplexity?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:25 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b2ae-8118-83ea-b067-7dfc0ad72c91?temporary-chat=true",
      "mentions": "Cited",
      "summary": "Content authority: Publishing useful, trustworthy information that Perplexity has a reason to reference in its answers.",
      "answerSnapshot": "Getting a new website cited by Perplexity involves two things:\n\nTechnical visibility: Making sure Perplexity can discover, crawl and access your website.\n\nContent authority: Publishing useful, trustworthy information that Perplexity has a reason to reference in its answers.\n\nThis is often called Generative Engine Optimisation (GEO) or Answer Engine Optimisation (AEO).\n\nThe important distinction is that getting indexed doesn't automatically mean getting cited. Perplexity doesn't publish a definitive formula for selecting citations.\n\n1. Make your website accessible to Perplexity\n\nThis is the first thing to check.\n\nPerplexity operates a crawler called PerplexityBot, designed to discover websites and surface them in search results.\n\nCheck your website's:\n\nhttps://yourdomain.com/robots.txt\n\nMake sure you aren't blocking PerplexityBot.\n\nFor example, you can explicitly allow it:\n\n纯文本\nUser-agent: PerplexityBot\nAllow: /\n\nAlso check:\n\nYour hosting provider isn't blocking the crawler.\n\nYour CDN or firewall isn't rejecting legitimate Perplexity requests.\n\nYour important pages return HTTP 200 responses.\n\nYour sitemap is accessible.\n\nYour pages aren't accidentally marked noindex.\n\nPerplexity publishes its crawler information and IP ranges in its developer documentation. \nGreat Ads\n+1\n\n2. Publish content that answers specific questions\n\nThis is probably the most important ongoing activity.\n\nPerplexity is designed to answer questions using relevant web sources. Your website should therefore contain pages that directly answer the questions your potential audience asks.\n\nFor example, instead of publishing:\n\nOur company\n\nOur services\n\nWhy choose us\n\nCreate more specific, informative articles:\n\nWhat is [your service]?\n\nHow much does [your service] cost?\n\nHow does [your service] work?\n\nWhat are the benefits of [your service]?\n\n[Your service] vs [alternative]: What's the difference?\n\nHow do you choose the right [product/service]?\n\nA good structure for an article:\n\nClear question as the H1 heading.\n\nA direct answer in the opening paragraph.\n\nA detailed explanation.\n\nSpecific facts, statistics and examples.\n\nRelevant expert commentary.\n\nLinks to reliable supporting sources.\n\nA short FAQ section.\n\nAvoid vague marketing language. Give Perplexity factual information it can independently reference.\n\n3. Build credibility and authority\n\nA brand-new website has a disadvantage: it hasn't yet established much online authority.\n\nYou can work on this by:\n\nGetting mentioned in reputable industry publications.\n\nPublishing original research and useful data.\n\nObtaining genuine expert contributions.\n\nBuilding relationships with relevant industry organisations.\n\nGetting legitimate backlinks from relevant websites.\n\nMaintaining consistent company information across the web.\n\nAn important point: brand mentions and third-party references matter too.\n\nDon't concentrate exclusively on publishing articles on your own domain. Build a reputation across your industry.\n\n4. Make your content easy for AI systems to understand\n\nUse a technically clean website.\n\nElement\n\n\t\n\nRecommendation\n\n\n\n\nPage titles\n\n\t\n\nDescriptive and specific\n\n\n\n\nH1 headings\n\n\t\n\nOne clear primary topic\n\n\n\n\nH2 headings\n\n\t\n\nOrganise related questions\n\n\n\n\nSchema markup\n\n\t\n\nRelevant Article, Organisation, Product or FAQPage schema where appropriate\n\n\n\n\nSitemap\n\n\t\n\nXML sitemap\n\n\n\n\nInternal links\n\n\t\n\nConnect related articles\n\n\n\n\nImages\n\n\t\n\nInclude descriptive alt text\n\n\n\n\nContent\n\n\t\n\nMake important information available as readable HTML\n\n\n\n\nDates\n\n\t\n\nDisplay publication and update dates accurately\n\nSchema markup helps machines understand your content, although it doesn't guarantee citations.\n\n5. Get your website discovered elsewhere\n\nDon't rely on Perplexity discovering your website entirely on its own.\n\nEstablish a wider online presence:\n\nGet your website indexed by Google and Bing.\n\nPublish useful articles on relevant industry websites.\n\nShare original research and findings.\n\nParticipate in relevant professional communities.\n\nMake sure your business name, expertise and website are consistently represented online.\n\nThe goal is to establish your website as a credible source of information within its subject area.\n\n6. Test whether Perplexity is citing you\n\nYou can do this manually.\n\nCreate a list of 10–20 questions that your ideal customers might ask Perplexity.\n\nFor example:\n\nWhat are the best companies offering [service]?\n\nHow much does [service] cost in 2026?\n\nWhich providers specialise in [specific requirement]?\n\nWhat should I look for when choosing [service]?\n\nRun these searches periodically.\n\nRecord:\n\nWhether your website appears.\n\nWhich individual pages are cited.\n\nWhich competitors are cited.\n\nWhich third-party websites appear repeatedly.\n\nWhat type of content Perplexity prefers for each question.\n\nThis gives you a practical way to identify content gaps.\n\n7. A practical 30-day action plan\n\nWeek 1 – Technical setup\n\nCheck robots.txt.\n\nAllow PerplexityBot.\n\nCheck firewall and CDN settings.\n\nSubmit your XML sitemap to Google and Bing.\n\nCheck indexing and crawl errors.\n\nWeek 2 – Content development\n\nIdentify 20 relevant customer questions.\n\nPublish 5 high-quality, informative articles.\n\nAdd clear headings and direct answers.\n\nInclude credible supporting references.\n\nWeek 3 – Authority building\n\nPublish original research.\n\nSeek relevant industry coverage.\n\nBuild genuine industry relationships.\n\nStrengthen internal linking.\n\nWeek 4 – Measurement\n\nTest your 20 target questions in Perplexity.\n\nRecord citations.\n\nIdentify competitors appearing frequently.\n\nImprove pages that aren't attracting visibility.\n\nOfficial documentation\n\nThese are useful starting points:\n\nPerplexity – Official crawler documentation.\n\nPerplexityBot official IP ranges.\n\nGoogle Search Central documentation.\n\nMy biggest recommendation\n\nDon't approach this as simply trying to get Perplexity to index a new website.\n\nApproach it as building a website that becomes a genuinely useful reference in its industry.\n\nThe strongest strategy combines:\n\nTechnical accessibility + excellent informational content + independent credibility + consistent measurement.\n\nThere is no guaranteed way to obtain a citation, but these steps improve your site's eligibility and give it a much stronger foundation.\n\nOne question that would help me tailor the strategy\n\nAre you trying to get citations for:\n\nA new business website?\n\nA SaaS or technology website?\n\nAn ecommerce website?\n\nA blog or content publication?\n\nA digital marketing or SEO agency website?\n\nThe strategy differs quite significantly depending on the type of website.",
      "citations": [
        {
          "title": "Getting Cited in Perplexity: robots.txt & IP Access",
          "host": "Great Ads",
          "url": "https://www.greatads.com.tr/en/guides/ai-search-and-geo/perplexity-source-citation?utm_source=chatgpt.com",
          "position": 1
        },
        {
          "title": "How to get cited by Perplexity",
          "host": "Ignura",
          "url": "https://ignura.com/blog/how-to-get-cited-by-perplexity?utm_source=chatgpt.com",
          "position": 2
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 2 sources and 2 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-get-cited-perplexity",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-get-cited-perplexity",
      "published_at": "2026-11-15T13:24:00.000Z",
      "updated_at": "2026-11-15T13:24:00.000Z",
      "revision": 1,
      "slug": "chatgpt-get-cited-perplexity",
      "panelId": "get-cited-perplexity",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-get-cited-perplexity",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "How can a new website get cited by Perplexity?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:24 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b26c-4758-83ea-92d8-2c1d5f74875a?temporary-chat=true",
      "mentions": "Cited",
      "summary": "Getting a new website cited by Perplexity involves a combination of technical SEO, high-quality content, topical authority and visibility across the web.",
      "answerSnapshot": "Getting a new website cited by Perplexity involves a combination of technical SEO, high-quality content, topical authority and visibility across the web.\n\nThe important distinction is that getting your website crawled is not the same as getting it cited. Perplexity needs to be able to access your pages, find them relevant to a user's question, and select them as sources.\n\nThere is no guaranteed submission process or publicly documented formula for earning citations.\n\nHere's a practical strategy for a new website.\n\n1. Make your website accessible to Perplexity\nEssential first step\n\nPerplexity operates a crawler called PerplexityBot, designed to discover and surface websites in its search results.\n\nCheck the following:\n\nYour website is publicly accessible.\n\nImportant pages return HTTP 200 responses.\n\nYour robots.txt file doesn't block PerplexityBot.\n\nYour hosting provider, CDN or firewall doesn't block its crawler.\n\nYour pages can be read without requiring a login.\n\nYour sitemap.xml is available and submitted to search engines.\n\nA basic robots.txt configuration could look like:\n\n纯文本\nUser-agent: PerplexityBot\nAllow: /\n\nSitemap: https://example.com/sitemap.xml\n\nOnly use this if you want to allow Perplexity to crawl your public content.\n\nPerplexity also publishes its crawler IP ranges, which can help with firewall configuration.\n\nThe official documentation confirms that Perplexity recommends allowing its crawler in robots.txt and permitting its published IP addresses. Changes can take up to 24 hours to appear in its systems. \nPerplexity\n\n2. Publish content that directly answers questions\n\nThis is probably the most important content strategy.\n\nPerplexity answers questions, so create pages that answer the questions your potential audience actually asks.\n\nFor example, instead of publishing:\n\nThe ultimate guide to our services\n\nWhy our company is different\n\nWelcome to our website\n\nPublish content such as:\n\nWhat is [your product category]?\n\nHow does [your service] work?\n\nHow much does [your service] cost?\n\nWhat are the benefits of [your solution]?\n\n[Product A] vs [Product B]: What's the difference?\n\nWhat should you consider before choosing [your service]?\n\nStructure each article like this:\n\nDirect answer: Explain the answer in the opening paragraph.\n\nDetailed explanation: Expand on the subject.\n\nFacts and figures: Include relevant statistics and evidence.\n\nExamples: Demonstrate how the information applies.\n\nSources: Link to authoritative references.\n\nConclusion: Summarise the key findings.\n\nMake each page useful as a standalone reference.\n\n3. Build topical authority\n\nA new website has little established reputation, so it needs to demonstrate expertise through consistently useful content.\n\nFor example, if your website specialises in digital marketing, create a collection of interconnected resources:\n\nExample: Digital marketing topical authority\n\nDigital Marketing\n\nCentral topic / pillar page\n\nSEO\n\nTechnical SEO, keyword research, on-page SEO\n\nContent Marketing\n\nContent strategy, blogging, content planning\n\nAI Search\n\nPerplexity, AI visibility, search citations\n\nAnalytics\n\nPerformance, measurement, reporting\n\nLink supporting articles back to the relevant central topic page.\n\nThe objective is to establish a recognisable area of expertise rather than publish hundreds of unrelated articles.\n\n4. Get mentioned by other reputable websites\n\nYour own website isn't the only place where your brand should appear.\n\nBuild a genuine presence through:\n\nIndustry publications.\n\nRelevant professional associations.\n\nIndependent reviews.\n\nExpert interviews.\n\nIndustry research and reports.\n\nReputable blogs and websites.\n\nRelevant community discussions.\n\nOne particularly valuable approach is to publish original research that other websites might reference.\n\nFor example:\n\nAn original industry survey.\n\nA comparison of industry trends.\n\nA useful dataset.\n\nA detailed case study.\n\nAn independently verifiable report.\n\nThese resources can generate references and links that help people discover your website.\n\n5. Make your content easy for AI systems to understand\n\nUse good technical and editorial practices.\n\nElement\n\n\t\n\nRecommendation\n\n\n\n\nPage titles\n\n\t\n\nClear, descriptive titles\n\n\n\n\nHeadings\n\n\t\n\nLogical H1, H2 and H3 structure\n\n\n\n\nSchema markup\n\n\t\n\nRelevant Article, Organisation, Product or FAQ schema\n\n\n\n\nAuthor information\n\n\t\n\nIdentify authors and their expertise\n\n\n\n\nPublication dates\n\n\t\n\nShow publication and update dates\n\n\n\n\nStatistics\n\n\t\n\nInclude sources and dates\n\n\n\n\nInternal links\n\n\t\n\nConnect related articles\n\n\n\n\nImages\n\n\t\n\nUse descriptive alt text\n\n\n\n\nMobile performance\n\n\t\n\nEnsure pages work well on mobile\n\nOne important point: structured data and llms.txt can be useful parts of a website's technical organisation, but Perplexity does not document either as a guaranteed citation-ranking factor.\n\n6. Get your website discovered through traditional search\n\nDon't abandon traditional SEO in favour of AI search optimisation.\n\nI recommend:\n\nSet up Google Search Console.\n\nSet up Bing Webmaster Tools.\n\nSubmit your XML sitemap.\n\nEnsure important pages are indexed.\n\nResearch relevant search queries.\n\nEarn relevant, legitimate backlinks.\n\nFix duplicate content and broken links.\n\nSearch visibility and external references can create additional discovery opportunities.\n\n7. Test whether Perplexity actually cites your website\n\nDon't rely on assumptions. Test actual queries.\n\nFor example, if your website sells accounting software, search Perplexity for:\n\nBest accounting software for small businesses.\n\nHow does accounting automation work?\n\nWhat features should accounting software include?\n\nAlternatives to traditional accounting software.\n\nRecord:\n\nWhether your website appears.\n\nWhich page gets cited.\n\nWhich competitors get cited.\n\nWhich questions generate citations.\n\nWhether citations remain consistent over time.\n\nRepeat the tests periodically and record the results.\n\nA practical 30-day implementation plan\n\nWeek 1 – Technical foundations\n\nConfigure robots.txt.\n\nCheck PerplexityBot access.\n\nCheck firewall restrictions.\n\nSet up Google Search Console and Bing Webmaster Tools.\n\nSubmit your sitemap.\n\nWeek 2 – Content development\n\nIdentify 20 relevant customer questions.\n\nCreate 5–8 comprehensive articles.\n\nAdd supporting evidence and references.\n\nImprove internal linking.\n\nWeek 3 – Authority building\n\nPublish original research or useful resources.\n\nSeek relevant industry coverage.\n\nBuild relationships with reputable publishers.\n\nImprove your author and company information.\n\nWeek 4 – Testing and improvement\n\nTest 20–30 relevant Perplexity questions.\n\nRecord citations and competing sources.\n\nIdentify content gaps.\n\nImprove pages that aren't being discovered.\n\nCommon misconceptions\n\nClaim\n\n\t\n\nReality\n\n\n\n\nInstalling an llms.txt guarantees citations\n\n\t\n\nNo\n\n\n\n\nAdding FAQ schema guarantees citations\n\n\t\n\nNo\n\n\n\n\nPublishing hundreds of articles guarantees visibility\n\n\t\n\nNo\n\n\n\n\nAllowing PerplexityBot guarantees citations\n\n\t\n\nNo\n\n\n\n\nPaying for backlinks guarantees citations\n\n\t\n\nNo\n\n\n\n\nQuality, accessible content improves your opportunity\n\n\t\n\nSensible strategy, but no guarantee\n\nPerplexity's own documentation is the best starting point: Perplexity crawler documentation. \nPerplexity\n\nMy recommendation\n\nIf you're launching a completely new website, I'd prioritise these five things:\n\nTechnical accessibility: Make sure Perplexity can crawl your website.\n\nQuestion-focused content: Publish genuinely useful answers to specific questions.\n\nTopical authority: Become a credible source within one particular subject.\n\nExternal recognition: Build a genuine reputation beyond your own website.\n\nContinuous testing: Measure actual Perplexity citations rather than vanity metrics.\n\nOne final question: Are you trying to get a new website cited by Perplexity for a business, an e-commerce store, a SaaS product, or a content/blog website?\n\nThe strategy differs considerably depending on the website type. I can put together a more targeted Perplexity SEO and AI visibility strategy, including a content plan and technical checklist.",
      "citations": [
        {
          "title": "Perplexity Crawlers",
          "host": "Perplexity",
          "url": "https://docs.perplexity.ai/docs/resources/perplexity-crawlers?utm_source=chatgpt.com",
          "position": 1
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 1 sources and 1 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-get-cited-perplexity-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-get-cited-perplexity-repeat-2",
      "published_at": "2026-11-15T13:24:00.000Z",
      "updated_at": "2026-11-15T13:24:00.000Z",
      "revision": 1,
      "slug": "chatgpt-get-cited-perplexity-repeat-2",
      "panelId": "get-cited-perplexity",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-get-cited-perplexity",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "How can a new website get cited by Perplexity?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:24 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b28e-7244-83e9-83fc-189bd6577842?temporary-chat=true",
      "mentions": "Cited",
      "summary": "Getting a new website cited by Perplexity AI is increasingly an important part of SEO, often called Generative Engine Optimisation (GEO) or Answer Engine Optimisation (AEO).",
      "answerSnapshot": "Getting a new website cited by Perplexity AI is increasingly an important part of SEO, often called Generative Engine Optimisation (GEO) or Answer Engine Optimisation (AEO).\n\nThe goal is to make your website a trustworthy, accessible source that Perplexity can discover, understand, and confidently reference when answering users' questions.\n\nUnlike traditional SEO, where you primarily compete for search rankings, GEO focuses on getting your content selected as a source in AI-generated answers.\n\nThere is no guaranteed way to get cited, but you can significantly improve your chances.\n\n1. Make sure Perplexity can crawl your website\n\nThis is the first thing to check.\n\nPerplexity uses a crawler called PerplexityBot, designed to discover websites and make them available as sources in its search results.\n\nCheck your website's:\n\nTechnical priority\n\nrobots.txt\n\nVisit:\n\nhttps://yourdomain.com/robots.txt\n\nMake sure you aren't blocking PerplexityBot.\n\nYou can explicitly allow it with:\n\ntxt\nUser-agent: PerplexityBot\nAllow: /\n\nAlso check:\n\nYour hosting provider isn't blocking AI crawlers.\n\nCloudflare or other firewalls aren't challenging PerplexityBot.\n\nYour pages return HTTP 200 responses.\n\nImportant content is available as readable HTML.\n\nYour website doesn't accidentally use noindex.\n\nPerplexity also publishes its crawler IP ranges for website owners to verify. Changes to crawler permissions can take up to 24 hours to take effect. \nIgnura\n+1\n\n2. Publish content that directly answers questions\n\nThis is arguably the most important content strategy.\n\nDon't just publish generic articles targeting keywords. Create pages that answer specific questions people might ask Perplexity.\n\nFor example, instead of:\n\nLESS EFFECTIVE\n\nOur Complete Guide to Digital Marketing\n\nA broad article covering digital marketing without providing much original information.\n\nBETTER\n\nHow Much Does Digital Marketing Cost for a Small Business in 2026?\n\nA specific article with pricing examples, comparisons, practical explanations and supporting evidence.\n\nOther useful content formats include:\n\nHow-to guides\n\nIndustry statistics and original research\n\nProduct comparisons\n\nFrequently asked questions\n\nDetailed case studies\n\nExpert explanations\n\nIndustry-specific reports\n\n3. Structure your articles for AI extraction\n\nMake it easy for Perplexity to find a useful answer within your page.\n\nA good article structure looks like this:\n\nHow Much Does Website Design Cost?\n\nBy Jane Smith · Updated November 2026\n\nShort answer\n\nWebsite design costs depend on the complexity, functionality and requirements of the project. Here are the main factors that influence pricing.\n\nWhat affects website design costs?\n\nExplain the main factors clearly.\n\nTypical pricing considerations\n\nProvide a detailed breakdown with sourced figures where available.\n\nExamples and case studies\n\nInclude real examples and evidence.\n\nFrequently asked questions\n\nAnswer related questions individually.\n\nSome important principles:\n\nUse descriptive H2 and H3 headings.\n\nPut the direct answer near the beginning.\n\nKeep paragraphs focused on one subject.\n\nUse tables for meaningful comparisons.\n\nInclude statistics and link to their original sources.\n\nIdentify the author and publication date.\n\nUpdate information when it becomes outdated.\n\n4. Build authority outside your website\n\nA brand-new website has a credibility challenge: there may be very little information about it elsewhere online.\n\nBuild a consistent presence across reputable external websites.\n\nConsider:\n\nChannel\n\n\t\n\nWhat to build\n\n\n\n\nIndustry publications\n\n\t\n\nExpert articles and contributions\n\n\n\n\nRelevant directories\n\n\t\n\nAccurate company listings\n\n\n\n\nLinkedIn\n\n\t\n\nCompany and expert profiles\n\n\n\n\nIndustry websites\n\n\t\n\nRelevant mentions and references\n\n\n\n\nResearch publications\n\n\t\n\nOriginal findings and reports\n\n\n\n\nProfessional communities\n\n\t\n\nHelpful, genuine contributions\n\nThe objective is to establish your website and its authors as credible sources within a particular subject area.\n\nDon't buy artificial mentions or manufacture reviews. Focus on genuine recognition.\n\n5. Implement structured data\n\nUse Schema.org structured data to help search engines understand your website.\n\nUseful schema types include:\n\nSchema\n\n\t\n\nPurpose\n\n\n\n\nOrganization\n\n\t\n\nIdentify your company\n\n\n\n\nArticle\n\n\t\n\nDescribe editorial content\n\n\n\n\nPerson\n\n\t\n\nIdentify authors\n\n\n\n\nProduct\n\n\t\n\nDescribe products\n\n\n\n\nService\n\n\t\n\nDescribe services\n\n\n\n\nBreadcrumbList\n\n\t\n\nExplain website structure\n\nFor example, Organization schema can identify your company's name, website and official social profiles.\n\nImportant: schema markup is useful technical housekeeping, but there is no established evidence that adding schema alone guarantees Perplexity citations.\n\n6. Get your website indexed and discoverable\n\nDon't focus exclusively on Perplexity.\n\nMake sure your website is discoverable through traditional search engines too.\n\nRecommended actions:\n\nCreate an XML sitemap.\n\nSubmit your sitemap through Google Search Console.\n\nSet up Bing Webmaster Tools.\n\nEnsure important pages are internally linked.\n\nFix broken links and duplicate pages.\n\nBuild relevant external backlinks.\n\nPublish useful content consistently.\n\nThese activities improve your website's overall discoverability.\n\n7. Test whether Perplexity actually cites your website\n\nThis is where many businesses fail: they implement GEO techniques but never measure whether they work.\n\nCreate a list of 20–30 questions that potential visitors might ask.\n\nFor example, if you operate a web design agency:\n\nWho are the best web design agencies for small businesses?\n\nHow much does a professional website cost?\n\nWhat should I look for when choosing a web design agency?\n\nWhich companies specialise in ecommerce website design?\n\nRun these questions through Perplexity periodically.\n\nRecord:\n\nMetric\n\n\t\n\nWhat to measure\n\n\n\n\nBrand mentions\n\n\t\n\nDoes Perplexity mention your business?\n\n\n\n\nWebsite citations\n\n\t\n\nDoes it link to your domain?\n\n\n\n\nPage citations\n\n\t\n\nWhich individual pages are cited?\n\n\n\n\nCompetitor citations\n\n\t\n\nWhich competitors appear instead?\n\n\n\n\nCitation frequency\n\n\t\n\nHow often does your website appear?\n\nRepeat monthly using consistent questions and fresh sessions.\n\nMy recommended 30-day action plan\n\nFor a completely new website, I'd prioritise the following:\n\nWeek 1 – Technical foundations\n\nAllow PerplexityBot.\n\nCheck robots.txt and indexing.\n\nConfigure Google Search Console and Bing Webmaster Tools.\n\nCreate an XML sitemap.\n\nWeek 2 – Content creation\n\nPublish 5–10 genuinely useful articles.\n\nTarget specific questions.\n\nInclude original information and supporting evidence.\n\nAdd author information and publication dates.\n\nWeek 3 – Authority building\n\nCreate consistent company profiles.\n\nSeek relevant industry mentions.\n\nPublish expert contributions.\n\nBuild relevant backlinks.\n\nWeek 4 – Measurement\n\nTest 20–30 relevant Perplexity questions.\n\nRecord citations and brand mentions.\n\nIdentify content gaps.\n\nImprove pages that aren't being cited.\n\nUseful resources\n\nPerplexity crawler documentation – Official crawler guidance.\n\nPerplexity – Test your website's visibility.\n\nGoogle Search Console – Monitor indexing.\n\nBing Webmaster Tools – Monitor search visibility.\n\nSchema.org – Structured data documentation.\n\nOne important question\n\nAre you trying to get citations for:\n\nA. A brand-new business website (to get your company recommended by Perplexity).\n\nB. An informational website or blog (to get your articles cited as sources).\n\nC. A SaaS or technology website (to get your product recommended in AI answers).\n\nThe strategies overlap, but the most effective GEO strategy differs considerably between these three. I can put together a more targeted Perplexity citation strategy once I know which type of website you're working with.",
      "citations": [
        {
          "title": "How to get cited by Perplexity",
          "host": "Ignura",
          "url": "https://ignura.com/blog/how-to-get-cited-by-perplexity?utm_source=chatgpt.com",
          "position": 1
        },
        {
          "title": "Perplexity SEO: how to get your website cited by Perplexity",
          "host": "JW2D",
          "url": "https://jw2d.com/en/blog/perplexity-seo/?cookie_consent=rejected&utm_source=chatgpt.com",
          "position": 2
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 2 sources and 2 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-get-cited-chatgpt-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-get-cited-chatgpt-repeat-3",
      "published_at": "2026-11-15T13:22:00.000Z",
      "updated_at": "2026-11-15T13:22:00.000Z",
      "revision": 1,
      "slug": "chatgpt-get-cited-chatgpt-repeat-3",
      "panelId": "get-cited-chatgpt",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-get-cited-chatgpt",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "How can a new website get cited by ChatGPT?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:22 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b1f4-88c8-83e9-a287-c4e5a2e17097?temporary-chat=true",
      "mentions": "Cited",
      "summary": "Becoming a trusted source: Publishing useful, original, authoritative information that ChatGPT has a reason to reference.",
      "answerSnapshot": "Getting a new website cited by ChatGPT is possible, even if the website has only recently launched.\n\nThe process involves two things:\n\nGetting your website discovered: Making sure ChatGPT can access and find your pages.\n\nBecoming a trusted source: Publishing useful, original, authoritative information that ChatGPT has a reason to reference.\n\nThis is often called Generative Engine Optimisation (GEO) or AI Search Optimisation.\n\nHere is a practical strategy for a new website.\n\n1. Allow ChatGPT to crawl your website\n\nThis is the first technical step.\n\nOpenAI uses a crawler called OAI-SearchBot to discover websites for ChatGPT Search.\n\nCheck your website's robots.txt file:\n\nhttps://yourdomain.com/robots.txt\n\nMake sure it doesn't block OAI-SearchBot.\n\nFor example, you can use:\n\ntxt\nUser-agent: OAI-SearchBot\nAllow: /\n\nSitemap: https://yourdomain.com/sitemap.xml\n\nAlso check that:\n\nYour hosting provider doesn't block OpenAI's crawler.\n\nYour website doesn't require users to log in.\n\nImportant pages aren't marked noindex.\n\nYour website works on mobile and desktop.\n\nImportant: OAI-SearchBot and GPTBot are different. OAI-SearchBot is relevant to appearing in ChatGPT Search. GPTBot relates to potential model training. You can control them separately.\n\nSource: OpenAI – Overview of Crawlers.\n\n2. Get your website indexed by search engines\n\nA new website needs to become discoverable outside ChatGPT too.\n\nRecommended steps:\n\nSet up Google Search Console.\n\nSubmit your XML sitemap.\n\nSet up Bing Webmaster Tools.\n\nSubmit your website to Bing.\n\nMake sure important pages are internally linked.\n\nFix broken links and indexing errors.\n\nSearch engine visibility can help your content become discoverable, although it doesn't guarantee ChatGPT citations.\n\n3. Publish content ChatGPT can confidently cite\n\nThis is arguably the most important long-term strategy.\n\nAvoid publishing hundreds of generic AI-generated articles. Instead, create genuinely useful pages that answer specific questions.\n\nFor example:\n\nContent type\n\n\t\n\nExample\n\n\n\n\nExpert guides\n\n\t\n\nHow does X work?\n\n\n\n\nComparisons\n\n\t\n\nX vs Y: What's the difference?\n\n\n\n\nStatistics\n\n\t\n\nIndustry statistics for 2026\n\n\n\n\nResearch\n\n\t\n\nOriginal research and findings\n\n\n\n\nFAQs\n\n\t\n\nFrequently asked questions about X\n\n\n\n\nCase studies\n\n\t\n\nHow a company achieved X\n\nA particularly useful approach is to publish original information that other websites don't already have.\n\nExamples include:\n\nYour own research.\n\nSurveys and statistics.\n\nOriginal case studies.\n\nExpert interviews.\n\nDetailed product comparisons.\n\nData-backed industry reports.\n\nThese give other websites and AI search systems something substantive to reference.\n\n4. Structure articles for AI search\n\nMake your content easy to understand and extract.\n\nA good article structure looks like this:\n\nWhat Is [Your Topic]?\n\nExample article structure\n\nDirect answer\n\nStart with a concise, factual explanation in the first paragraph.\n\nHow does it work?\n\nExplain the topic in clear sections.\n\nWhat are the benefits?\n\nInclude specific facts and examples.\n\nWhat does the research show?\n\nInclude credible sources and evidence.\n\nFrequently asked questions\n\nAnswer 4–6 relevant questions.\n\nReferences\n\nLink to reliable primary sources.\n\nOther recommendations:\n\nUse descriptive H1, H2 and H3 headings.\n\nInclude author names and relevant expertise.\n\nDisplay publication and update dates.\n\nCite trustworthy references.\n\nUse structured data (Schema.org) where appropriate.\n\nKeep important information in accessible HTML rather than relying entirely on images.\n\n5. Build your website's authority\n\nA brand-new website has little established reputation.\n\nYou can gradually build authority through:\n\nGetting mentioned by established industry websites.\n\nPublishing guest articles.\n\nGetting included in relevant industry directories.\n\nEarning links from reputable websites.\n\nParticipating in relevant professional communities.\n\nPublishing research that others reference.\n\nThe goal is to establish your website as a credible source, rather than simply increasing the number of backlinks.\n\n6. Monitor whether ChatGPT references your website\n\nCreate a list of 20–30 questions that potential customers might ask ChatGPT.\n\nFor example:\n\nWhat are the best companies providing X?\n\nHow does X work?\n\nWhat are the latest developments in X?\n\nWhich providers offer X?\n\nWhat should I consider when choosing X?\n\nTest these questions periodically using ChatGPT Search.\n\nRecord:\n\nWhether your website appears.\n\nWhich pages are cited.\n\nWhich competitors are cited.\n\nWhich topics generate citations.\n\nWhether citations increase over time.\n\nYou can also monitor website analytics for referral traffic containing:\n\nutm_source=chatgpt.com\n\nOpenAI confirms that this tracking parameter is added to ChatGPT referral URLs. \nOpenAI Help Center\n\n7. A practical 30-day implementation plan\n\nWeek 1 – Technical setup\n\nCheck robots.txt.\n\nAllow OAI-SearchBot.\n\nSet up XML sitemap.\n\nConfigure Google Search Console and Bing Webmaster Tools.\n\nCheck indexing and accessibility.\n\nWeek 2 – Content development\n\nPublish 5–10 high-quality articles.\n\nTarget specific questions.\n\nAdd FAQs and factual explanations.\n\nInclude reliable references.\n\nWeek 3 – Authority building\n\nPublish original research.\n\nSeek relevant industry mentions.\n\nBuild relationships with reputable websites.\n\nStrengthen internal linking.\n\nWeek 4 – AI visibility testing\n\nTest 20–30 relevant ChatGPT queries.\n\nRecord citations.\n\nIdentify content gaps.\n\nImprove pages based on findings.\n\nNote: Thirty days is an initial implementation period, not a guarantee of indexing or citations.\n\nMy recommendation\n\nFor a new website, I would prioritise these three things:\n\nTechnical accessibility to OAI-SearchBot.\n\nHigh-quality, original content answering specific questions.\n\nBuilding credibility through independent third-party mentions and references.\n\nThere is no guaranteed way to make ChatGPT cite a website, and OpenAI does not offer a guaranteed citation placement service. \nOpenAI Help Center\n+1\n\nLet me help you develop a strategy for your website\n\nOne important question:\n\nWhat type of website are you trying to get cited by ChatGPT?\n\nFor example, an e-commerce website, SaaS business, consultancy, news website, local business, blog, or something else.\n\nI can then put together a more targeted ChatGPT SEO / GEO strategy, including technical setup, content topics, keyword opportunities and a plan to increase the likelihood of being cited in ChatGPT answers.",
      "citations": [
        {
          "title": "Publishers and Developers - FAQ",
          "host": "OpenAI Help Center",
          "url": "https://help.openai.com/en/articles/20001371-chatgpt-atlas-retirement?utm_source=chatgpt.com",
          "position": 1
        },
        {
          "title": "Overview of OpenAI Crawlers",
          "host": "OpenAI Developers",
          "url": "https://developers.openai.com/api/docs/bots?app=1&utm_source=chatgpt.com",
          "position": 2
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 2 sources and 2 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-get-cited-chatgpt",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-get-cited-chatgpt",
      "published_at": "2026-11-15T13:20:00.000Z",
      "updated_at": "2026-11-15T13:20:00.000Z",
      "revision": 1,
      "slug": "chatgpt-get-cited-chatgpt",
      "panelId": "get-cited-chatgpt",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-get-cited-chatgpt",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "How can a new website get cited by ChatGPT?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:20 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b15e-440c-83e9-b527-f0478314e43d?temporary-chat=true",
      "mentions": "Cited",
      "summary": "Getting a new website cited by ChatGPT is possible, but it requires more than simply publishing a website or submitting it to Google.",
      "answerSnapshot": "Getting a new website cited by ChatGPT is possible, but it requires more than simply publishing a website or submitting it to Google.\n\nThe goal is to make your website discoverable, trustworthy, easy to understand, and useful enough to be referenced in AI-generated answers.\n\nThis is often called Generative Engine Optimization (GEO) or Answer Engine Optimization (AEO).\n\nHere is a practical strategy for getting a new website discovered and potentially cited by ChatGPT.\n\n1. Make your website accessible to ChatGPT\n\nThe first step is ensuring that OpenAI's search crawler can access your content.\n\nOpenAI operates a crawler called OAI-SearchBot, which is used to discover content for ChatGPT search.\n\nCheck your robots.txt file and make sure you aren't accidentally blocking it.\n\nExample:\n\ntxt\nUser-agent: OAI-SearchBot\nAllow: /\n\nSitemap: https://example.com/sitemap.xml\n\nAlso:\n\nMake important pages publicly accessible.\n\nAvoid requiring login to read your content.\n\nEnsure your pages return HTTP 200 status codes.\n\nUse a sitemap.\n\nMake sure your website works on mobile.\n\nAvoid hiding important information behind JavaScript that search crawlers might not process.\n\nImportantly, allowing the crawler does not guarantee that ChatGPT will cite your website.\n\nImportant distinction: OAI-SearchBot and GPTBot are different.\n\nCrawler\n\n\t\n\nPurpose\n\n\n\n\nOAI-SearchBot\n\n\t\n\nHelps your website appear in ChatGPT Search\n\n\n\n\nGPTBot\n\n\t\n\nCrawls content that may be used to train OpenAI models\n\n\n\n\nChatGPT-User\n\n\t\n\nFetches pages in response to certain user-initiated requests\n\nYou can allow OAI-SearchBot while blocking GPTBot if you want search visibility without permitting training-related crawling. \nOpenAI Developers\n+1\n\nOfficial documentation: OpenAI crawler documentation.\n\n2. Publish content that answers specific questions\n\nThis is one of the most important content strategies.\n\nChatGPT is more likely to find a page useful when it provides a clear answer to a question someone might ask.\n\nFor example, instead of publishing:\n\n- Our company provides excellent digital marketing services.\n\nPublish useful articles such as:\n\nWhat is digital marketing?\n\nHow much does digital marketing cost?\n\nWhat are the benefits of SEO for small businesses?\n\nHow do you choose an SEO agency?\n\nSEO vs PPC: What is the difference?\n\nStructure each article clearly:\n\nA direct answer near the beginning.\n\nA detailed explanation.\n\nPractical examples.\n\nRelevant statistics or original research.\n\nReferences to authoritative sources.\n\nA short FAQ section.\n\nAvoid publishing dozens of nearly identical, superficial articles. A smaller collection of genuinely useful resources is a better starting point.\n\n3. Establish credibility and authority\n\nA new website has a disadvantage: it has little history or external recognition.\n\nBuild credibility through:\n\nAuthor biographies and relevant qualifications.\n\nAn accessible About Us page.\n\nClear contact information.\n\nOriginal research and statistics.\n\nReferences to reliable sources.\n\nCase studies with verifiable results.\n\nConsistent and accurate information.\n\nMentions and links from reputable, relevant websites.\n\nFor example, if you operate a website about accounting, publishing an original guide supported by credible financial sources is more valuable than simply rewriting existing articles.\n\n4. Make your content easy for AI systems to understand\n\nUse a logical website structure.\n\nA recommended article format:\n\nExample article structure\n\nWhat Is SEO and How Does It Work?\nDirect answer\n\nSEO (Search Engine Optimisation) is the process of improving a website's visibility in organic search results.\n\nWhat is SEO?\n\nA concise explanation of the subject.\n\nHow does SEO work?\n\nA detailed, structured explanation.\n\nWhy is SEO important?\n\nBenefits and practical examples.\n\nCommon SEO questions\n\nFrequently asked questions and direct answers.\n\nSources and references\n\nLinks to authoritative resources.\n\nTechnical recommendations:\n\nUse descriptive page titles.\n\nWrite meaningful meta descriptions.\n\nUse H1, H2 and H3 headings logically.\n\nAdd relevant internal links.\n\nUse descriptive image alt text.\n\nImplement appropriate structured data (Schema.org).\n\nKeep important information in readable HTML.\n\nUse canonical URLs to avoid duplicate-content confusion.\n\nStructured data helps search engines interpret your content, but it does not guarantee ChatGPT citations.\n\n5. Build visibility outside your website\n\nChatGPT can discover information from the wider web, so your website should not exist in isolation.\n\nDevelop a consistent online presence through:\n\nIndustry publications.\n\nRelevant professional directories.\n\nReputable educational resources.\n\nInterviews and expert contributions.\n\nGenuine industry partnerships.\n\nOriginal research that other websites may reference.\n\nThe objective is to establish your website as a credible source rather than simply creating more backlinks.\n\n6. Make sure traditional search engines can discover you\n\nDo not ignore conventional SEO.\n\nI would recommend:\n\nSetting up Google Search Console.\n\nSubmitting your XML sitemap.\n\nChecking indexing issues.\n\nSetting up Bing Webmaster Tools.\n\nMonitoring organic search impressions and clicks.\n\nImproving internal linking and site architecture.\n\nTraditional search visibility and ChatGPT visibility are not identical, but good technical SEO makes your content easier to discover and understand.\n\n7. Test whether ChatGPT actually cites your website\n\nDon't assume that being indexed means you will be cited.\n\nCreate a list of 10–20 questions that potential visitors might ask ChatGPT.\n\nFor example, if your website offers web design services:\n\nWho are the best web design companies for small businesses?\n\nHow much does a professional website cost?\n\nWhat should I look for when choosing a web designer?\n\nWhich companies specialise in e-commerce websites?\n\nTest these questions using ChatGPT Search and record:\n\nWhether your website appears.\n\nWhich page is cited.\n\nWhich competitors are cited.\n\nWhether the information attributed to your website is accurate.\n\nYou can also monitor referral traffic in your website analytics. OpenAI documents the use of utm_source=chatgpt.com in referral URLs. \nOpenAI Help Center\n\n8. A practical 30-day implementation plan\n\nPeriod\n\n\t\n\nMain activity\n\n\n\n\nDays 1–5\n\n\t\n\nTechnical SEO, robots.txt, sitemap and crawler access\n\n\n\n\nDays 6–10\n\n\t\n\nPublish 5–10 high-quality articles\n\n\n\n\nDays 11–15\n\n\t\n\nAdd structured data, FAQs and internal links\n\n\n\n\nDays 16–20\n\n\t\n\nBuild credible industry mentions and references\n\n\n\n\nDays 21–25\n\n\t\n\nImprove existing content and identify gaps\n\n\n\n\nDays 26–30\n\n\t\n\nTest ChatGPT citations and measure traffic\n\nThese are suggested implementation stages, not guaranteed indexing or citation deadlines.\n\nWhat matters most?\n\nMy recommended priorities are:\n\n01\n\nCrawler accessibility\n\nAllow OAI-SearchBot and check hosting restrictions.\n\n02\n\nHigh-quality content\n\nPublish original, detailed answers to real questions.\n\n03\n\nCredibility\n\nBuild a reputation through reliable information and industry references.\n\n04\n\nTechnical SEO\n\nMake pages accessible, structured and easy to navigate.\n\n05\n\nMeasurement\n\nTest real ChatGPT queries and monitor citations.\n\nI can help you develop a ChatGPT SEO strategy for your website\n\nOne useful starting point is to understand what type of website you're trying to get cited.\n\nWhat is your website about?\n\nFor example, is it:\n\nA business or company website?\n\nAn e-commerce website?\n\nA blog or information website?\n\nA professional services website?\n\nA newly launched startup?\n\nIf you share the website URL, I can also help assess its technical accessibility, content structure and opportunities to improve its chances of appearing in ChatGPT Search.\n\nThe key point: There is no guaranteed way to get cited, but you can systematically improve your website's eligibility, discoverability and usefulness as a source.\n\n你正在使用 ChatGPT 免费版。升级即可使用高级模型、更强的记忆功能，并生成更高质量的图像。\n免费试用 Plus",
      "citations": [
        {
          "title": "Overview of OpenAI Crawlers",
          "host": "OpenAI Developers",
          "url": "https://developers.openai.com/api/docs/bots?utm_source=chatgpt.com",
          "position": 1
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 1 sources and 1 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-get-cited-chatgpt-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-get-cited-chatgpt-repeat-2",
      "published_at": "2026-11-15T13:20:00.000Z",
      "updated_at": "2026-11-15T13:20:00.000Z",
      "revision": 1,
      "slug": "chatgpt-get-cited-chatgpt-repeat-2",
      "panelId": "get-cited-chatgpt",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-get-cited-chatgpt",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "How can a new website get cited by ChatGPT?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:20 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b187-0db4-83e9-aed7-94645fc3c0dd?temporary-chat=true",
      "mentions": "Cited",
      "summary": "Getting a new website cited by ChatGPT is possible, but it requires more than traditional SEO. You need to make your website discoverable, crawlable, trustworthy, and useful enough for ChatGPT to select as a source.",
      "answerSnapshot": "Getting a new website cited by ChatGPT is possible, but it requires more than traditional SEO. You need to make your website discoverable, crawlable, trustworthy, and useful enough for ChatGPT to select as a source.\n\nThis is increasingly known as Generative Engine Optimisation (GEO) or AI Search Optimisation (AISO).\n\nThe important distinction is that getting your website indexed by search engines does not automatically mean ChatGPT will cite it.\n\n1. Make sure ChatGPT can access your website\n\nStart with the technical foundations.\n\nEnsure your website is publicly accessible.\n\nAllow relevant search crawlers in your robots.txt.\n\nAvoid blocking important pages behind logins or paywalls.\n\nSubmit an XML sitemap to search engines.\n\nUse clear URLs and logical internal links.\n\nEnsure important content is available as readable HTML, rather than only inside images or scripts.\n\n2. Allow OpenAI's search crawler\nMost important technical step\n\nOpenAI operates a crawler called OAI-SearchBot. This is the crawler specifically used to discover websites for ChatGPT Search.\n\nIt is different from GPTBot, which is associated with crawling content that may be used to train OpenAI's AI models. You can allow one and block the other. \nOpenAI Developers\n+1\n\nCheck your website's robots.txt file:\n\nExample:\n\ntxt\nUser-agent: OAI-SearchBot\nAllow: /\n\nUser-agent: GPTBot\nDisallow: /\n\nSitemap: https://www.example.com/sitemap.xml\n\nReplace www.example.com with your domain.\n\nThis example allows ChatGPT Search to crawl your site while opting out of GPTBot crawling for potential training use.\n\nAlso check: Your hosting provider, CDN or firewall must not block OpenAI's published crawler IP addresses.\n\nOpenAI says changes to robots.txt can take approximately 24 hours to be reflected in its search systems. \nOpenAI Developers\n\nOfficial documentation: OpenAI – Overview of OpenAI Crawlers.\n\n3. Create content that ChatGPT wants to cite\n\nThis is where the real SEO opportunity lies.\n\nChatGPT needs useful information to answer questions. Your website should contain original, detailed and factual information that directly answers questions people ask.\n\nFor example, instead of publishing:\n\n- Our company provides excellent digital marketing services.\n\nPublish useful resources such as:\n\nHow much does digital marketing cost in 2026?\n\nWhat is the difference between SEO and GEO?\n\nHow long does SEO take to produce results?\n\nWhat are the most common SEO mistakes?\n\nHow do you measure organic search performance?\n\nA good article should contain:\n\nA clear answer near the beginning.\n\nDescriptive headings organised around specific questions.\n\nFacts, figures and examples.\n\nOriginal research or practical experience.\n\nLinks to credible supporting sources.\n\nAn author name and relevant credentials.\n\nA publication date and update date.\n\nMy recommendation: Build pages that provide information competitors don't already offer. Original research, useful statistics, case studies and expert explanations give your website something distinctive to contribute.\n\n4. Build your website's authority\n\nChatGPT needs to find reliable information, not just well-written marketing content.\n\nWork on building your website's reputation across the wider internet.\n\nSome useful activities:\n\nGet mentioned by reputable industry publications.\n\nPublish original research that others may reference.\n\nEarn links from relevant, established websites.\n\nBuild relationships with industry organisations.\n\nMaintain consistent company information across legitimate business directories.\n\nPublish expert commentary and educational resources.\n\nAvoid buying hundreds of low-quality backlinks or generating large volumes of repetitive AI-written articles.\n\n5. Structure your content for AI discovery\n\nMake it easy for search engines and AI systems to understand what each page contains.\n\nConsider implementing:\n\nElement\n\n\t\n\nPurpose\n\n\n\n\nSchema markup\n\n\t\n\nDescribes content in a structured format\n\n\n\n\nArticle schema\n\n\t\n\nIdentifies articles and their authors\n\n\n\n\nOrganisation schema\n\n\t\n\nDescribes your business\n\n\n\n\nFAQ content\n\n\t\n\nAnswers specific user questions\n\n\n\n\nCanonical tags\n\n\t\n\nIdentifies the preferred version of a page\n\n\n\n\nXML sitemap\n\n\t\n\nHelps crawlers discover pages\n\n\n\n\nInternal links\n\n\t\n\nConnects related information\n\nSchema markup helps machines interpret your content, but it does not guarantee ChatGPT citations.\n\nSimilarly, an llms.txt file is not a documented requirement for ChatGPT Search inclusion.\n\n6. Get your website mentioned elsewhere\n\nAn important part of AI visibility is your website's presence beyond its own domain.\n\nFor example, imagine you own a website about accounting software.\n\nYou could publish an in-depth guide on your website, then build genuine industry recognition through:\n\nRelevant professional publications.\n\nIndependent product reviews.\n\nIndustry research reports.\n\nEducational articles on established websites.\n\nInterviews and expert contributions.\n\nThe goal is to establish your website as a recognisable and credible source within its subject area.\n\n7. Test whether ChatGPT cites your website\n\nDon't assume that being indexed means you're being cited.\n\nTest ChatGPT using questions your potential visitors might ask.\n\nFor example:\n\nSearch question\n\n\t\n\nWhat to check\n\n\n\n\nWhat is the best solution for X?\n\n\t\n\nDoes your website appear?\n\n\n\n\nHow do I solve problem Y?\n\n\t\n\nIs your article cited?\n\n\n\n\nWhich companies provide X?\n\n\t\n\nIs your business mentioned?\n\n\n\n\nWhat does X cost?\n\n\t\n\nDoes ChatGPT reference your pricing page?\n\nRepeat these tests periodically and record the results.\n\nYou can also monitor referral traffic in Google Analytics. OpenAI adds utm_source=chatgpt.com to referral URLs, making ChatGPT traffic identifiable. \nOpenAI Help Center\n\nA practical 30-day implementation plan\n\nHere's how I would approach launching a new website with ChatGPT visibility as one of its objectives.\n\nWeek 1 – Technical foundations\n\nCheck robots.txt.\n\nAllow OAI-SearchBot.\n\nCheck firewall and CDN access.\n\nCreate an XML sitemap.\n\nSet up Google Search Console and Bing Webmaster Tools.\n\nWeek 2 – Content creation\n\nPublish 5–10 genuinely useful articles.\n\nTarget specific questions.\n\nInclude original information and examples.\n\nAdd author information and references.\n\nWeek 3 – Authority building\n\nSeek relevant industry mentions.\n\nPublish original research.\n\nBuild relationships with reputable websites.\n\nImprove internal linking.\n\nWeek 4 – AI visibility testing\n\nTest 20–30 relevant questions in ChatGPT.\n\nRecord citations and competitor appearances.\n\nImprove pages that aren't being discovered.\n\nMonitor referral traffic.\n\n8. What matters most?\n\nMy suggested priorities:\n\nActivity\n\n\t\n\nPriority\n\n\n\n\nAllow OAI-SearchBot\n\n\t\n\nEssential\n\n\n\n\nPublish genuinely useful content\n\n\t\n\nVery high\n\n\n\n\nBuild website authority\n\n\t\n\nVery high\n\n\n\n\nEarn reputable external mentions\n\n\t\n\nHigh\n\n\n\n\nImplement structured data\n\n\t\n\nMedium\n\n\n\n\nCreate an llms.txt file\n\n\t\n\nExperimental\n\n\n\n\nPublish hundreds of AI articles\n\n\t\n\nNot recommended\n\nOne important point: OpenAI does not guarantee placement or citations, even when a website allows its crawler. There is no guaranteed shortcut to getting cited. \nOpenAI Developers\n+1\n\nLet's take this a step further\n\nIf your objective is to get a new website cited by ChatGPT, I can help develop a complete ChatGPT SEO / GEO strategy for it.\n\nThe most useful starting point is:\n\nDo you already have a website, or are you planning to launch a new one?\n\nI can then help you create a technical checklist, content strategy and AI visibility plan appropriate to its stage.",
      "citations": [
        {
          "title": "Overview of OpenAI Crawlers",
          "host": "OpenAI Developers",
          "url": "https://developers.openai.com/api/docs/bots?utm_source=chatgpt.com",
          "position": 1
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 1 sources and 1 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-source-types-cited-most-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-source-types-cited-most-repeat-3",
      "published_at": "2026-11-15T13:19:00.000Z",
      "updated_at": "2026-11-15T13:19:00.000Z",
      "revision": 1,
      "slug": "chatgpt-source-types-cited-most-repeat-3",
      "panelId": "source-types-cited-most",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-source-types-cited-most",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "Which source types do AI search engines cite most often?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:19 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b135-ee94-83ea-82cb-5a8db8d25264?temporary-chat=true",
      "mentions": "Cited",
      "summary": "AI search engines most often cite web pages, especially established informational websites, news publications, specialist blogs, and reference sources.",
      "answerSnapshot": "AI search engines most often cite web pages, especially established informational websites, news publications, specialist blogs, and reference sources.\n\nThe most common source types generally fall into these categories:\n\n1. Most frequently cited source types\n\nSource type\n\n\t\n\nExamples\n\n\t\n\nWhy AI engines cite them\n\n\n\n\nOfficial websites\n\n\t\n\nCompany websites, government sites, product documentation\n\n\t\n\nFirst-hand, authoritative information\n\n\n\n\nNews and media\n\n\t\n\nReuters, BBC, Forbes, industry publications\n\n\t\n\nCurrent information and reporting\n\n\n\n\nExpert blogs and niche publishers\n\n\t\n\nSpecialist blogs, technical guides, industry websites\n\n\t\n\nDetailed subject knowledge\n\n\n\n\nReview and comparison sites\n\n\t\n\nProduct reviews, software comparisons, buyer guides\n\n\t\n\nUseful for evaluating alternatives\n\n\n\n\nListicles and rankings\n\n\t\n\nBest-of articles, top-10 lists, recommended tools\n\n\t\n\nStructured recommendations\n\n\n\n\nCommunity discussions\n\n\t\n\nReddit, forums, Q&A websites\n\n\t\n\nReal-world experiences and opinions\n\n\n\n\nVideo platforms\n\n\t\n\nYouTube tutorials, demonstrations, interviews\n\n\t\n\nVisual explanations and practical demonstrations\n\n\n\n\nResearch and reference sources\n\n\t\n\nAcademic papers, universities, encyclopedias\n\n\t\n\nEvidence and verifiable facts\n\nA useful finding from a 2026 academic citation analysis is that three broad categories accounted for most citations:\n\nOfficial sources: 34–46%, depending on the AI engine.\n\nNews sources: 16–31%.\n\nSpecialist or vertical publications: approximately 19–22%.\n\nThe study analysed 21,143 citations across ChatGPT, Google AI surfaces and Perplexity. \nBavior\n\n2. Differences between AI search engines\n\nDifferent platforms have different citation preferences.\n\nAI search engine\n\n\t\n\nNotable source preferences\n\n\n\n\nChatGPT\n\n\t\n\nOfficial company pages, product pages, reviews and comparison articles\n\n\n\n\nPerplexity\n\n\t\n\nNews publications, specialist publishers and comparison articles\n\n\n\n\nGoogle AI Overviews\n\n\t\n\nCommunity discussions, official websites and video\n\n\n\n\nGemini\n\n\t\n\nOfficial websites, community content, specialist publishers and video\n\n\n\n\nClaude\n\n\t\n\nSpecialist publications, comparison sites and news\n\nFor example, a 2026 study of 2,309 citations across four AI engines found that review and comparison publishers represented 14% of citations, company websites 11.3%, and general news/business press another 11.3%. Google AI Overviews showed a substantially different pattern, with community and social sources accounting for approximately 34% of citations. \n5W Public Relations\n+1\n\n3. Which content formats get cited most?\n\nThe type of page matters just as much as the type of website.\n\nCommonly cited formats include:\n\nComparison articles: Product A vs Product B.\n\nBest-of lists: Best software, services or products in a category.\n\nIn-depth guides: Explanations of specific topics.\n\nOriginal research: Surveys, statistics and industry reports.\n\nProduct pages: Features, specifications and pricing.\n\nHow-to articles: Step-by-step educational content.\n\nFAQ pages: Direct answers to specific questions.\n\nA 2026 study analysing 75,000 AI answers found that listicles represented 21.88% of cited page types, articles 16.68%, and product pages 13.66%. \nAI Rank Lab\n\n4. The biggest takeaway\n\nIf you're studying AI search visibility (also called Generative Engine Optimisation or GEO), the evidence suggests three important things:\n\nAuthority matters: AI engines frequently reference official and specialist sources.\n\nThird-party coverage matters: Independent reviews, comparisons and industry publications can provide valuable citations.\n\nContent format matters: Pages that directly answer questions, compare options and provide verifiable evidence are useful citation candidates.\n\nOne important caveat: there is no universal ranking. Citation patterns change depending on the platform, industry, question and research methodology.\n\nMy overall conclusion: Official sources, specialist publications, news websites and comparison articles are particularly important source categories to understand when analysing AI search citations. Community discussions and video also matter, particularly on certain platforms.\n\n你正在使用 ChatGPT 免费版。升级即可使用高级模型、更强的记忆功能，并生成更高质量的图像。\n免费试用 Plus",
      "citations": [
        {
          "title": "What Sources Get Cited by AI Search Engines",
          "host": "Bavior",
          "url": "https://tunnel-check.bavior.com/learn/what-sources-get-cited?utm_source=chatgpt.com",
          "position": 1
        },
        {
          "title": "What 1,454 AI Citations Across 7 Engines Show",
          "host": "Conduit",
          "url": "https://www.conduit.ai/blog/ai-search-citation-study?utm_source=chatgpt.com",
          "position": 2
        },
        {
          "title": "Content Formats for AI Search: 19 Citation Studies Compared",
          "host": "AI Rank Lab",
          "url": "https://www.airanklab.com/zh-CN/blog/content-formats-ai-citations?utm_source=chatgpt.com",
          "position": 3
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 3 sources and 3 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-source-types-cited-most-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-source-types-cited-most-repeat-2",
      "published_at": "2026-11-15T13:18:00.000Z",
      "updated_at": "2026-11-15T13:18:00.000Z",
      "revision": 1,
      "slug": "chatgpt-source-types-cited-most-repeat-2",
      "panelId": "source-types-cited-most",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-source-types-cited-most",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "Which source types do AI search engines cite most often?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:18 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b10d-6554-83ea-9cf1-07746801c859?temporary-chat=true",
      "mentions": "Cited",
      "summary": "AI search engines most often cite official websites, news publishers, specialist publications, company websites, and community platforms. However, the ranking varies considerably between engines such as ChatGPT, Perplexity, Gemini and Google AI Overviews.",
      "answerSnapshot": "AI search engines most often cite official websites, news publishers, specialist publications, company websites, and community platforms. However, the ranking varies considerably between engines such as ChatGPT, Perplexity, Gemini and Google AI Overviews.\n\nBased on several 2026 citation studies, here's a breakdown.\n\n1. Most frequently cited source types\n\nA 2026 study analysing 5.31 million citations across five AI search engines found significant differences in their source preferences. Other smaller studies show similar patterns, although their classifications differ.\n\nSource type\n\n\t\n\nExamples\n\n\t\n\nWhy AI cites them\n\n\n\n\nOfficial websites\n\n\t\n\nGovernment, universities, organisations\n\n\t\n\nAuthoritative information\n\n\n\n\nNews publishers\n\n\t\n\nBBC, Reuters, Forbes\n\n\t\n\nReporting and current information\n\n\n\n\nSpecialist publications\n\n\t\n\nIndustry blogs, technical journals\n\n\t\n\nSubject expertise\n\n\n\n\nCompany websites\n\n\t\n\nProduct pages, documentation\n\n\t\n\nFirst-party information\n\n\n\n\nReview websites\n\n\t\n\nG2, Trustpilot, comparison sites\n\n\t\n\nProduct evaluations\n\n\n\n\nCommunity platforms\n\n\t\n\nReddit, forums, Q&A sites\n\n\t\n\nFirst-hand experiences\n\n\n\n\nReference websites\n\n\t\n\nWikipedia, encyclopedias\n\n\t\n\nGeneral knowledge\n\n\n\n\nVideo platforms\n\n\t\n\nYouTube\n\n\t\n\nDemonstrations and explanations\n\nVercite\n+2\n2. Which sources do individual AI engines prefer?\n\nChatGPT\n\nStrong use of Reddit, company websites, technical sources and news publishers. One large citation dataset identified Reddit as its most-cited individual domain.\n\nPerplexity\n\nStrong use of news publishers, review websites, YouTube and community discussions.\n\nGemini\n\nUses company websites, community platforms, YouTube and review websites.\n\nGoogle AI Overviews\n\nParticularly strong use of community and social sources, alongside company websites.\n\nSources: Vercite's 5.31-million-citation study and 5WPR's 2026 research. \nVercite\n+1\n\n3. What does the research tell us?\n\nThree particularly interesting findings:\n\nReddit matters: ChatGPT frequently cites Reddit, while other engines show different preferences.\n\nOfficial sources matter: Government, university and company websites are important sources of factual information.\n\nThere is no universal winner: A website frequently cited by one AI engine might rarely appear in another.\n\nVercite\n+2\n4. Overall takeaway\n\nIf you're studying how AI search engines select their sources, I'd group the most important categories into five:\n\nOfficial and authoritative websites.\n\nEstablished news and editorial publishers.\n\nSpecialist and industry publications.\n\nCompany-owned websites and documentation.\n\nCommunity-generated content and discussion platforms.\n\nOne important distinction: The most frequently cited source isn't necessarily the most trusted or influential source. Citation frequency, factual reliability and influence on the generated answer are different measurements.\n\nFor further reading:\n\nVercite – The AI Citation Landscape\n\n5WPR – State of AI Search 2026\n\nBavior – AI Search Citation Research",
      "citations": [
        {
          "title": "The AI citation landscape",
          "host": "Vercite",
          "url": "https://vercite.io/research/citation-landscape?utm_source=chatgpt.com",
          "position": 1
        },
        {
          "title": "State of AI Search 2026: No Single Brand Leads",
          "host": "5W Public Relations",
          "url": "https://www.5wpr.com/research/state-of-ai-search-2026/?utm_source=chatgpt.com",
          "position": 2
        },
        {
          "title": "What Sources Get Cited by AI Search Engines",
          "host": "Bavior",
          "url": "https://tunnel-check.bavior.com/learn/what-sources-get-cited?utm_source=chatgpt.com",
          "position": 3
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 3 sources and 3 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/chatgpt-source-types-cited-most",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/chatgpt-source-types-cited-most",
      "published_at": "2026-11-15T13:17:00.000Z",
      "updated_at": "2026-11-15T13:17:00.000Z",
      "revision": 1,
      "slug": "chatgpt-source-types-cited-most",
      "panelId": "source-types-cited-most",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "chatgpt-source-types-cited-most",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "Which source types do AI search engines cite most often?",
      "product": "ChatGPT",
      "observedAt": "15 November 2026, 21:17 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "ChatGPT web with web search",
      "model": "ChatGPT automatic model (model version not exposed)",
      "responseUrl": "https://chatgpt.com/c/6af9b029-54dc-83e9-b4ad-3bcfc7c843bc?temporary-chat=true",
      "mentions": "Cited",
      "summary": "AI search engines most often cite websites with clear, authoritative, well-structured information. However, the source types they favour vary depending on the search engine, the topic, and the user's question.",
      "answerSnapshot": "AI search engines most often cite websites with clear, authoritative, well-structured information. However, the source types they favour vary depending on the search engine, the topic, and the user's question.\n\nAcross AI-powered search experiences, the main source categories to pay attention to are:\n\nNews and media websites – reporting, current events, expert commentary and industry news.\n\nGovernment and official websites – statistics, regulations, public information and official guidance.\n\nResearch and academic publications – scientific studies, journal articles and university research.\n\nCorporate and brand websites – product information, company details, service descriptions and original research.\n\nReference and educational websites – encyclopedias, specialist explainers and educational resources.\n\nForums and community platforms – Reddit discussions, Q&A websites and first-hand experiences.\n\nIndustry publications and specialist blogs – niche expertise, tutorials, comparisons and technical analysis.\n\nWhat research suggests about citation frequency\n\nA 2026 study by 5WPR analysed 2,309 citations across four AI engines. Its findings provide a useful indication of the source mix:\n\nSource type\n\n\t\n\nShare of citations\n\n\n\n\nReview and comparison publishers\n\n\t\n\n14.0%\n\n\n\n\nCompany and brand websites\n\n\t\n\n11.3%\n\n\n\n\nGeneral news and business press\n\n\t\n\n11.3%\n\n\n\n\nSports and general news media\n\n\t\n\n5.3%\n\n\n\n\nCommunity and social platforms\n\n\t\n\n4.0%\n\n\n\n\nReference websites\n\n\t\n\n2.1%\n\n\n\n\nHealth publishers\n\n\t\n\n1.9%\n\n\n\n\nResearch and data providers\n\n\t\n\n1.7%\n\n\n\n\nOfficial government websites\n\n\t\n\n0.7%\n\n44.3% were classified as other or unclassified. \n5W Public Relations\n\nHow the engines differ\n\nChatGPT\n\nCompany and brand websites, followed by review and comparison publishers.\n\nPerplexity\n\nNews and business press, followed by review and comparison publishers.\n\nGemini\n\nCompany websites, community platforms and comparison publishers.\n\nGoogle AI Overviews\n\nCommunity and social platforms, followed by company websites.\n\nClaude\n\nReview and comparison publishers, followed by company websites and media.\n\nSource: 5WPR's 2026 citation analysis. \n5W Public Relations\n\nWhat this means for content strategy\n\nIf your objective is to increase the likelihood of being cited by AI search engines, consider prioritising:\n\nOriginal research: Publish proprietary statistics, surveys and data.\n\nDetailed guides: Answer specific questions with clear, factual explanations.\n\nComparison articles: Provide useful, evidence-based product and service comparisons.\n\nStrong brand pages: Make your products, services and expertise easy to understand.\n\nIndustry coverage: Build visibility through reputable publications.\n\nCommunity participation: Contribute useful expertise to relevant discussions.\n\nOne important qualification: these figures are observational, not universal rankings. Different studies use different queries, industries, engines and classification methods. For example, a separate 2026 study found that page formats such as listicles dominated certain industries but not others. \nDeltaV Digital\n+1\n\nBottom line: There isn't one source type that dominates every AI search engine. Company websites, specialist publishers, comparison articles and news sources are particularly important, while community sources can be especially influential in Google AI Overviews.",
      "citations": [
        {
          "title": "State of AI Search 2026: No Single Brand Leads",
          "host": "5W Public Relations",
          "url": "https://www.5wpr.com/research/state-of-ai-search-2026/?utm_source=chatgpt.com",
          "position": 1
        },
        {
          "title": "AI search citations study: what 25,000+ citations reveal",
          "host": "DeltaV Digital",
          "url": "https://www.deltavdigital.com/resources/reports/ai-citation-study/?utm_source=chatgpt.com",
          "position": 2
        },
        {
          "title": "What 1,454 AI Citations Across 7 Engines Show",
          "host": "Conduit",
          "url": "https://www.conduit.ai/blog/ai-search-citation-study?utm_source=chatgpt.com",
          "position": 3
        }
      ],
      "limitations": "Authenticated free-account ChatGPT browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 3 sources and 3 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-audit-ai-answer-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-audit-ai-answer-repeat-2",
      "published_at": "2026-11-15T13:07:00.000Z",
      "updated_at": "2026-11-15T13:07:00.000Z",
      "revision": 1,
      "slug": "perplexity-audit-ai-answer-repeat-2",
      "panelId": "audit-ai-answer",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-audit-ai-answer",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "How should a publisher audit an AI-generated answer?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 21:07 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/49c3cca0-093d-48e3-ba87-269368541130",
      "mentions": "Cited",
      "summary": "Audit it claim by claim before publication: check each factual statement against a current, authoritative source; verify quotes, figures, dates, and named entities directly; and flag unsupported claims, misattributions, stale information, or contradictions.",
      "answerSnapshot": "8秒\n\nAudit it claim by claim before publication: check each factual statement against a current, authoritative source; verify quotes, figures, dates, and named entities directly; and flag unsupported claims, misattributions, stale information, or contradictions.\ncertant\n+2\n\nAssign a named editor to approve the answer, and log the question, answer, sources, model version, review date, and verdict. Sample high-risk answers first, then broaden the review if errors appear.\ncertant\n+2",
      "citations": [
        {
          "title": "How to Audit AI Generated Answers: A 2026 Guide — Certant Blog",
          "host": "certant",
          "url": "https://certant.ai/blog/how-to-audit-ai-generated-answers",
          "position": 1
        },
        {
          "title": "How to Verify AI-Generated Answers and Citations - AOFIRS",
          "host": "aofirs",
          "url": "https://aofirs.org/articles/how-to-verify-ai-generated-answers-and-citations/",
          "position": 2
        },
        {
          "title": "Frequently Asked Questions",
          "host": "drdaveheath",
          "url": "https://drdaveheath.com/blog/a-simple-audit-process-for-ai-written-content-before-publishing",
          "position": 3
        },
        {
          "title": "Fact Check AI Content: 7 Steps to Catch False Claims",
          "host": "aiflowreview",
          "url": "https://aiflowreview.com/fact-check-ai-content/",
          "position": 4
        },
        {
          "title": "How to Audit AI-Generated Content for Accuracy, EEAT & Trust",
          "host": "trustedaiseo",
          "url": "https://trustedaiseo.com/audit-ai-generated-content-eeat/",
          "position": 5
        },
        {
          "title": "AI Search Audit for Publishers and Review Websites",
          "host": "citeworksstudio",
          "url": "https://citeworksstudio.com/resources/research/ai-search-audit-publishers-review-websites",
          "position": 6
        },
        {
          "title": "Fact-checking AI content: why verification has to be a pipeline stage",
          "host": "avoidcontent",
          "url": "https://avoidcontent.com/blog/fact-checking-ai-content/",
          "position": 7
        },
        {
          "title": "AI Content Publishing Controls: Review, Provenance, and Audience Trust",
          "host": "aitrendfast",
          "url": "https://aitrendfast.com/blog/ai-content-publishing-controls",
          "position": 8
        },
        {
          "title": "How to Fact-Check AI Content Before Publishing",
          "host": "searchfit",
          "url": "https://searchfit.ai/blog/fact-check-ai-content-before-publish",
          "position": 9
        },
        {
          "title": "AI-Generated Content for SEO: What to Verify Before Publishing",
          "host": "seoinformatica",
          "url": "https://seoinformatica.com/ai-generated-content-seo/",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-audit-ai-answer-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-audit-ai-answer-repeat-3",
      "published_at": "2026-11-15T13:07:00.000Z",
      "updated_at": "2026-11-15T13:07:00.000Z",
      "revision": 1,
      "slug": "perplexity-audit-ai-answer-repeat-3",
      "panelId": "audit-ai-answer",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-audit-ai-answer",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "How should a publisher audit an AI-generated answer?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 21:07 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/e713733c-db0f-4a4b-946b-6f743702fe00",
      "mentions": "Cited",
      "summary": "Audit it claim by claim before publication: verify each factual statement against a current, authoritative source, ensure the citation supports that exact claim, and check for fabricated details, misattribution, outdated information, contradictions, and claims that exceed the evidence.",
      "answerSnapshot": "7秒\n\nAudit it claim by claim before publication: verify each factual statement against a current, authoritative source, ensure the citation supports that exact claim, and check for fabricated details, misattribution, outdated information, contradictions, and claims that exceed the evidence. Prioritize high-risk claims—such as statistics, quotations, legal or regulatory statements, and product claims—and remove or qualify anything you can’t verify.\ncertant\n+2\n\nKeep a record of the answer, sources, model/version, review date, and editor’s decision; then sample published answers for errors and correct them when needed.\ncertant\n+1",
      "citations": [
        {
          "title": "How to Audit AI Generated Answers: A 2026 Guide — Certant Blog",
          "host": "certant",
          "url": "https://certant.ai/blog/how-to-audit-ai-generated-answers",
          "position": 1
        },
        {
          "title": "How to Verify AI-Generated Answers and Citations - AOFIRS",
          "host": "aofirs",
          "url": "https://aofirs.org/articles/how-to-verify-ai-generated-answers-and-citations/",
          "position": 2
        },
        {
          "title": "Frequently Asked Questions",
          "host": "drdaveheath",
          "url": "https://drdaveheath.com/blog/a-simple-audit-process-for-ai-written-content-before-publishing",
          "position": 3
        },
        {
          "title": "AI Content Publishing Controls: Review, Provenance, and Audience Trust",
          "host": "aitrendfast",
          "url": "https://aitrendfast.com/blog/ai-content-publishing-controls",
          "position": 4
        },
        {
          "title": "How to Audit AI-Generated Content for Accuracy, EEAT & Trust",
          "host": "trustedaiseo",
          "url": "https://trustedaiseo.com/audit-ai-generated-content-eeat/",
          "position": 5
        },
        {
          "title": "AI Search Audit for Publishers and Review Websites",
          "host": "citeworksstudio",
          "url": "https://citeworksstudio.com/resources/research/ai-search-audit-publishers-review-websites",
          "position": 6
        },
        {
          "title": "Fact-checking AI content: why verification has to be a pipeline stage",
          "host": "avoidcontent",
          "url": "https://avoidcontent.com/blog/fact-checking-ai-content/",
          "position": 7
        },
        {
          "title": "Fact Check AI Content: 7 Steps to Catch False Claims",
          "host": "aiflowreview",
          "url": "https://aiflowreview.com/fact-check-ai-content/",
          "position": 8
        },
        {
          "title": "How to Fact-Check AI Content Before Publishing",
          "host": "searchfit",
          "url": "https://searchfit.ai/blog/fact-check-ai-content-before-publish",
          "position": 9
        },
        {
          "title": "AI-Generated Content for SEO: What to Verify Before Publishing",
          "host": "seoinformatica",
          "url": "https://seoinformatica.com/ai-generated-content-seo/",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-audit-ai-answer",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-audit-ai-answer",
      "published_at": "2026-11-15T13:06:00.000Z",
      "updated_at": "2026-11-15T13:06:00.000Z",
      "revision": 1,
      "slug": "perplexity-audit-ai-answer",
      "panelId": "audit-ai-answer",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-audit-ai-answer",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "How should a publisher audit an AI-generated answer?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 21:06 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/b4ee4844-b42e-487d-8b55-0e74a3167a3e",
      "mentions": "Cited",
      "summary": "Audit it claim by claim before publication: check each factual statement against a current, authoritative source, verify that citations genuinely support the claims, and flag contradictions, fabricated details, stale facts, or claims outside the source’s scope.",
      "answerSnapshot": "9秒\n\nAudit it claim by claim before publication: check each factual statement against a current, authoritative source, verify that citations genuinely support the claims, and flag contradictions, fabricated details, stale facts, or claims outside the source’s scope. Prioritize high-risk claims such as statistics, dates, quotations, and legal or product information.\ncertant\n+2\n\nKeep a record of the answer, sources, model version, review date, and editor’s verdict; publish only after a named editor approves it, and monitor for corrections afterward.\ncertant\n+1",
      "citations": [
        {
          "title": "How to Audit AI Generated Answers: A 2026 Guide — Certant Blog",
          "host": "certant",
          "url": "https://certant.ai/blog/how-to-audit-ai-generated-answers",
          "position": 1
        },
        {
          "title": "How to Verify AI-Generated Answers and Citations - AOFIRS",
          "host": "aofirs",
          "url": "https://aofirs.org/articles/how-to-verify-ai-generated-answers-and-citations/",
          "position": 2
        },
        {
          "title": "Frequently Asked Questions",
          "host": "drdaveheath",
          "url": "https://drdaveheath.com/blog/a-simple-audit-process-for-ai-written-content-before-publishing",
          "position": 3
        },
        {
          "title": "AI Content Publishing Controls: Review, Provenance, and Audience Trust",
          "host": "aitrendfast",
          "url": "https://aitrendfast.com/blog/ai-content-publishing-controls",
          "position": 4
        },
        {
          "title": "How to Audit AI-Generated Content for Accuracy, EEAT & Trust",
          "host": "trustedaiseo",
          "url": "https://trustedaiseo.com/audit-ai-generated-content-eeat/",
          "position": 5
        },
        {
          "title": "AI Search Audit for Publishers and Review Websites",
          "host": "citeworksstudio",
          "url": "https://citeworksstudio.com/resources/research/ai-search-audit-publishers-review-websites",
          "position": 6
        },
        {
          "title": "Fact-checking AI content: why verification has to be a pipeline stage",
          "host": "avoidcontent",
          "url": "https://avoidcontent.com/blog/fact-checking-ai-content/",
          "position": 7
        },
        {
          "title": "Fact Check AI Content: 7 Steps to Catch False Claims",
          "host": "aiflowreview",
          "url": "https://aiflowreview.com/fact-check-ai-content/",
          "position": 8
        },
        {
          "title": "How to Fact-Check AI Content Before Publishing",
          "host": "searchfit",
          "url": "https://searchfit.ai/blog/fact-check-ai-content-before-publish",
          "position": 9
        },
        {
          "title": "AI-Generated Content for SEO: What to Verify Before Publishing",
          "host": "seoinformatica",
          "url": "https://seoinformatica.com/ai-generated-content-seo/",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-inaccurate-ai-citations-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-inaccurate-ai-citations-repeat-2",
      "published_at": "2026-11-15T13:06:00.000Z",
      "updated_at": "2026-11-15T13:06:00.000Z",
      "revision": 1,
      "slug": "perplexity-inaccurate-ai-citations-repeat-2",
      "panelId": "inaccurate-ai-citations",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-inaccurate-ai-citations",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "What causes inaccurate AI citations?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 21:06 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/35240fdd-dc48-43c5-9f1e-1ebed1acd21b",
      "mentions": "Cited",
      "summary": "AI citations can be inaccurate because language models generate the most plausible next text, not necessarily references verified against authoritative databases. Noisy training data, missing or conflicting source details, weak fact-checking, and retrieval mistakes can lead to fabricated citations or mismatches between a claim and its source.",
      "answerSnapshot": "7秒\n\nAI citations can be inaccurate because language models generate the most plausible next text, not necessarily references verified against authoritative databases. Noisy training data, missing or conflicting source details, weak fact-checking, and retrieval mistakes can lead to fabricated citations or mismatches between a claim and its source. Even a citation that looks complete and correctly formatted may still be wrong.\nacm\n+1",
      "citations": [
        {
          "title": "A Review of Techniques for Detecting and Alleviating Academic …",
          "host": "dl.acm",
          "url": "https://dl.acm.org/doi/full/10.1145/3838323.3838361",
          "position": 1
        },
        {
          "title": "AI Tools for Citations: Why GPT Versions Fail",
          "host": "linkedin",
          "url": "https://www.linkedin.com/posts/deepa-jaganathan-1a8a6238_how-all-my-citations-went-wrong-this-activity-7403048235524726784-DBFJ",
          "position": 2
        },
        {
          "title": "Fabrication and errors in the bibliographic citations ...",
          "host": "nature",
          "url": "https://www.nature.com/articles/s41598-023-41032-5?error=cookies_not_supported&code=dd348cd4-3155-4068-8d18-d8dcaff8c673",
          "position": 3
        },
        {
          "title": "Fabrication and errors in the bibliographic citations generated by ...",
          "host": "nature",
          "url": "https://www.nature.com/articles/s41598-023-41032-5?error=cookies_not_supported&code=48aca44f-4766-45f8-8217-a5e9011ba669",
          "position": 4
        },
        {
          "title": "Significance and implications of accurate and proper ...",
          "host": "pmc.ncbi.nlm.nih",
          "url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC8712974/",
          "position": 5
        },
        {
          "title": "ChatGPT Provided Non-Existent Citations and… – AI Incident 464",
          "host": "airisk.mit",
          "url": "https://airisk.mit.edu/ai-incidents/464-chatgpt-provided-non-existent-citations-and-links-when",
          "position": 6
        },
        {
          "title": "lseo.com · answer-engine-optimization-services · howWrong AI Citations: How to Fix Them at the Source",
          "host": "lseo",
          "url": "https://lseo.com/answer-engine-optimization-services/how-to-correct-wrong-citations-without-chasing-every-prompt/",
          "position": 7
        },
        {
          "title": "Why AI Makes Up Sources and Fake Citations",
          "host": "aitoolsusageguide",
          "url": "https://aitoolsusageguide.org/why-ai-makes-up-sources/",
          "position": 8
        },
        {
          "title": "AI Search's Citation Problem: Accuracy, Tracking and Fixes",
          "host": "itallstartedwithaidea",
          "url": "https://itallstartedwithaidea.com/truths-in-ai-search-has-a-citation-problem/",
          "position": 9
        },
        {
          "title": "Warum KI-Tools akademische Referenzen halluzinieren",
          "host": "citely",
          "url": "https://citely.ai/de/posts/why-ai-tools-hallucinate-academic-references",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-inaccurate-ai-citations-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-inaccurate-ai-citations-repeat-3",
      "published_at": "2026-11-15T13:06:00.000Z",
      "updated_at": "2026-11-15T13:06:00.000Z",
      "revision": 1,
      "slug": "perplexity-inaccurate-ai-citations-repeat-3",
      "panelId": "inaccurate-ai-citations",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-inaccurate-ai-citations",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "What causes inaccurate AI citations?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 21:06 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/48518d3e-2736-4ddc-af04-14078aafec06",
      "mentions": "Cited",
      "summary": "AI citations go wrong mainly because language models generate text probabilistically: they can produce plausible-looking sources without verifying that the work exists or that its details are correct.",
      "answerSnapshot": "10秒\n\nAI citations go wrong mainly because language models generate text probabilistically: they can produce plausible-looking sources without verifying that the work exists or that its details are correct. Errors are more likely when training data is noisy, source information is incomplete, retrieval finds the wrong material, or the system lacks a fact-checking step.\nacm\n+1",
      "citations": [
        {
          "title": "A Review of Techniques for Detecting and Alleviating Academic …",
          "host": "dl.acm",
          "url": "https://dl.acm.org/doi/full/10.1145/3838323.3838361",
          "position": 1
        },
        {
          "title": "Fabrication and errors in the bibliographic citations ...",
          "host": "nature",
          "url": "https://www.nature.com/articles/s41598-023-41032-5?error=cookies_not_supported&code=dd348cd4-3155-4068-8d18-d8dcaff8c673",
          "position": 2
        },
        {
          "title": "Fabrication and errors in the bibliographic citations generated by ...",
          "host": "nature",
          "url": "https://www.nature.com/articles/s41598-023-41032-5?error=cookies_not_supported&code=48aca44f-4766-45f8-8217-a5e9011ba669",
          "position": 3
        },
        {
          "title": "Significance and implications of accurate and proper ...",
          "host": "pmc.ncbi.nlm.nih",
          "url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC8712974/",
          "position": 4
        },
        {
          "title": "ChatGPT Provided Non-Existent Citations and… – AI Incident 464",
          "host": "airisk.mit",
          "url": "https://airisk.mit.edu/ai-incidents/464-chatgpt-provided-non-existent-citations-and-links-when",
          "position": 5
        },
        {
          "title": "lseo.com · answer-engine-optimization-services · howWrong AI Citations: How to Fix Them at the Source",
          "host": "lseo",
          "url": "https://lseo.com/answer-engine-optimization-services/how-to-correct-wrong-citations-without-chasing-every-prompt/",
          "position": 6
        },
        {
          "title": "Why AI Makes Up Sources and Fake Citations",
          "host": "aitoolsusageguide",
          "url": "https://aitoolsusageguide.org/why-ai-makes-up-sources/",
          "position": 7
        },
        {
          "title": "AI Tools for Citations: Why GPT Versions Fail",
          "host": "linkedin",
          "url": "https://www.linkedin.com/posts/deepa-jaganathan-1a8a6238_how-all-my-citations-went-wrong-this-activity-7403048235524726784-DBFJ",
          "position": 8
        },
        {
          "title": "AI Search's Citation Problem: Accuracy, Tracking and Fixes",
          "host": "itallstartedwithaidea",
          "url": "https://itallstartedwithaidea.com/truths-in-ai-search-has-a-citation-problem/",
          "position": 9
        },
        {
          "title": "Warum KI-Tools akademische Referenzen halluzinieren",
          "host": "citely",
          "url": "https://citely.ai/de/posts/why-ai-tools-hallucinate-academic-references",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-inaccurate-ai-citations",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-inaccurate-ai-citations",
      "published_at": "2026-11-15T13:05:00.000Z",
      "updated_at": "2026-11-15T13:05:00.000Z",
      "revision": 1,
      "slug": "perplexity-inaccurate-ai-citations",
      "panelId": "inaccurate-ai-citations",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-inaccurate-ai-citations",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "What causes inaccurate AI citations?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 21:05 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/09176708-18f2-4f86-b481-d6c20b617262",
      "mentions": "Cited",
      "summary": "AI citations are often inaccurate because language models generate the most plausible continuation, not necessarily a verified source. No fact-checking or citation-database check means noisy or incomplete training data, missing source details, retrieval mistakes, and prompt or domain limitations can lead to invented references or mismatched links.",
      "answerSnapshot": "9秒\n\nAI citations are often inaccurate because language models generate the most plausible continuation, not necessarily a verified source. No fact-checking or citation-database check means noisy or incomplete training data, missing source details, retrieval mistakes, and prompt or domain limitations can lead to invented references or mismatched links.\nacm\n+1",
      "citations": [
        {
          "title": "A Review of Techniques for Detecting and Alleviating Academic …",
          "host": "dl.acm",
          "url": "https://dl.acm.org/doi/full/10.1145/3838323.3838361",
          "position": 1
        },
        {
          "title": "Fabrication and errors in the bibliographic citations ...",
          "host": "nature",
          "url": "https://www.nature.com/articles/s41598-023-41032-5?error=cookies_not_supported&code=dd348cd4-3155-4068-8d18-d8dcaff8c673",
          "position": 2
        },
        {
          "title": "Fabrication and errors in the bibliographic citations generated by ...",
          "host": "nature",
          "url": "https://www.nature.com/articles/s41598-023-41032-5?error=cookies_not_supported&code=48aca44f-4766-45f8-8217-a5e9011ba669",
          "position": 3
        },
        {
          "title": "Significance and implications of accurate and proper ...",
          "host": "pmc.ncbi.nlm.nih",
          "url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC8712974/",
          "position": 4
        },
        {
          "title": "ChatGPT Provided Non-Existent Citations and… – AI Incident 464",
          "host": "airisk.mit",
          "url": "https://airisk.mit.edu/ai-incidents/464-chatgpt-provided-non-existent-citations-and-links-when",
          "position": 5
        },
        {
          "title": "lseo.com · answer-engine-optimization-services · howWrong AI Citations: How to Fix Them at the Source",
          "host": "lseo",
          "url": "https://lseo.com/answer-engine-optimization-services/how-to-correct-wrong-citations-without-chasing-every-prompt/",
          "position": 6
        },
        {
          "title": "Why AI Makes Up Sources and Fake Citations",
          "host": "aitoolsusageguide",
          "url": "https://aitoolsusageguide.org/why-ai-makes-up-sources/",
          "position": 7
        },
        {
          "title": "AI Tools for Citations: Why GPT Versions Fail",
          "host": "linkedin",
          "url": "https://www.linkedin.com/posts/deepa-jaganathan-1a8a6238_how-all-my-citations-went-wrong-this-activity-7403048235524726784-DBFJ",
          "position": 8
        },
        {
          "title": "AI Search's Citation Problem: Accuracy, Tracking and Fixes",
          "host": "itallstartedwithaidea",
          "url": "https://itallstartedwithaidea.com/truths-in-ai-search-has-a-citation-problem/",
          "position": 9
        },
        {
          "title": "Warum KI-Tools akademische Referenzen halluzinieren",
          "host": "citely",
          "url": "https://citely.ai/de/posts/why-ai-tools-hallucinate-academic-references",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-source-change-answer-drift",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-source-change-answer-drift",
      "published_at": "2026-11-15T13:05:00.000Z",
      "updated_at": "2026-11-15T13:05:00.000Z",
      "revision": 1,
      "slug": "perplexity-source-change-answer-drift",
      "panelId": "source-change-answer-drift",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-source-change-answer-drift",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "How do source changes affect future AI citations?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 21:05 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/24fa34da-4484-40b3-bdbd-4e004479d82f",
      "mentions": "Cited",
      "summary": "Changes to a source page can affect whether AI systems retrieve and cite it later: clearer facts, structure, and corrected contradictions may make it easier to use, while a drop in relevance or freshness may cause citations to shift.",
      "answerSnapshot": "9秒\n\nChanges to a source page can affect whether AI systems retrieve and cite it later: clearer facts, structure, and corrected contradictions may make it easier to use, while a drop in relevance or freshness may cause citations to shift. But citations can also change because of model or search-system updates, query wording, and competing sources, so updating a page does not guarantee it will be cited.\ninboundpros\n+2",
      "citations": [
        {
          "title": "Do updates change AI citations?",
          "host": "inboundpros",
          "url": "https://inboundpros.io/blog/how-citations-change-when-source-page-keeps-updating/",
          "position": 1
        },
        {
          "title": "AI Citation Volatility: How Much Do AI Answers Change?",
          "host": "addlly",
          "url": "https://addlly.ai/blog/ai-citation-volatility/",
          "position": 2
        },
        {
          "title": "AI Citation Volatility, Persistence and Durability Explained",
          "host": "answerenginee",
          "url": "https://answerenginee.com/ai-citation-volatility/",
          "position": 3
        },
        {
          "title": "Why AI Search Sources Change: Track Brand Visibility",
          "host": "wisoftsolutions",
          "url": "https://www.wisoftsolutions.com/blog/ai-search-citations-brand-visibility-tracking",
          "position": 4
        },
        {
          "title": "The Future of Citations: How AI is Changing Academic ...",
          "host": "sourcely",
          "url": "https://www.sourcely.net/post/the-future-of-citations-how-ai-is-changing-academic-referencing",
          "position": 5
        },
        {
          "title": "How stable are AI citations? What the research shows, and how to measure yours",
          "host": "katama",
          "url": "https://katama.io/blog/how-stable-are-ai-citations/",
          "position": 6
        },
        {
          "title": "How AI Engines Choose Sources to Cite: 2026 Guide",
          "host": "searchcounselco",
          "url": "https://searchcounselco.com/how-ai-engines-choose-sources/",
          "position": 7
        },
        {
          "title": "AI Citation Decay: Why Your Brand Can Disappear From AI Search Answers",
          "host": "besourceable",
          "url": "https://besourceable.com/blog/ai-citation-decay-why-your-brand-can-disappear-from-ai-search-answers",
          "position": 8
        },
        {
          "title": "AI citation tracking: Everything you need to know | Asky",
          "host": "askylabs",
          "url": "https://askylabs.com/learn/ai-visibility-tracking/ai-citation-tracking-everything-you-need-to-know",
          "position": 9
        },
        {
          "title": "Analyzing ChatGPT’s Source Citation and Search Result Matching - Adsroid",
          "host": "adsroid",
          "url": "https://adsroid.com/analyzing-chatgpts-source-citation-search-result-matching/",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-source-change-answer-drift-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-source-change-answer-drift-repeat-2",
      "published_at": "2026-11-15T13:05:00.000Z",
      "updated_at": "2026-11-15T13:05:00.000Z",
      "revision": 1,
      "slug": "perplexity-source-change-answer-drift-repeat-2",
      "panelId": "source-change-answer-drift",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-source-change-answer-drift",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "How do source changes affect future AI citations?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 21:05 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/a4981997-b3de-41f2-8048-434e3b5e1fef",
      "mentions": "Cited",
      "summary": "Changes to a source—especially clearer definitions, structured data, updated facts, or removed contradictions—can make it easier for AI systems to extract and cite, but minor edits may have little effect.",
      "answerSnapshot": "11秒\n\nChanges to a source—especially clearer definitions, structured data, updated facts, or removed contradictions—can make it easier for AI systems to extract and cite, but minor edits may have little effect. Whether the page is cited later also depends on query wording, competing sources, freshness signals, and changes to the AI system, so citations can shift even when your page stays the same.\ninboundpros\n+3",
      "citations": [
        {
          "title": "Do updates change AI citations?",
          "host": "inboundpros",
          "url": "https://inboundpros.io/blog/how-citations-change-when-source-page-keeps-updating/",
          "position": 1
        },
        {
          "title": "How stable are AI citations? What the research shows, and how to measure yours",
          "host": "katama",
          "url": "https://katama.io/blog/how-stable-are-ai-citations/",
          "position": 2
        },
        {
          "title": "AI Citation Volatility: How Much Do AI Answers Change?",
          "host": "addlly",
          "url": "https://addlly.ai/blog/ai-citation-volatility/",
          "position": 3
        },
        {
          "title": "AI Citation Volatility, Persistence and Durability Explained",
          "host": "answerenginee",
          "url": "https://answerenginee.com/ai-citation-volatility/",
          "position": 4
        },
        {
          "title": "Why AI Search Sources Change: Track Brand Visibility",
          "host": "wisoftsolutions",
          "url": "https://www.wisoftsolutions.com/blog/ai-search-citations-brand-visibility-tracking",
          "position": 5
        },
        {
          "title": "The Future of Citations: How AI is Changing Academic ...",
          "host": "sourcely",
          "url": "https://www.sourcely.net/post/the-future-of-citations-how-ai-is-changing-academic-referencing",
          "position": 6
        },
        {
          "title": "How AI Engines Choose Sources to Cite: 2026 Guide",
          "host": "searchcounselco",
          "url": "https://searchcounselco.com/how-ai-engines-choose-sources/",
          "position": 7
        },
        {
          "title": "AI Citation Decay: Why Your Brand Can Disappear From AI Search Answers",
          "host": "besourceable",
          "url": "https://besourceable.com/blog/ai-citation-decay-why-your-brand-can-disappear-from-ai-search-answers",
          "position": 8
        },
        {
          "title": "AI citation tracking: Everything you need to know | Asky",
          "host": "askylabs",
          "url": "https://askylabs.com/learn/ai-visibility-tracking/ai-citation-tracking-everything-you-need-to-know",
          "position": 9
        },
        {
          "title": "Analyzing ChatGPT’s Source Citation and Search Result Matching - Adsroid",
          "host": "adsroid",
          "url": "https://adsroid.com/analyzing-chatgpts-source-citation-search-result-matching/",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-source-change-answer-drift-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-source-change-answer-drift-repeat-3",
      "published_at": "2026-11-15T13:05:00.000Z",
      "updated_at": "2026-11-15T13:05:00.000Z",
      "revision": 1,
      "slug": "perplexity-source-change-answer-drift-repeat-3",
      "panelId": "source-change-answer-drift",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-source-change-answer-drift",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "How do source changes affect future AI citations?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 21:05 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/316a3095-22fe-489f-ad7f-37397a3265d5",
      "mentions": "Cited",
      "summary": "When a source page changes, AI systems may cite it more often, less often, or stop citing it—especially if the update improves or weakens clarity, factual precision, structure, or consistency. Small cosmetic edits may have little effect.",
      "answerSnapshot": "11秒\n\nWhen a source page changes, AI systems may cite it more often, less often, or stop citing it—especially if the update improves or weakens clarity, factual precision, structure, or consistency. Small cosmetic edits may have little effect.\ninboundpros\n\nChanges aren’t guaranteed to persist: citations also vary with query wording, freshness, and AI platform updates, so a citation shift may reflect the system rather than your page.\nkatama\n+2",
      "citations": [
        {
          "title": "Do updates change AI citations?",
          "host": "inboundpros",
          "url": "https://inboundpros.io/blog/how-citations-change-when-source-page-keeps-updating/",
          "position": 1
        },
        {
          "title": "How stable are AI citations? What the research shows, and how to measure yours",
          "host": "katama",
          "url": "https://katama.io/blog/how-stable-are-ai-citations/",
          "position": 2
        },
        {
          "title": "AI Citation Volatility: How Much Do AI Answers Change?",
          "host": "addlly",
          "url": "https://addlly.ai/blog/ai-citation-volatility/",
          "position": 3
        },
        {
          "title": "AI Citation Volatility, Persistence and Durability Explained",
          "host": "answerenginee",
          "url": "https://answerenginee.com/ai-citation-volatility/",
          "position": 4
        },
        {
          "title": "Why AI Search Sources Change: Track Brand Visibility",
          "host": "wisoftsolutions",
          "url": "https://www.wisoftsolutions.com/blog/ai-search-citations-brand-visibility-tracking",
          "position": 5
        },
        {
          "title": "The Future of Citations: How AI is Changing Academic ...",
          "host": "sourcely",
          "url": "https://www.sourcely.net/post/the-future-of-citations-how-ai-is-changing-academic-referencing",
          "position": 6
        },
        {
          "title": "How AI Engines Choose Sources to Cite: 2026 Guide",
          "host": "searchcounselco",
          "url": "https://searchcounselco.com/how-ai-engines-choose-sources/",
          "position": 7
        },
        {
          "title": "AI Citation Decay: Why Your Brand Can Disappear From AI Search Answers",
          "host": "besourceable",
          "url": "https://besourceable.com/blog/ai-citation-decay-why-your-brand-can-disappear-from-ai-search-answers",
          "position": 8
        },
        {
          "title": "AI citation tracking: Everything you need to know | Asky",
          "host": "askylabs",
          "url": "https://askylabs.com/learn/ai-visibility-tracking/ai-citation-tracking-everything-you-need-to-know",
          "position": 9
        },
        {
          "title": "Analyzing ChatGPT’s Source Citation and Search Result Matching - Adsroid",
          "host": "adsroid",
          "url": "https://adsroid.com/analyzing-chatgpts-source-citation-search-result-matching/",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-track-citations-over-time",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-track-citations-over-time",
      "published_at": "2026-11-15T13:04:00.000Z",
      "updated_at": "2026-11-15T13:04:00.000Z",
      "revision": 1,
      "slug": "perplexity-track-citations-over-time",
      "panelId": "track-citations-over-time",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-track-citations-over-time",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "How should a team track AI citations over time?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 21:04 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/02a61b44-8d48-4a7d-95ae-25614b582b18",
      "mentions": "Cited",
      "summary": "Use a fixed set of priority prompts across the AI platforms that matter to your audience, and rerun them on a consistent schedule with the same wording, settings, and location. Because answers vary, repeat each prompt several times before treating a change as meaningful, and compare trends rather than single snapshots.",
      "answerSnapshot": "11秒\n\nUse a fixed set of priority prompts across the AI platforms that matter to your audience, and rerun them on a consistent schedule with the same wording, settings, and location. Because answers vary, repeat each prompt several times before treating a change as meaningful, and compare trends rather than single snapshots.\nuseomnia\n\nTrack citation rate and share of voice by platform and topic, record the exact cited URLs and domains, and include each metric’s formula and answer count in reports. Review priority queries weekly, summarize broader trends monthly, and investigate sustained changes rather than reacting to one-off dips.\nslatehq\n+2",
      "citations": [
        {
          "title": "AI Citation Tracking Guide: What it is, How to Measure it, and How to Turn it into Action",
          "host": "useomnia",
          "url": "https://www.useomnia.com/blog/ai-citation-tracking",
          "position": 1
        },
        {
          "title": "10 Best AI Citation Tracking Tools For Content Leaders In 2026 - SlateHQ",
          "host": "slatehq",
          "url": "https://slatehq.com/blog/ai-citation-tracking-tools",
          "position": 2
        },
        {
          "title": "AI Citation Tracking: How to Track AI Citations (2026)",
          "host": "atomicagi",
          "url": "https://www.atomicagi.com/blog/ai-citation-tracking-the-complete-guide-for-seo-and-marketing-teams",
          "position": 3
        },
        {
          "title": "Best AI Citation Tracking Tools for 2026",
          "host": "pixis",
          "url": "https://pixis.ai/blog/best-ai-citation-tracking-tools-for-2026/",
          "position": 4
        },
        {
          "title": "AI Citation Tracking Tools for Brands: The Complete Guide to Measuring ...",
          "host": "clearscope",
          "url": "https://www.clearscope.io/blog/ai-citation-tracking-tools-for-brands",
          "position": 5
        },
        {
          "title": "AI citation tracking: How to track (and grow) AI engine citations",
          "host": "blog.hubspot",
          "url": "https://blog.hubspot.com/marketing/ai-citation-tracking",
          "position": 6
        },
        {
          "title": "AI Citation Tracking: How to Track & Beat the Sources AI Cites",
          "host": "mentionlytics",
          "url": "https://www.mentionlytics.com/blog/ai-citation-tracking/",
          "position": 7
        },
        {
          "title": "AI citation tracking tools to monitor and increase visibility",
          "host": "blog.hubspot",
          "url": "https://blog.hubspot.com/marketing/ai-citation-tracking-tools",
          "position": 8
        },
        {
          "title": "How to Track AI Citations: A Weekly Framework - AI-Advisors",
          "host": "ai-advisors",
          "url": "https://www.ai-advisors.ai/blog/how-to-track-ai-citations",
          "position": 9
        },
        {
          "title": "AI Citation Tracking: What It Is, Why It Matters, and Which Tools to Use",
          "host": "foundationinc",
          "url": "https://foundationinc.co/lab/ai-citation-tracking",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-track-citations-over-time-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-track-citations-over-time-repeat-2",
      "published_at": "2026-11-15T13:04:00.000Z",
      "updated_at": "2026-11-15T13:04:00.000Z",
      "revision": 1,
      "slug": "perplexity-track-citations-over-time-repeat-2",
      "panelId": "track-citations-over-time",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-track-citations-over-time",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "How should a team track AI citations over time?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 21:04 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/5abdd061-33cd-428c-b078-4ba324aed97c",
      "mentions": "Cited",
      "summary": "Use a fixed set of prompts across the same AI engines and settings, and repeat each prompt several times so normal answer variation doesn’t skew the result. Log the exact cited URLs, then track citation rate and share of voice by engine and topic, with the formula and answer count alongside each metric.",
      "answerSnapshot": "11秒\n\nUse a fixed set of prompts across the same AI engines and settings, and repeat each prompt several times so normal answer variation doesn’t skew the result. Log the exact cited URLs, then track citation rate and share of voice by engine and topic, with the formula and answer count alongside each metric.\natomicagi\n+2\n\nReview priority queries weekly and broader trends monthly; use a rolling baseline, investigate sustained shifts rather than reacting to a single dip, and revisit the prompt set quarterly.\nslatehq\n+2",
      "citations": [
        {
          "title": "AI Citation Tracking: How to Track AI Citations (2026)",
          "host": "atomicagi",
          "url": "https://www.atomicagi.com/blog/ai-citation-tracking-the-complete-guide-for-seo-and-marketing-teams",
          "position": 1
        },
        {
          "title": "Best AI Citation Tracking Tools for 2026",
          "host": "pixis",
          "url": "https://pixis.ai/blog/best-ai-citation-tracking-tools-for-2026/",
          "position": 2
        },
        {
          "title": "AI Citation Tracking Guide: What it is, How to Measure it, and How to Turn it into Action",
          "host": "useomnia",
          "url": "https://www.useomnia.com/blog/ai-citation-tracking",
          "position": 3
        },
        {
          "title": "10 Best AI Citation Tracking Tools For Content Leaders In 2026 - SlateHQ",
          "host": "slatehq",
          "url": "https://slatehq.com/blog/ai-citation-tracking-tools",
          "position": 4
        },
        {
          "title": "AI Citation Tracking: How to Track & Beat the Sources AI Cites",
          "host": "mentionlytics",
          "url": "https://www.mentionlytics.com/blog/ai-citation-tracking/",
          "position": 5
        },
        {
          "title": "How to Track AI Citations: A Weekly Framework - AI-Advisors",
          "host": "ai-advisors",
          "url": "https://www.ai-advisors.ai/blog/how-to-track-ai-citations",
          "position": 6
        },
        {
          "title": "AI Citation Tracking Tools for Brands: The Complete Guide to Measuring ...",
          "host": "clearscope",
          "url": "https://www.clearscope.io/blog/ai-citation-tracking-tools-for-brands",
          "position": 7
        },
        {
          "title": "AI citation tracking: How to track (and grow) AI engine citations",
          "host": "blog.hubspot",
          "url": "https://blog.hubspot.com/marketing/ai-citation-tracking",
          "position": 8
        },
        {
          "title": "AI citation tracking tools to monitor and increase visibility",
          "host": "blog.hubspot",
          "url": "https://blog.hubspot.com/marketing/ai-citation-tracking-tools",
          "position": 9
        },
        {
          "title": "AI Citation Tracking: What It Is, Why It Matters, and Which Tools to Use",
          "host": "foundationinc",
          "url": "https://foundationinc.co/lab/ai-citation-tracking",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-track-citations-over-time-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-track-citations-over-time-repeat-3",
      "published_at": "2026-11-15T13:04:00.000Z",
      "updated_at": "2026-11-15T13:04:00.000Z",
      "revision": 1,
      "slug": "perplexity-track-citations-over-time-repeat-3",
      "panelId": "track-citations-over-time",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-track-citations-over-time",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "How should a team track AI citations over time?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 21:04 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/401fd8a4-8305-4f94-ab08-f2ac059411aa",
      "mentions": "Cited",
      "summary": "Track a fixed set of priority prompts across the same AI engines and settings on a regular schedule. Repeat each prompt several times to smooth out answer variability, and log citation URLs, domains, brand mentions, and competitors.",
      "answerSnapshot": "14秒\n\nTrack a fixed set of priority prompts across the same AI engines and settings on a regular schedule. Repeat each prompt several times to smooth out answer variability, and log citation URLs, domains, brand mentions, and competitors.\npixis\n+1\n\nReport citation rate and share of voice by engine and topic, with the formula and number of answers sampled; review priority queries weekly, broader trends monthly, and refresh the prompt set quarterly. Use a multiweek baseline or rolling average, and investigate sustained shifts rather than reacting to one-off changes.\nslatehq\n+2",
      "citations": [
        {
          "title": "Best AI Citation Tracking Tools for 2026",
          "host": "pixis",
          "url": "https://pixis.ai/blog/best-ai-citation-tracking-tools-for-2026/",
          "position": 1
        },
        {
          "title": "AI Citation Tracking Guide: What it is, How to Measure it, and How to Turn it into Action",
          "host": "useomnia",
          "url": "https://www.useomnia.com/blog/ai-citation-tracking",
          "position": 2
        },
        {
          "title": "10 Best AI Citation Tracking Tools For Content Leaders In 2026 - SlateHQ",
          "host": "slatehq",
          "url": "https://slatehq.com/blog/ai-citation-tracking-tools",
          "position": 3
        },
        {
          "title": "AI Citation Tracking: How to Track AI Citations (2026)",
          "host": "atomicagi",
          "url": "https://www.atomicagi.com/blog/ai-citation-tracking-the-complete-guide-for-seo-and-marketing-teams",
          "position": 4
        },
        {
          "title": "How to Track AI Citations: A Weekly Framework - AI-Advisors",
          "host": "ai-advisors",
          "url": "https://www.ai-advisors.ai/blog/how-to-track-ai-citations",
          "position": 5
        },
        {
          "title": "AI Citation Tracking Tools for Brands: The Complete Guide to Measuring ...",
          "host": "clearscope",
          "url": "https://www.clearscope.io/blog/ai-citation-tracking-tools-for-brands",
          "position": 6
        },
        {
          "title": "AI citation tracking: How to track (and grow) AI engine citations",
          "host": "blog.hubspot",
          "url": "https://blog.hubspot.com/marketing/ai-citation-tracking",
          "position": 7
        },
        {
          "title": "AI Citation Tracking: How to Track & Beat the Sources AI Cites",
          "host": "mentionlytics",
          "url": "https://www.mentionlytics.com/blog/ai-citation-tracking/",
          "position": 8
        },
        {
          "title": "AI citation tracking tools to monitor and increase visibility",
          "host": "blog.hubspot",
          "url": "https://blog.hubspot.com/marketing/ai-citation-tracking-tools",
          "position": 9
        },
        {
          "title": "AI Citation Tracking: What It Is, Why It Matters, and Which Tools to Use",
          "host": "foundationinc",
          "url": "https://foundationinc.co/lab/ai-citation-tracking",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-comparison-page-citations-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-comparison-page-citations-repeat-3",
      "published_at": "2026-11-15T13:03:00.000Z",
      "updated_at": "2026-11-15T13:03:00.000Z",
      "revision": 1,
      "slug": "perplexity-comparison-page-citations-repeat-3",
      "panelId": "comparison-page-citations",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-comparison-page-citations",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "How do AI search engines choose comparison pages as sources?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 21:03 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/8b1c81f7-c17d-4dac-a3ce-8d8d8f021e29",
      "mentions": "Cited",
      "summary": "AI search engines generally retrieve comparison pages when their content directly answers the query, then cite useful passages that support the generated response. Selection varies by engine and query; there’s no known universal formula.",
      "answerSnapshot": "6秒\n\nAI search engines generally retrieve comparison pages when their content directly answers the query, then cite useful passages that support the generated response. Selection varies by engine and query; there’s no known universal formula.\nleadtoconversion\n+1\n\nComparison pages are more useful when they clearly state who each option suits, provide specific and verifiable differences, and explain trade-offs; honest, relevant matchups are preferable to generic “best” claims.\ngetdigitize",
      "citations": [
        {
          "title": "How Do AI Search Engines Choose Sources? Citations ...",
          "host": "leadtoconversion",
          "url": "https://leadtoconversion.com/blog/how-do-ai-search-engines-choose-sources/",
          "position": 1
        },
        {
          "title": "How Does AI Search Work? AI Citation Mechanics Explained",
          "host": "straightnorth",
          "url": "https://www.straightnorth.com/blog/how-do-ai-search-engines-choose-which-sources-to-cite/",
          "position": 2
        },
        {
          "title": "Comparison Pages for AI Search: Win Best X for Y Answers",
          "host": "getdigitize",
          "url": "https://getdigitize.com/blog/comparison-pages-for-ai-search/",
          "position": 3
        },
        {
          "title": "Comparison Pages in AI Search: What 1.8 Million Answers Show",
          "host": "searchable",
          "url": "https://www.searchable.com/es/blog/comparison-pages-ai-search",
          "position": 4
        },
        {
          "title": "How AI Engines Choose Sources to Cite: 2026 Guide",
          "host": "searchcounselco",
          "url": "https://searchcounselco.com/how-ai-engines-choose-sources/",
          "position": 5
        },
        {
          "title": "How AI Search Engines Choose Sources",
          "host": "postking",
          "url": "https://postking.app/blog/how-ai-search-engines-choose-sources",
          "position": 6
        },
        {
          "title": "How AI search engines choose sources",
          "host": "extendsclass",
          "url": "https://extendsclass.com/blog/how-ai-search-engines-choose-sources",
          "position": 7
        },
        {
          "title": "Under the Hood: The AI Search Workflow | AI Search Engine Optimization Guide | AI Chat Watch (AICW)",
          "host": "aichatwatch",
          "url": "https://aichatwatch.com/books/ai-seo-guide/03-under-the-hood/",
          "position": 8
        },
        {
          "title": "How AI Chooses Sources: ChatGPT, AI Overviews and More | Rankry",
          "host": "rankry",
          "url": "https://rankry.ai/guides/how-ai-chooses-sources/",
          "position": 9
        },
        {
          "title": "How AI decides which sources to cite | AI Visibility Engineering",
          "host": "searchtides",
          "url": "https://searchtides.com/ai-visibility-engineering/how-ai-citation-works/",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-product-page-evidence",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-product-page-evidence",
      "published_at": "2026-11-15T13:03:00.000Z",
      "updated_at": "2026-11-15T13:03:00.000Z",
      "revision": 1,
      "slug": "perplexity-product-page-evidence",
      "panelId": "product-page-evidence",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-product-page-evidence",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "What evidence makes a product page citable by AI search?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 21:03 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/b9cef103-c113-4b21-85f1-48c1ebdf0db8",
      "mentions": "Cited",
      "summary": "A product page is more citable when it gives clear, specific, verifiable facts in crawlable text: what the product is, who it suits, what it includes, its specifications, price, availability, and any limits or exclusions.",
      "answerSnapshot": "7秒\n\nA product page is more citable when it gives clear, specific, verifiable facts in crawlable text: what the product is, who it suits, what it includes, its specifications, price, availability, and any limits or exclusions.\nnugentive\n+2\n\nSupport those facts with trust evidence—such as genuine reviews, certifications, warranties, and clear shipping or return terms—and make the page easy to parse with descriptive headings and accurate product schema. Schema helps machines interpret the details, but does not guarantee a citation.\ncapston\n+2",
      "citations": [
        {
          "title": "Product Pages and AI Citations",
          "host": "nugentive",
          "url": "https://nugentive.com/blog/product-pages-ai-citations/",
          "position": 1
        },
        {
          "title": "How to Structure Product Pages for AI Citations: A Technical Guide for E-Commerce Brands",
          "host": "mentionowl",
          "url": "https://www.mentionowl.com/blog/how-to-structure-product-pages-for-ai-citations",
          "position": 2
        },
        {
          "title": "GEO for Product Pages: Can Commercial URLs Earn Citations in AI Answers?",
          "host": "geoseochecker",
          "url": "https://geoseochecker.com/blog/geo-for-product-pages-can-commercial-urls-earn-citations-in-ai-answers",
          "position": 3
        },
        {
          "title": "How AI Web Search Engines Evaluate Product Evidence",
          "host": "capston",
          "url": "https://capston.ai/how-ai-web-search-engines-evaluate-product-evidence/",
          "position": 4
        },
        {
          "title": "eCommerce SEO: expert guide",
          "host": "makdigitaldesign",
          "url": "https://makdigitaldesign.com/search-engine-optimization/get-products-cited-by-chatgpt/",
          "position": 5
        },
        {
          "title": "aysa.ai › ecommerce-ai-search-citations-product-pagesEcommerce AI Search Citations: Why Product Pages Are No…",
          "host": "aysa",
          "url": "https://aysa.ai/ecommerce-ai-search-citations-product-pages/",
          "position": 6
        },
        {
          "title": "What makes a page citable in AI search? | Articles | 2-UA.com",
          "host": "2-ua",
          "url": "https://2-ua.com/articles/what-makes-a-page-citable-in-ai-search",
          "position": 7
        },
        {
          "title": "Product Pages Win 24% Of AI Citations: Content Format Guide",
          "host": "seovendor",
          "url": "https://seovendor.co/product-pages-win-24-of-ai-citations-content-format-guide/",
          "position": 8
        },
        {
          "title": "How to Structure Product Content for AI Citations",
          "host": "windmillstrategy",
          "url": "https://www.windmillstrategy.com/optimize-content-for-ai-search-citations/",
          "position": 9
        },
        {
          "title": "AI Search Visibility for Product Sites: Structure and Schema",
          "host": "softwarecrafting",
          "url": "https://softwarecrafting.in/blog/ai-search-visibility-product-sites",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-product-page-evidence-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-product-page-evidence-repeat-2",
      "published_at": "2026-11-15T13:03:00.000Z",
      "updated_at": "2026-11-15T13:03:00.000Z",
      "revision": 1,
      "slug": "perplexity-product-page-evidence-repeat-2",
      "panelId": "product-page-evidence",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-product-page-evidence",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "What evidence makes a product page citable by AI search?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 21:03 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/bee25ced-713f-4eb1-bef0-bb273d183934",
      "mentions": "Cited",
      "summary": "A product page is more citable when it presents clear, specific, verifiable facts in crawlable text: what the product is, who it’s for, what’s included, its specifications, price, availability, and relevant limitations.",
      "answerSnapshot": "7秒\n\nA product page is more citable when it presents clear, specific, verifiable facts in crawlable text: what the product is, who it’s for, what’s included, its specifications, price, availability, and relevant limitations.\nnugentive\n+2\n\nBuild credibility with evidence such as genuine reviews, certifications, warranties, and clear shipping or return terms; use structured data that matches the visible page content to make those facts easier to interpret. None of these guarantees a citation, but together they make the page easier to retrieve, understand, and trust.\ncapston\n+2",
      "citations": [
        {
          "title": "Product Pages and AI Citations",
          "host": "nugentive",
          "url": "https://nugentive.com/blog/product-pages-ai-citations/",
          "position": 1
        },
        {
          "title": "How to Structure Product Pages for AI Citations: A Technical Guide for E-Commerce Brands",
          "host": "mentionowl",
          "url": "https://www.mentionowl.com/blog/how-to-structure-product-pages-for-ai-citations",
          "position": 2
        },
        {
          "title": "GEO for Product Pages: Can Commercial URLs Earn Citations in AI Answers?",
          "host": "geoseochecker",
          "url": "https://geoseochecker.com/blog/geo-for-product-pages-can-commercial-urls-earn-citations-in-ai-answers",
          "position": 3
        },
        {
          "title": "How AI Web Search Engines Evaluate Product Evidence",
          "host": "capston",
          "url": "https://capston.ai/how-ai-web-search-engines-evaluate-product-evidence/",
          "position": 4
        },
        {
          "title": "eCommerce SEO: expert guide",
          "host": "makdigitaldesign",
          "url": "https://makdigitaldesign.com/search-engine-optimization/get-products-cited-by-chatgpt/",
          "position": 5
        },
        {
          "title": "aysa.ai › ecommerce-ai-search-citations-product-pagesEcommerce AI Search Citations: Why Product Pages Are No…",
          "host": "aysa",
          "url": "https://aysa.ai/ecommerce-ai-search-citations-product-pages/",
          "position": 6
        },
        {
          "title": "What makes a page citable in AI search? | Articles | 2-UA.com",
          "host": "2-ua",
          "url": "https://2-ua.com/articles/what-makes-a-page-citable-in-ai-search",
          "position": 7
        },
        {
          "title": "Product Pages Win 24% Of AI Citations: Content Format Guide",
          "host": "seovendor",
          "url": "https://seovendor.co/product-pages-win-24-of-ai-citations-content-format-guide/",
          "position": 8
        },
        {
          "title": "How to Structure Product Content for AI Citations",
          "host": "windmillstrategy",
          "url": "https://www.windmillstrategy.com/optimize-content-for-ai-search-citations/",
          "position": 9
        },
        {
          "title": "AI Search Visibility for Product Sites: Structure and Schema",
          "host": "softwarecrafting",
          "url": "https://softwarecrafting.in/blog/ai-search-visibility-product-sites",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-product-page-evidence-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-product-page-evidence-repeat-3",
      "published_at": "2026-11-15T13:03:00.000Z",
      "updated_at": "2026-11-15T13:03:00.000Z",
      "revision": 1,
      "slug": "perplexity-product-page-evidence-repeat-3",
      "panelId": "product-page-evidence",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-product-page-evidence",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "What evidence makes a product page citable by AI search?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 21:03 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/9a87d11b-0196-4eba-a313-1003403b3e0c",
      "mentions": "Cited",
      "summary": "A product page is more citable when it presents clear, specific, verifiable facts in crawlable text: what the product is, who it’s for, what’s included, specifications, price, availability, and any limitations.",
      "answerSnapshot": "10秒\n\nA product page is more citable when it presents clear, specific, verifiable facts in crawlable text: what the product is, who it’s for, what’s included, specifications, price, availability, and any limitations.\nnugentive\n+2\n\nSupport those facts with visible reviews, certifications, warranty and shipping details; make the brand and product identity unambiguous; and use structured data that matches the page’s visible content. These signals make information easier to retrieve and assess, but none guarantees a citation.\ncapston\n+2",
      "citations": [
        {
          "title": "Product Pages and AI Citations",
          "host": "nugentive",
          "url": "https://nugentive.com/blog/product-pages-ai-citations/",
          "position": 1
        },
        {
          "title": "How to Structure Product Pages for AI Citations: A Technical Guide for E-Commerce Brands",
          "host": "mentionowl",
          "url": "https://www.mentionowl.com/blog/how-to-structure-product-pages-for-ai-citations",
          "position": 2
        },
        {
          "title": "GEO for Product Pages: Can Commercial URLs Earn Citations in AI Answers?",
          "host": "geoseochecker",
          "url": "https://geoseochecker.com/blog/geo-for-product-pages-can-commercial-urls-earn-citations-in-ai-answers",
          "position": 3
        },
        {
          "title": "How AI Web Search Engines Evaluate Product Evidence",
          "host": "capston",
          "url": "https://capston.ai/how-ai-web-search-engines-evaluate-product-evidence/",
          "position": 4
        },
        {
          "title": "eCommerce SEO: expert guide",
          "host": "makdigitaldesign",
          "url": "https://makdigitaldesign.com/search-engine-optimization/get-products-cited-by-chatgpt/",
          "position": 5
        },
        {
          "title": "AI Search Visibility for Product Sites: Structure and Schema",
          "host": "softwarecrafting",
          "url": "https://softwarecrafting.in/blog/ai-search-visibility-product-sites",
          "position": 6
        },
        {
          "title": "What makes a page citable in AI search? | Articles | 2-UA.com",
          "host": "2-ua",
          "url": "https://2-ua.com/articles/what-makes-a-page-citable-in-ai-search",
          "position": 7
        },
        {
          "title": "Product Pages Win 24% Of AI Citations: Content Format Guide",
          "host": "seovendor",
          "url": "https://seovendor.co/product-pages-win-24-of-ai-citations-content-format-guide/",
          "position": 8
        },
        {
          "title": "How to Structure Product Content for AI Citations",
          "host": "windmillstrategy",
          "url": "https://www.windmillstrategy.com/optimize-content-for-ai-search-citations/",
          "position": 9
        },
        {
          "title": "aysa.ai › ecommerce-ai-search-citations-product-pagesEcommerce AI Search Citations: Why Product Pages Are No…",
          "host": "aysa",
          "url": "https://aysa.ai/ecommerce-ai-search-citations-product-pages/",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-comparison-page-citations",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-comparison-page-citations",
      "published_at": "2026-11-15T13:02:00.000Z",
      "updated_at": "2026-11-15T13:02:00.000Z",
      "revision": 1,
      "slug": "perplexity-comparison-page-citations",
      "panelId": "comparison-page-citations",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-comparison-page-citations",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "How do AI search engines choose comparison pages as sources?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 21:02 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/aae707e4-74aa-40c2-9be2-0741454b633d",
      "mentions": "Cited",
      "summary": "AI search engines typically find comparison pages through their search or retrieval systems, then select passages that directly support the user’s question; the exact ranking and citation rules vary by platform and aren’t fully public.",
      "answerSnapshot": "3秒\n\nAI search engines typically find comparison pages through their search or retrieval systems, then select passages that directly support the user’s question; the exact ranking and citation rules vary by platform and aren’t fully public.\nleadtoconversion\n+1\n\nComparison pages are more likely to be useful when they give a clear, verifiable side-by-side comparison, explain trade-offs and “best for” use cases, and are accessible to the engine’s crawler.\nleadtoconversion\n+1",
      "citations": [
        {
          "title": "How Do AI Search Engines Choose Sources? Citations ...",
          "host": "leadtoconversion",
          "url": "https://leadtoconversion.com/blog/how-do-ai-search-engines-choose-sources/",
          "position": 1
        },
        {
          "title": "How Does AI Search Work? AI Citation Mechanics Explained",
          "host": "straightnorth",
          "url": "https://www.straightnorth.com/blog/how-do-ai-search-engines-choose-which-sources-to-cite/",
          "position": 2
        },
        {
          "title": "Comparison Pages for AI Search: Win Best X for Y Answers",
          "host": "getdigitize",
          "url": "https://getdigitize.com/blog/comparison-pages-for-ai-search/",
          "position": 3
        },
        {
          "title": "Comparison Pages in AI Search: What 1.8 Million Answers Show",
          "host": "searchable",
          "url": "https://www.searchable.com/es/blog/comparison-pages-ai-search",
          "position": 4
        },
        {
          "title": "How AI Engines Choose Sources to Cite: 2026 Guide",
          "host": "searchcounselco",
          "url": "https://searchcounselco.com/how-ai-engines-choose-sources/",
          "position": 5
        },
        {
          "title": "How AI Search Engines Choose Sources",
          "host": "postking",
          "url": "https://postking.app/blog/how-ai-search-engines-choose-sources",
          "position": 6
        },
        {
          "title": "How AI search engines choose sources",
          "host": "extendsclass",
          "url": "https://extendsclass.com/blog/how-ai-search-engines-choose-sources",
          "position": 7
        },
        {
          "title": "Under the Hood: The AI Search Workflow | AI Search Engine Optimization Guide | AI Chat Watch (AICW)",
          "host": "aichatwatch",
          "url": "https://aichatwatch.com/books/ai-seo-guide/03-under-the-hood/",
          "position": 8
        },
        {
          "title": "How AI Chooses Sources: ChatGPT, AI Overviews and More | Rankry",
          "host": "rankry",
          "url": "https://rankry.ai/guides/how-ai-chooses-sources/",
          "position": 9
        },
        {
          "title": "How AI decides which sources to cite | AI Visibility Engineering",
          "host": "searchtides",
          "url": "https://searchtides.com/ai-visibility-engineering/how-ai-citation-works/",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-comparison-page-citations-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-comparison-page-citations-repeat-2",
      "published_at": "2026-11-15T13:02:00.000Z",
      "updated_at": "2026-11-15T13:02:00.000Z",
      "revision": 1,
      "slug": "perplexity-comparison-page-citations-repeat-2",
      "panelId": "comparison-page-citations",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-comparison-page-citations",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "How do AI search engines choose comparison pages as sources?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 21:02 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/fd64a076-6ce5-4c08-a755-d5b49a2c0467",
      "mentions": "Cited",
      "summary": "AI search engines generally find comparison pages through search, sometimes by splitting a question into related searches, then cite passages that directly support the answer. There’s no universal published formula; selection varies by engine and query.",
      "answerSnapshot": "5秒\n\nAI search engines generally find comparison pages through search, sometimes by splitting a question into related searches, then cite passages that directly support the answer. There’s no universal published formula; selection varies by engine and query.\nsearchcounselco\n+1\n\nComparison pages are more useful as sources when they provide clear, verifiable differences—such as who each option suits, factual tables, and trade-offs—rather than vague or unsupported claims.\ngetdigitize",
      "citations": [
        {
          "title": "How AI Engines Choose Sources to Cite: 2026 Guide",
          "host": "searchcounselco",
          "url": "https://searchcounselco.com/how-ai-engines-choose-sources/",
          "position": 1
        },
        {
          "title": "How Does AI Search Work? AI Citation Mechanics Explained",
          "host": "straightnorth",
          "url": "https://www.straightnorth.com/blog/how-do-ai-search-engines-choose-which-sources-to-cite/",
          "position": 2
        },
        {
          "title": "Comparison Pages for AI Search: Win Best X for Y Answers",
          "host": "getdigitize",
          "url": "https://getdigitize.com/blog/comparison-pages-for-ai-search/",
          "position": 3
        },
        {
          "title": "Comparison Pages in AI Search: What 1.8 Million Answers Show",
          "host": "searchable",
          "url": "https://www.searchable.com/es/blog/comparison-pages-ai-search",
          "position": 4
        },
        {
          "title": "How AI Search Engines Choose Sources",
          "host": "postking",
          "url": "https://postking.app/blog/how-ai-search-engines-choose-sources",
          "position": 5
        },
        {
          "title": "How AI search engines choose sources",
          "host": "extendsclass",
          "url": "https://extendsclass.com/blog/how-ai-search-engines-choose-sources",
          "position": 6
        },
        {
          "title": "How Do AI Search Engines Choose Sources? Citations ...",
          "host": "leadtoconversion",
          "url": "https://leadtoconversion.com/blog/how-do-ai-search-engines-choose-sources/",
          "position": 7
        },
        {
          "title": "Under the Hood: The AI Search Workflow | AI Search Engine Optimization Guide | AI Chat Watch (AICW)",
          "host": "aichatwatch",
          "url": "https://aichatwatch.com/books/ai-seo-guide/03-under-the-hood/",
          "position": 8
        },
        {
          "title": "How AI Chooses Sources: ChatGPT, AI Overviews and More | Rankry",
          "host": "rankry",
          "url": "https://rankry.ai/guides/how-ai-chooses-sources/",
          "position": 9
        },
        {
          "title": "How AI decides which sources to cite | AI Visibility Engineering",
          "host": "searchtides",
          "url": "https://searchtides.com/ai-visibility-engineering/how-ai-citation-works/",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-local-business-ai-search-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-local-business-ai-search-repeat-2",
      "published_at": "2026-11-15T13:02:00.000Z",
      "updated_at": "2026-11-15T13:02:00.000Z",
      "revision": 1,
      "slug": "perplexity-local-business-ai-search-repeat-2",
      "panelId": "local-business-ai-search",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-local-business-ai-search",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "Which pages should a local business optimize for AI search?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 21:02 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/d1a94f9c-2d55-41ce-b688-c9070319a45b",
      "mentions": "Cited",
      "summary": "Prioritize dedicated pages for each main service and each location or service area, plus an FAQ page that answers real customer questions. Make each page clear about what you offer, whom you serve, where you operate, and how customers can contact or book you; don’t rely on the homepage alone.",
      "answerSnapshot": "11秒\n\nPrioritize dedicated pages for each main service and each location or service area, plus an FAQ page that answers real customer questions. Make each page clear about what you offer, whom you serve, where you operate, and how customers can contact or book you; don’t rely on the homepage alone.\nthryv\n+3",
      "citations": [
        {
          "title": "AI Search Optimization for Local Business",
          "host": "thryv",
          "url": "https://www.thryv.com/blog/ai-search-optimization-for-local-business/",
          "position": 1
        },
        {
          "title": "AI Search Optimization for Local Businesses in 2026",
          "host": "searchable",
          "url": "https://www.searchable.com/blog/local-business-ai-search-optimization-2026",
          "position": 2
        },
        {
          "title": "10 Ways to Show Up in AI Search - soci.ai",
          "host": "soci",
          "url": "https://www.soci.ai/blog/10-ways-to-show-up-in-ai-search/",
          "position": 3
        },
        {
          "title": "Practical, step-by-step optimization guide for local and small ...",
          "host": "uandai",
          "url": "https://uandai.co/blog/chatgpt-visibility-local-businesses-2026",
          "position": 4
        },
        {
          "title": "Local AI SEO: Optimize Your Site for AI Search",
          "host": "squarespace",
          "url": "https://www.squarespace.com/blog/local-ai-seo",
          "position": 5
        },
        {
          "title": "How to Optimize Your Local Business for AI Search and Discovery",
          "host": "keyword",
          "url": "https://keyword.com/blog/local-seo-for-ai-search/",
          "position": 6
        },
        {
          "title": "Home Services SEO Playbook: Get Found on Google, Maps, and AI Search",
          "host": "vendasta",
          "url": "https://www.vendasta.com/blog/home-services-seo-google-maps-ai-search/",
          "position": 7
        },
        {
          "title": "How to do AI search optimization for local businesses",
          "host": "semrush",
          "url": "https://www.semrush.com/blog/ai-search-optimization-for-local-businesses/",
          "position": 8
        },
        {
          "title": "Google's Ai Overviews, Ai...",
          "host": "aischemagen",
          "url": "https://www.aischemagen.com/blog/ai-search-optimization-for-local-business",
          "position": 9
        },
        {
          "title": "How to Get Your Local Business Recommended by ChatGPT, Gemini ...",
          "host": "seoinc",
          "url": "https://www.seoinc.com/seo-blog/get-your-local-business-recommended-by-ai/",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-local-business-ai-search-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-local-business-ai-search-repeat-3",
      "published_at": "2026-11-15T13:02:00.000Z",
      "updated_at": "2026-11-15T13:02:00.000Z",
      "revision": 1,
      "slug": "perplexity-local-business-ai-search-repeat-3",
      "panelId": "local-business-ai-search",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-local-business-ai-search",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "Which pages should a local business optimize for AI search?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 21:02 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/874819f9-74bb-47d9-b24f-f8b565702a8e",
      "mentions": "Cited",
      "summary": "Prioritize dedicated pages for each main service and service area, rather than relying on your homepage to explain everything. Make them clearly describe what you offer, who you serve, where you operate, what customers can expect, and how to book or contact you.",
      "answerSnapshot": "10秒\n\nPrioritize dedicated pages for each main service and service area, rather than relying on your homepage to explain everything. Make them clearly describe what you offer, who you serve, where you operate, what customers can expect, and how to book or contact you.\nthryv\n+2\n\nAlso optimize your homepage, FAQ page, and location pages; use natural-language headings and answer common questions directly, such as pricing, service options, and what to look for in a provider.\nsearchable\n+3",
      "citations": [
        {
          "title": "AI Search Optimization for Local Business",
          "host": "thryv",
          "url": "https://www.thryv.com/blog/ai-search-optimization-for-local-business/",
          "position": 1
        },
        {
          "title": "Home Services SEO Playbook: Get Found on Google, Maps, and AI Search",
          "host": "vendasta",
          "url": "https://www.vendasta.com/blog/home-services-seo-google-maps-ai-search/",
          "position": 2
        },
        {
          "title": "Practical, step-by-step optimization guide for local and small ...",
          "host": "uandai",
          "url": "https://uandai.co/blog/chatgpt-visibility-local-businesses-2026",
          "position": 3
        },
        {
          "title": "AI Search Optimization for Local Businesses in 2026",
          "host": "searchable",
          "url": "https://www.searchable.com/blog/local-business-ai-search-optimization-2026",
          "position": 4
        },
        {
          "title": "Local AI SEO: Optimize Your Site for AI Search",
          "host": "squarespace",
          "url": "https://www.squarespace.com/blog/local-ai-seo",
          "position": 5
        },
        {
          "title": "10 Ways to Show Up in AI Search - soci.ai",
          "host": "soci",
          "url": "https://www.soci.ai/blog/10-ways-to-show-up-in-ai-search/",
          "position": 6
        },
        {
          "title": "How to Optimize Your Local Business for AI Search and Discovery",
          "host": "keyword",
          "url": "https://keyword.com/blog/local-seo-for-ai-search/",
          "position": 7
        },
        {
          "title": "How to do AI search optimization for local businesses",
          "host": "semrush",
          "url": "https://www.semrush.com/blog/ai-search-optimization-for-local-businesses/",
          "position": 8
        },
        {
          "title": "Google's Ai Overviews, Ai...",
          "host": "aischemagen",
          "url": "https://www.aischemagen.com/blog/ai-search-optimization-for-local-business",
          "position": 9
        },
        {
          "title": "How to Get Your Local Business Recommended by ChatGPT, Gemini ...",
          "host": "seoinc",
          "url": "https://www.seoinc.com/seo-blog/get-your-local-business-recommended-by-ai/",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-b2b-saas-visibility",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-b2b-saas-visibility",
      "published_at": "2026-11-15T13:01:00.000Z",
      "updated_at": "2026-11-15T13:01:00.000Z",
      "revision": 1,
      "slug": "perplexity-b2b-saas-visibility",
      "panelId": "b2b-saas-visibility",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-b2b-saas-visibility",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "How should a B2B SaaS company measure AI search visibility?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 21:01 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/57e38e02-0bd3-487d-8b7c-960c57d20bae",
      "mentions": "Cited",
      "summary": "Presence: Track a fixed set of high-intent prompts across key AI assistants, segmented by buying stage and prompt type. Record whether your brand appears, how often, and whether it is cited; 40–100 well-chosen prompts can be a practical starting point for a single-product company.",
      "answerSnapshot": "7秒\n\nMeasure AI search visibility as a funnel, not one score:\n\nPresence: Track a fixed set of high-intent prompts across key AI assistants, segmented by buying stage and prompt type. Record whether your brand appears, how often, and whether it is cited; 40–100 well-chosen prompts can be a practical starting point for a single-product company.\ngrowthract\n+1\n\nQuality: Log your share of mentions versus competitors, citation URLs, answer position, and whether the description is accurate and favorable. Use first-party reporting where available, and keep a dated prompt log for assistants without it.\nlinkedin\n+1\n\nBusiness impact: Track AI-referred sessions, conversions, and pipeline in analytics/CRM, alongside AI-feature impressions where reported. These show whether visibility is reaching your site and contributing to demand—not just whether a model mentions you.\nrampkit\n+2\n\nKeep prompts and checks consistent over time, report trends by assistant and intent, and test changes against control prompts so model updates or market shifts aren’t mistaken for your work.\ncrawlspider",
      "citations": [
        {
          "title": "AI Search Visibility for B2B SaaS: Measure & Improve - Growthract",
          "host": "growthract",
          "url": "https://www.growthract.com/ai-search-visibility",
          "position": 1
        },
        {
          "title": "AI Search Visibility Strategy for B2B SaaS: A 90-Day Plan in 15 Moves (2026) | CrawlSpider",
          "host": "crawlspider",
          "url": "https://www.crawlspider.com/ai-search-visibility-strategy-b2b-saas-90-day-plan/",
          "position": 2
        },
        {
          "title": "How Do You Measure AI Search Visibility for Your SaaS? - LinkedIn",
          "host": "linkedin",
          "url": "https://www.linkedin.com/pulse/how-do-you-measure-ai-search-visibility-your-saas-bige-besikci-yaman-sfkte",
          "position": 3
        },
        {
          "title": "GEO Metrics for B2B SaaS: What to Measure in AI Search, and What Not to Report to the Board | Rampkit.co Blog",
          "host": "rampkit",
          "url": "https://www.rampkit.co/blog/geo-metrics-b2b-saas/",
          "position": 4
        },
        {
          "title": "How to Measure AI Search Visibility for B2B SaaS - madx.digital",
          "host": "madx",
          "url": "https://www.madx.digital/learn/how-to-measure-ai-search-visibility",
          "position": 5
        },
        {
          "title": "AI Search Visibility Tracking: Metrics and Tools (2027)",
          "host": "xql",
          "url": "https://xql.group/blog/how-to-measure-ai-search-visibility",
          "position": 6
        },
        {
          "title": "How to Measure Your AI Visibility as a B2B Provider",
          "host": "clutch",
          "url": "https://clutch.co/resources/ai-visibility-measurement-b2b",
          "position": 7
        },
        {
          "title": "AI Search Visibility: Metrics, Tracking and Fixes",
          "host": "shila",
          "url": "https://shila.so/blog/ai-search-visibility",
          "position": 8
        },
        {
          "title": "AI Search Visibility Optimization: How B2B SaaS Gets Cited",
          "host": "entlify",
          "url": "https://www.entlify.com/blog/ai-search-visibility-optimization",
          "position": 9
        },
        {
          "title": "AI Visibility for B2B SaaS: The 2026 Measurement Framework ...",
          "host": "seoauthori",
          "url": "https://www.seoauthori.com/en/blog/ai-visibility-b2b-saas-guide",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-b2b-saas-visibility-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-b2b-saas-visibility-repeat-2",
      "published_at": "2026-11-15T13:01:00.000Z",
      "updated_at": "2026-11-15T13:01:00.000Z",
      "revision": 1,
      "slug": "perplexity-b2b-saas-visibility-repeat-2",
      "panelId": "b2b-saas-visibility",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-b2b-saas-visibility",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "How should a B2B SaaS company measure AI search visibility?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 21:01 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/072907d9-892f-48f0-a2c2-93a7af31c2c1",
      "mentions": "Cited",
      "summary": "Measure AI search visibility as a set of signals, not one universal score. Track a representative set of buyer prompts across major assistants, then measure:",
      "answerSnapshot": "9秒\n\nMeasure AI search visibility as a set of signals, not one universal score. Track a representative set of buyer prompts across major assistants, then measure:\n\nPresence: how often your brand appears in answers, how often your pages are cited, and your share of mentions versus competitors.\n\nQuality: whether the answer describes your product accurately, recommends it in relevant comparisons, and cites useful pages.\n\nBusiness impact: AI-referred visits, conversions, and pipeline, alongside impressions in AI search features.\n\nUse first-party data where available, and keep a dated prompt log for assistants without reporting. Segment prompts by buyer journey and product area, and compare results over time; sampled answers can be noisy, so test changes against similar prompts you did not target. For one product, 40–100 well-chosen prompts can be a practical starting point.\nrampkit\n+4",
      "citations": [
        {
          "title": "GEO Metrics for B2B SaaS: What to Measure in AI Search, and What Not to Report to the Board | Rampkit.co Blog",
          "host": "rampkit",
          "url": "https://www.rampkit.co/blog/geo-metrics-b2b-saas/",
          "position": 1
        },
        {
          "title": "AI Search Visibility Strategy for B2B SaaS: A 90-Day Plan in 15 Moves (2026) | CrawlSpider",
          "host": "crawlspider",
          "url": "https://www.crawlspider.com/ai-search-visibility-strategy-b2b-saas-90-day-plan/",
          "position": 2
        },
        {
          "title": "AI Search Visibility Tracking: Metrics and Tools (2027)",
          "host": "xql",
          "url": "https://xql.group/blog/how-to-measure-ai-search-visibility",
          "position": 3
        },
        {
          "title": "How to Measure AI Search Visibility for B2B SaaS - madx.digital",
          "host": "madx",
          "url": "https://www.madx.digital/learn/how-to-measure-ai-search-visibility",
          "position": 4
        },
        {
          "title": "AI Search Visibility for B2B SaaS: Measure & Improve - Growthract",
          "host": "growthract",
          "url": "https://www.growthract.com/ai-search-visibility",
          "position": 5
        },
        {
          "title": "How Do You Measure AI Search Visibility for Your SaaS? - LinkedIn",
          "host": "linkedin",
          "url": "https://www.linkedin.com/pulse/how-do-you-measure-ai-search-visibility-your-saas-bige-besikci-yaman-sfkte",
          "position": 6
        },
        {
          "title": "How to Measure Your AI Visibility as a B2B Provider",
          "host": "clutch",
          "url": "https://clutch.co/resources/ai-visibility-measurement-b2b",
          "position": 7
        },
        {
          "title": "AI Search Visibility: Metrics, Tracking and Fixes",
          "host": "shila",
          "url": "https://shila.so/blog/ai-search-visibility",
          "position": 8
        },
        {
          "title": "AI Search Visibility Optimization: How B2B SaaS Gets Cited",
          "host": "entlify",
          "url": "https://www.entlify.com/blog/ai-search-visibility-optimization",
          "position": 9
        },
        {
          "title": "AI Visibility for B2B SaaS: The 2026 Measurement Framework ...",
          "host": "seoauthori",
          "url": "https://www.seoauthori.com/en/blog/ai-visibility-b2b-saas-guide",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-b2b-saas-visibility-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-b2b-saas-visibility-repeat-3",
      "published_at": "2026-11-15T13:01:00.000Z",
      "updated_at": "2026-11-15T13:01:00.000Z",
      "revision": 1,
      "slug": "perplexity-b2b-saas-visibility-repeat-3",
      "panelId": "b2b-saas-visibility",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-b2b-saas-visibility",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "How should a B2B SaaS company measure AI search visibility?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 21:01 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/e3a958ce-5d9d-4341-9248-f1f17309cc87",
      "mentions": "Cited",
      "summary": "Measure AI search visibility as a set of signals—not one universal score—and connect them to qualified pipeline. Track:",
      "answerSnapshot": "9秒\n\nMeasure AI search visibility as a set of signals—not one universal score—and connect them to qualified pipeline. Track:\n\nPresence: how often your brand appears in a fixed set of relevant prompts, segmented by buying stage, product, competitor, and AI platform.\n\nQuality: citation rate, prominence in the answer, accuracy of product claims, and whether you’re recommended in comparison or “best tool” answers. Keep a dated prompt log; tools’ sampled answers can vary.\n\nBusiness impact: AI-referred visits, conversions, qualified opportunities, and pipeline. Track referral sources in analytics, but don’t treat traffic alone as proof that AI visibility is driving demand.\n\nOn-site evidence: AI-search impressions and citations in available first-party reporting, which show whether your pages appear or are used—not necessarily whether an assistant recommends you.\nlinkedin\n+2\n\nStart with roughly 40–100 high-intent prompts for a single-product company, repeat them consistently, and compare changes against a control group so model updates or market shifts aren’t mistaken for the impact of your work.\ncrawlspider",
      "citations": [
        {
          "title": "How Do You Measure AI Search Visibility for Your SaaS? - LinkedIn",
          "host": "linkedin",
          "url": "https://www.linkedin.com/pulse/how-do-you-measure-ai-search-visibility-your-saas-bige-besikci-yaman-sfkte",
          "position": 1
        },
        {
          "title": "GEO Metrics for B2B SaaS: What to Measure in AI Search, and What Not to Report to the Board | Rampkit.co Blog",
          "host": "rampkit",
          "url": "https://www.rampkit.co/blog/geo-metrics-b2b-saas/",
          "position": 2
        },
        {
          "title": "How to Measure AI Search Visibility for B2B SaaS - madx.digital",
          "host": "madx",
          "url": "https://www.madx.digital/learn/how-to-measure-ai-search-visibility",
          "position": 3
        },
        {
          "title": "AI Search Visibility Strategy for B2B SaaS: A 90-Day Plan in 15 Moves (2026) | CrawlSpider",
          "host": "crawlspider",
          "url": "https://www.crawlspider.com/ai-search-visibility-strategy-b2b-saas-90-day-plan/",
          "position": 4
        },
        {
          "title": "AI Search Visibility for B2B SaaS: Measure & Improve - Growthract",
          "host": "growthract",
          "url": "https://www.growthract.com/ai-search-visibility",
          "position": 5
        },
        {
          "title": "How to Measure Your AI Visibility as a B2B Provider",
          "host": "clutch",
          "url": "https://clutch.co/resources/ai-visibility-measurement-b2b",
          "position": 6
        },
        {
          "title": "AI Search Visibility: Metrics, Tracking and Fixes",
          "host": "shila",
          "url": "https://shila.so/blog/ai-search-visibility",
          "position": 7
        },
        {
          "title": "AI Search Visibility Tracking: Metrics and Tools (2027)",
          "host": "xql",
          "url": "https://xql.group/blog/how-to-measure-ai-search-visibility",
          "position": 8
        },
        {
          "title": "AI Search Visibility Optimization: How B2B SaaS Gets Cited",
          "host": "entlify",
          "url": "https://www.entlify.com/blog/ai-search-visibility-optimization",
          "position": 9
        },
        {
          "title": "AI Visibility for B2B SaaS: The 2026 Measurement Framework ...",
          "host": "seoauthori",
          "url": "https://www.seoauthori.com/en/blog/ai-visibility-b2b-saas-guide",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-local-business-ai-search",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-local-business-ai-search",
      "published_at": "2026-11-15T13:01:00.000Z",
      "updated_at": "2026-11-15T13:01:00.000Z",
      "revision": 1,
      "slug": "perplexity-local-business-ai-search",
      "panelId": "local-business-ai-search",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-local-business-ai-search",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "Which pages should a local business optimize for AI search?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 21:01 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/38e7d664-6219-4ba8-96c4-4cb6d476720f",
      "mentions": "Cited",
      "summary": "Prioritize dedicated pages for each core service and each location or service area, rather than relying on your homepage alone. Add an FAQ page or clear Q&A sections answering common customer questions, and make sure each page plainly explains what you offer, who you serve, where you operate, and how to contact or book you.",
      "answerSnapshot": "9秒\n\nPrioritize dedicated pages for each core service and each location or service area, rather than relying on your homepage alone. Add an FAQ page or clear Q&A sections answering common customer questions, and make sure each page plainly explains what you offer, who you serve, where you operate, and how to contact or book you.\nthryv\n+2",
      "citations": [
        {
          "title": "AI Search Optimization for Local Business",
          "host": "thryv",
          "url": "https://www.thryv.com/blog/ai-search-optimization-for-local-business/",
          "position": 1
        },
        {
          "title": "AI Search Optimization for Local Businesses in 2026",
          "host": "searchable",
          "url": "https://www.searchable.com/blog/local-business-ai-search-optimization-2026",
          "position": 2
        },
        {
          "title": "10 Ways to Show Up in AI Search - soci.ai",
          "host": "soci",
          "url": "https://www.soci.ai/blog/10-ways-to-show-up-in-ai-search/",
          "position": 3
        },
        {
          "title": "Local AI SEO: Optimize Your Site for AI Search",
          "host": "squarespace",
          "url": "https://www.squarespace.com/blog/local-ai-seo",
          "position": 4
        },
        {
          "title": "How to Optimize Your Local Business for AI Search and Discovery",
          "host": "keyword",
          "url": "https://keyword.com/blog/local-seo-for-ai-search/",
          "position": 5
        },
        {
          "title": "Home Services SEO Playbook: Get Found on Google, Maps, and AI Search",
          "host": "vendasta",
          "url": "https://www.vendasta.com/blog/home-services-seo-google-maps-ai-search/",
          "position": 6
        },
        {
          "title": "How to do AI search optimization for local businesses",
          "host": "semrush",
          "url": "https://www.semrush.com/blog/ai-search-optimization-for-local-businesses/",
          "position": 7
        },
        {
          "title": "Google's Ai Overviews, Ai...",
          "host": "aischemagen",
          "url": "https://www.aischemagen.com/blog/ai-search-optimization-for-local-business",
          "position": 8
        },
        {
          "title": "How to Get Your Local Business Recommended by ChatGPT, Gemini ...",
          "host": "seoinc",
          "url": "https://www.seoinc.com/seo-blog/get-your-local-business-recommended-by-ai/",
          "position": 9
        },
        {
          "title": "Practical, step-by-step optimization guide for local and small ...",
          "host": "uandai",
          "url": "https://uandai.co/blog/chatgpt-visibility-local-businesses-2026",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-mention-vs-citation",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-mention-vs-citation",
      "published_at": "2026-11-15T13:00:00.000Z",
      "updated_at": "2026-11-15T13:00:00.000Z",
      "revision": 1,
      "slug": "perplexity-mention-vs-citation",
      "panelId": "mention-vs-citation",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-mention-vs-citation",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "What is the difference between an AI mention and an AI citation?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 21:00 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/d0cc48fb-185e-4634-8b63-2718245203d5",
      "mentions": "Cited",
      "summary": "An AI mention is when an AI-generated answer names your brand; an AI citation is when it credits or links to your page as a source. A brand can be mentioned without being cited, or cited without being named in the answer.",
      "answerSnapshot": "9秒\n\nAn AI mention is when an AI-generated answer names your brand; an AI citation is when it credits or links to your page as a source. A brand can be mentioned without being cited, or cited without being named in the answer.\naisearch.similarweb\n+1",
      "citations": [
        {
          "title": "AI Mentions vs Citations: Key Differences for GEO",
          "host": "aisearch.similarweb",
          "url": "https://aisearch.similarweb.com/blog/ai-mentions-vs-ai-citations/",
          "position": 1
        },
        {
          "title": "AI Citations Vs. Mentions: What Do They Mean? - Victorious",
          "host": "victorious",
          "url": "https://victorious.com/blog/ai-citations-vs-mentions/",
          "position": 2
        },
        {
          "title": "AI citations vs mentions: key differences and the strategy to ...",
          "host": "digitalhitmen.com",
          "url": "https://www.digitalhitmen.com.au/blog/ai-citations-vs-mentions/",
          "position": 3
        },
        {
          "title": "AI Citations vs Mentions: What's the SEO Difference?",
          "host": "searchatlas",
          "url": "https://searchatlas.com/blog/ai-citations-vs-mentions/",
          "position": 4
        },
        {
          "title": "AI citation tracking: How to track (and grow) AI engine citations",
          "host": "blog.hubspot",
          "url": "https://blog.hubspot.com/marketing/ai-citation-tracking",
          "position": 5
        },
        {
          "title": "LLM Mentions vs Citations: What They Mean & Why Both ...",
          "host": "airops",
          "url": "https://www.airops.com/blog/difference-mentions-citations-chatgpt",
          "position": 6
        },
        {
          "title": "AI mentions versus citations: what each tool counts",
          "host": "aiwiz",
          "url": "https://aiwiz.uk/blog/ai-mentions-vs-citations/",
          "position": 7
        },
        {
          "title": "AI Mention: How It Works & Tips to Get Your Brand Mentioned",
          "host": "cmlabs",
          "url": "https://cmlabs.co/en/blog/ai-mention",
          "position": 8
        },
        {
          "title": "AI Citation vs AI Mention: Differences, Examples and Metrics ...",
          "host": "novaverb",
          "url": "https://novaverb.com/blog/ai-citation-vs-ai-mention",
          "position": 9
        },
        {
          "title": "AI Mentions vs Citations: What's the Difference?",
          "host": "rankvia",
          "url": "https://www.rankvia.ai/guides/ai-mentions-vs-citations",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-mention-vs-citation-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-mention-vs-citation-repeat-2",
      "published_at": "2026-11-15T13:00:00.000Z",
      "updated_at": "2026-11-15T13:00:00.000Z",
      "revision": 1,
      "slug": "perplexity-mention-vs-citation-repeat-2",
      "panelId": "mention-vs-citation",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-mention-vs-citation",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "What is the difference between an AI mention and an AI citation?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 21:00 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/cc206fb5-de26-490f-9b8c-e26a6af0ffe0",
      "mentions": "Cited",
      "summary": "An AI mention is when a brand or product name appears in an AI-generated answer; an AI citation is when the AI credits a specific webpage or other source, often with a link. They can happen independently: a brand may be mentioned without its content being cited, or a page may be cited without naming the brand.",
      "answerSnapshot": "8秒\n\nAn AI mention is when a brand or product name appears in an AI-generated answer; an AI citation is when the AI credits a specific webpage or other source, often with a link. They can happen independently: a brand may be mentioned without its content being cited, or a page may be cited without naming the brand.\naisearch.similarweb\n+1",
      "citations": [
        {
          "title": "AI Mentions vs Citations: Key Differences for GEO",
          "host": "aisearch.similarweb",
          "url": "https://aisearch.similarweb.com/blog/ai-mentions-vs-ai-citations/",
          "position": 1
        },
        {
          "title": "AI Citations Vs. Mentions: What Do They Mean? - Victorious",
          "host": "victorious",
          "url": "https://victorious.com/blog/ai-citations-vs-mentions/",
          "position": 2
        },
        {
          "title": "AI citations vs mentions: key differences and the strategy to ...",
          "host": "digitalhitmen.com",
          "url": "https://www.digitalhitmen.com.au/blog/ai-citations-vs-mentions/",
          "position": 3
        },
        {
          "title": "AI Citations vs Mentions: What's the SEO Difference?",
          "host": "searchatlas",
          "url": "https://searchatlas.com/blog/ai-citations-vs-mentions/",
          "position": 4
        },
        {
          "title": "AI citation tracking: How to track (and grow) AI engine citations",
          "host": "blog.hubspot",
          "url": "https://blog.hubspot.com/marketing/ai-citation-tracking",
          "position": 5
        },
        {
          "title": "LLM Mentions vs Citations: What They Mean & Why Both ...",
          "host": "airops",
          "url": "https://www.airops.com/blog/difference-mentions-citations-chatgpt",
          "position": 6
        },
        {
          "title": "AI mentions versus citations: what each tool counts",
          "host": "aiwiz",
          "url": "https://aiwiz.uk/blog/ai-mentions-vs-citations/",
          "position": 7
        },
        {
          "title": "AI Mention: How It Works & Tips to Get Your Brand Mentioned",
          "host": "cmlabs",
          "url": "https://cmlabs.co/en/blog/ai-mention",
          "position": 8
        },
        {
          "title": "AI Citation vs AI Mention: Differences, Examples and Metrics ...",
          "host": "novaverb",
          "url": "https://novaverb.com/blog/ai-citation-vs-ai-mention",
          "position": 9
        },
        {
          "title": "AI Mentions vs Citations: What's the Difference?",
          "host": "rankvia",
          "url": "https://www.rankvia.ai/guides/ai-mentions-vs-citations",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-mention-vs-citation-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-mention-vs-citation-repeat-3",
      "published_at": "2026-11-15T13:00:00.000Z",
      "updated_at": "2026-11-15T13:00:00.000Z",
      "revision": 1,
      "slug": "perplexity-mention-vs-citation-repeat-3",
      "panelId": "mention-vs-citation",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-mention-vs-citation",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "What is the difference between an AI mention and an AI citation?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 21:00 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/919d6c81-c9ec-4b65-bbf2-1e6532c99f1a",
      "mentions": "Cited",
      "summary": "An AI mention is when a brand or product name appears in an AI-generated answer; an AI citation is when the answer credits a specific webpage or other content as a source, often with a link. A brand can be mentioned without being cited, cited without being named, or both.",
      "answerSnapshot": "11秒\n\nAn AI mention is when a brand or product name appears in an AI-generated answer; an AI citation is when the answer credits a specific webpage or other content as a source, often with a link. A brand can be mentioned without being cited, cited without being named, or both.\naisearch.similarweb\n+1",
      "citations": [
        {
          "title": "AI Mentions vs Citations: Key Differences for GEO",
          "host": "aisearch.similarweb",
          "url": "https://aisearch.similarweb.com/blog/ai-mentions-vs-ai-citations/",
          "position": 1
        },
        {
          "title": "AI Citations Vs. Mentions: What Do They Mean? - Victorious",
          "host": "victorious",
          "url": "https://victorious.com/blog/ai-citations-vs-mentions/",
          "position": 2
        },
        {
          "title": "AI citations vs mentions: key differences and the strategy to ...",
          "host": "digitalhitmen.com",
          "url": "https://www.digitalhitmen.com.au/blog/ai-citations-vs-mentions/",
          "position": 3
        },
        {
          "title": "AI Citations vs Mentions: What's the SEO Difference?",
          "host": "searchatlas",
          "url": "https://searchatlas.com/blog/ai-citations-vs-mentions/",
          "position": 4
        },
        {
          "title": "AI citation tracking: How to track (and grow) AI engine citations",
          "host": "blog.hubspot",
          "url": "https://blog.hubspot.com/marketing/ai-citation-tracking",
          "position": 5
        },
        {
          "title": "LLM Mentions vs Citations: What They Mean & Why Both ...",
          "host": "airops",
          "url": "https://www.airops.com/blog/difference-mentions-citations-chatgpt",
          "position": 6
        },
        {
          "title": "AI mentions versus citations: what each tool counts",
          "host": "aiwiz",
          "url": "https://aiwiz.uk/blog/ai-mentions-vs-citations/",
          "position": 7
        },
        {
          "title": "AI Mention: How It Works & Tips to Get Your Brand Mentioned",
          "host": "cmlabs",
          "url": "https://cmlabs.co/en/blog/ai-mention",
          "position": 8
        },
        {
          "title": "AI Citation vs AI Mention: Differences, Examples and Metrics ...",
          "host": "novaverb",
          "url": "https://novaverb.com/blog/ai-citation-vs-ai-mention",
          "position": 9
        },
        {
          "title": "AI Mentions vs Citations: What's the Difference?",
          "host": "rankvia",
          "url": "https://www.rankvia.ai/guides/ai-mentions-vs-citations",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-verify-ai-citation-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-verify-ai-citation-repeat-3",
      "published_at": "2026-11-15T13:00:00.000Z",
      "updated_at": "2026-11-15T13:00:00.000Z",
      "revision": 1,
      "slug": "perplexity-verify-ai-citation-repeat-3",
      "panelId": "verify-ai-citation",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-verify-ai-citation",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "How can a publisher verify that an AI answer cited its source correctly?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 21:00 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/19668c25-454b-45bf-96db-69aec686d036",
      "mentions": "Cited",
      "summary": "Open the cited page and verify that it exists, matches the stated title, author, publisher, and date, and—most importantly—that the specific passage supports the answer’s exact claim, including its context and qualifications. A working link alone doesn’t prove the citation is correct.",
      "answerSnapshot": "9秒\n\nOpen the cited page and verify that it exists, matches the stated title, author, publisher, and date, and—most importantly—that the specific passage supports the answer’s exact claim, including its context and qualifications. A working link alone doesn’t prove the citation is correct.\nizood\n+2\n\nFor a publisher’s own content, compare the answer with the original page and flag misquotes, outdated details, missing caveats, or claims the page doesn’t support. Check each factual claim separately when a sentence contains several.\nrillor\n+1",
      "citations": [
        {
          "title": "How to Verify AI Citations: 5 Checks Beyond the Link",
          "host": "izood",
          "url": "https://izood.net/ai/how-to-verify-ai-citations",
          "position": 1
        },
        {
          "title": "How to Verify AI-Generated Information Before You Trust It",
          "host": "collegenp",
          "url": "https://www.collegenp.com/technology/verify-ai-generated-information-before-trust",
          "position": 2
        },
        {
          "title": "Do Source Citations Guarantee That an AI-Generated Answer Is Correct?",
          "host": "manuelgarcia",
          "url": "https://manuelgarcia.info/guides/research/guide/ai-source-citations-answer-accuracy",
          "position": 3
        },
        {
          "title": "How to check LLM citations against their sources | Rillor",
          "host": "rillor",
          "url": "https://rillor.com/insights/checking-llm-citations-against-sources",
          "position": 4
        },
        {
          "title": "What Is AI Citation Accuracy? Why Getting Cited Isn't Enough",
          "host": "thepromptinsider",
          "url": "https://thepromptinsider.com/aeo/ai-citation-accuracy/",
          "position": 5
        },
        {
          "title": "How to Check AI Answers Before You Trust Them",
          "host": "aiforbeginners.substack",
          "url": "https://aiforbeginners.substack.com/p/how-to-check-ai-answers-before-you",
          "position": 6
        },
        {
          "title": "How to Check Whether an AI Answer Is True: A Practical Verification ...",
          "host": "opentechtips",
          "url": "https://opentechtips.com/how-to-verify-ai-answers/",
          "position": 7
        },
        {
          "title": "How to Verify AI-Generated Answers and Citations - AOFIRS",
          "host": "aofirs",
          "url": "https://aofirs.org/articles/how-to-verify-ai-generated-answers-and-citations/",
          "position": 8
        },
        {
          "title": "How to Check if an AI Citation Is Real and Reliable — WorkWithAI.expert",
          "host": "workwithai",
          "url": "https://workwithai.expert/read/how-to-check-if-ai-citation-is-real-relevant-up-to-date",
          "position": 9
        },
        {
          "title": "Why AI Answers Have No Citations and How to Investigate",
          "host": "cooperaiq",
          "url": "https://cooperaiq.com/news/guides/why-ai-answers-have-no-citations",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-source-freshness-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-source-freshness-repeat-2",
      "published_at": "2026-11-15T12:59:00.000Z",
      "updated_at": "2026-11-15T12:59:00.000Z",
      "revision": 1,
      "slug": "perplexity-source-freshness-repeat-2",
      "panelId": "source-freshness",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-source-freshness",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "How do AI search engines decide whether information is fresh enough?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 20:59 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/047fee23-565a-4cf8-8bfd-f4f5dee8e7dd",
      "mentions": "Cited",
      "summary": "AI search engines don’t use one universal “freshness” cutoff. They weigh whether the query is time-sensitive, when a page was published or updated, whether its facts and data are still current, and whether the system has recently crawled and indexed it. A visible “last updated” date alone doesn’t prove the engine has seen or verified the change.",
      "answerSnapshot": "9秒\n\nAI search engines don’t use one universal “freshness” cutoff. They weigh whether the query is time-sensitive, when a page was published or updated, whether its facts and data are still current, and whether the system has recently crawled and indexed it. A visible “last updated” date alone doesn’t prove the engine has seen or verified the change.\nsearchable\n+2\n\nSo an older source may still suit a stable topic, while current prices, news, or product details call for recently retrieved and corroborated sources.\nkime\n+1",
      "citations": [
        {
          "title": "Content Freshness Signals: How AI Determines If Your ... - Searchable",
          "host": "searchable",
          "url": "https://www.searchable.com/blog/content-freshness-signals",
          "position": 1
        },
        {
          "title": "When Content Freshness Affects AI Search Citations",
          "host": "deepsmith",
          "url": "https://deepsmith.ai/blog/content-freshness-ai-search-citations",
          "position": 2
        },
        {
          "title": "Maintain content freshness for AI search",
          "host": "siteimprove",
          "url": "https://www.siteimprove.com/blog/maintaining-content-freshness-for-ai-search/",
          "position": 3
        },
        {
          "title": "What Is Content Freshness? Why It Matters for AI | KIME",
          "host": "kime",
          "url": "https://kime.ai/glossary/content-freshness",
          "position": 4
        },
        {
          "title": "How Fresh Is the Data Returned by Search APIs?",
          "host": "olostep",
          "url": "https://www.olostep.com/glossary/agentic-search-apis/how-fresh-data-search-apis",
          "position": 5
        },
        {
          "title": "Source Freshness: Check Dates in a Live Web Search ...",
          "host": "tinyfish",
          "url": "https://www.tinyfish.ai/blog/source-freshness-live-web-search",
          "position": 6
        },
        {
          "title": "Does Content Freshness Influence AI Search? [Data Insights] - OMNIUS",
          "host": "omnius",
          "url": "https://www.omnius.so/blog/does-content-freshness-influence-ai-search",
          "position": 7
        },
        {
          "title": "Content freshness signals for generative AI systems",
          "host": "spawned",
          "url": "https://spawned.com/learn/content-freshness-signals-for-generative-ai-systems",
          "position": 8
        },
        {
          "title": "Content Freshness Signals That AI Search Engines Look For | AY ...",
          "host": "ayrank",
          "url": "https://www.ayrank.com/blog/content-freshness-signals-ai-search",
          "position": 9
        },
        {
          "title": "What's the role of freshness in keeping information visible in AI ...",
          "host": "senso",
          "url": "https://www.senso.ai/prompts-content/whats-the-role-of-freshness-in-keeping-information-visible-in-ai-search",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-source-freshness-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-source-freshness-repeat-3",
      "published_at": "2026-11-15T12:59:00.000Z",
      "updated_at": "2026-11-15T12:59:00.000Z",
      "revision": 1,
      "slug": "perplexity-source-freshness-repeat-3",
      "panelId": "source-freshness",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-source-freshness",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "How do AI search engines decide whether information is fresh enough?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 20:59 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/dd788d2c-727f-41c7-99c7-72cbd8187896",
      "mentions": "Cited",
      "summary": "AI search engines don’t use one universal “freshness” cutoff. They weigh whether the query needs current information, when a page was published or updated, whether its facts still match the present, and whether the engine has recently crawled and indexed it.",
      "answerSnapshot": "8秒\n\nAI search engines don’t use one universal “freshness” cutoff. They weigh whether the query needs current information, when a page was published or updated, whether its facts still match the present, and whether the engine has recently crawled and indexed it. A new date alone doesn’t prove the content was meaningfully updated—or that the engine has seen the change.\nsearchable\n+2\n\nSo stable facts can remain useful for years, while prices, product features, and “latest” comparisons can go stale quickly; freshness is judged against the topic, not just the calendar.\nkime\n+2",
      "citations": [
        {
          "title": "Content Freshness Signals: How AI Determines If Your ... - Searchable",
          "host": "searchable",
          "url": "https://www.searchable.com/blog/content-freshness-signals",
          "position": 1
        },
        {
          "title": "When Content Freshness Affects AI Search Citations",
          "host": "deepsmith",
          "url": "https://deepsmith.ai/blog/content-freshness-ai-search-citations",
          "position": 2
        },
        {
          "title": "Maintain content freshness for AI search",
          "host": "siteimprove",
          "url": "https://www.siteimprove.com/blog/maintaining-content-freshness-for-ai-search/",
          "position": 3
        },
        {
          "title": "What Is Content Freshness? Why It Matters for AI | KIME",
          "host": "kime",
          "url": "https://kime.ai/glossary/content-freshness",
          "position": 4
        },
        {
          "title": "How Fresh Is the Data Returned by Search APIs?",
          "host": "olostep",
          "url": "https://www.olostep.com/glossary/agentic-search-apis/how-fresh-data-search-apis",
          "position": 5
        },
        {
          "title": "Content Freshness Signals That AI Search Engines Look For | AY ...",
          "host": "ayrank",
          "url": "https://www.ayrank.com/blog/content-freshness-signals-ai-search",
          "position": 6
        },
        {
          "title": "Source Freshness: Check Dates in a Live Web Search ...",
          "host": "tinyfish",
          "url": "https://www.tinyfish.ai/blog/source-freshness-live-web-search",
          "position": 7
        },
        {
          "title": "Does Content Freshness Influence AI Search? [Data Insights] - OMNIUS",
          "host": "omnius",
          "url": "https://www.omnius.so/blog/does-content-freshness-influence-ai-search",
          "position": 8
        },
        {
          "title": "Content freshness signals for generative AI systems",
          "host": "spawned",
          "url": "https://spawned.com/learn/content-freshness-signals-for-generative-ai-systems",
          "position": 9
        },
        {
          "title": "What's the role of freshness in keeping information visible in AI ...",
          "host": "senso",
          "url": "https://www.senso.ai/prompts-content/whats-the-role-of-freshness-in-keeping-information-visible-in-ai-search",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-verify-ai-citation",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-verify-ai-citation",
      "published_at": "2026-11-15T12:59:00.000Z",
      "updated_at": "2026-11-15T12:59:00.000Z",
      "revision": 1,
      "slug": "perplexity-verify-ai-citation",
      "panelId": "verify-ai-citation",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-verify-ai-citation",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "How can a publisher verify that an AI answer cited its source correctly?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 20:59 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/86cc9ca5-68e8-45c9-a89e-af836730389d",
      "mentions": "Cited",
      "summary": "Copy the citation exactly as shown, then verify that the source exists and the title, author, publisher, and date match. Open the original source and check that its text supports the specific claim—including relevant context and qualifications; a working link alone isn’t enough.",
      "answerSnapshot": "9秒\n\nCopy the citation exactly as shown, then verify that the source exists and the title, author, publisher, and date match. Open the original source and check that its text supports the specific claim—including relevant context and qualifications; a working link alone isn’t enough.\nworkwithai\n+2",
      "citations": [
        {
          "title": "How to Check if an AI Citation Is Real and Reliable — WorkWithAI.expert",
          "host": "workwithai",
          "url": "https://workwithai.expert/read/how-to-check-if-ai-citation-is-real-relevant-up-to-date",
          "position": 1
        },
        {
          "title": "Do Source Citations Guarantee That an AI-Generated Answer Is Correct?",
          "host": "manuelgarcia",
          "url": "https://manuelgarcia.info/guides/research/guide/ai-source-citations-answer-accuracy",
          "position": 2
        },
        {
          "title": "How to Verify AI Citations: 5 Checks Beyond the Link",
          "host": "izood",
          "url": "https://izood.net/ai/how-to-verify-ai-citations",
          "position": 3
        },
        {
          "title": "How to Check AI Answers Before You Trust Them",
          "host": "aiforbeginners.substack",
          "url": "https://aiforbeginners.substack.com/p/how-to-check-ai-answers-before-you",
          "position": 4
        },
        {
          "title": "How to Check Whether an AI Answer Is True: A Practical Verification ...",
          "host": "opentechtips",
          "url": "https://opentechtips.com/how-to-verify-ai-answers/",
          "position": 5
        },
        {
          "title": "How to Verify AI-Generated Information Before You Trust It",
          "host": "collegenp",
          "url": "https://www.collegenp.com/technology/verify-ai-generated-information-before-trust",
          "position": 6
        },
        {
          "title": "How to Verify AI-Generated Answers and Citations - AOFIRS",
          "host": "aofirs",
          "url": "https://aofirs.org/articles/how-to-verify-ai-generated-answers-and-citations/",
          "position": 7
        },
        {
          "title": "Why AI Answers Have No Citations and How to Investigate",
          "host": "cooperaiq",
          "url": "https://cooperaiq.com/news/guides/why-ai-answers-have-no-citations",
          "position": 8
        },
        {
          "title": "How to check LLM citations against their sources | Rillor",
          "host": "rillor",
          "url": "https://rillor.com/insights/checking-llm-citations-against-sources",
          "position": 9
        },
        {
          "title": "What Is AI Citation Accuracy? Why Getting Cited Isn't Enough",
          "host": "thepromptinsider",
          "url": "https://thepromptinsider.com/aeo/ai-citation-accuracy/",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-verify-ai-citation-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-verify-ai-citation-repeat-2",
      "published_at": "2026-11-15T12:59:00.000Z",
      "updated_at": "2026-11-15T12:59:00.000Z",
      "revision": 1,
      "slug": "perplexity-verify-ai-citation-repeat-2",
      "panelId": "verify-ai-citation",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-verify-ai-citation",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "How can a publisher verify that an AI answer cited its source correctly?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 20:59 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/451fecaa-f586-4a70-807c-76b7e0bd2be1",
      "mentions": "Cited",
      "summary": "Compare the answer’s exact claim with the cited page—not just whether the link works. Check that the source exists and the title, author, publisher, and date are correct; then find the relevant passage and confirm it supports the claim with the same scope and qualifications. A genuine citation can still be misattributed or fail to support the statement.",
      "answerSnapshot": "8秒\n\nCompare the answer’s exact claim with the cited page—not just whether the link works. Check that the source exists and the title, author, publisher, and date are correct; then find the relevant passage and confirm it supports the claim with the same scope and qualifications. A genuine citation can still be misattributed or fail to support the statement.\nizood\n+2",
      "citations": [
        {
          "title": "How to Verify AI Citations: 5 Checks Beyond the Link",
          "host": "izood",
          "url": "https://izood.net/ai/how-to-verify-ai-citations",
          "position": 1
        },
        {
          "title": "How to Verify AI-Generated Information Before You Trust It",
          "host": "collegenp",
          "url": "https://www.collegenp.com/technology/verify-ai-generated-information-before-trust",
          "position": 2
        },
        {
          "title": "Do Source Citations Guarantee That an AI-Generated Answer Is Correct?",
          "host": "manuelgarcia",
          "url": "https://manuelgarcia.info/guides/research/guide/ai-source-citations-answer-accuracy",
          "position": 3
        },
        {
          "title": "How to Check AI Answers Before You Trust Them",
          "host": "aiforbeginners.substack",
          "url": "https://aiforbeginners.substack.com/p/how-to-check-ai-answers-before-you",
          "position": 4
        },
        {
          "title": "How to Check Whether an AI Answer Is True: A Practical Verification ...",
          "host": "opentechtips",
          "url": "https://opentechtips.com/how-to-verify-ai-answers/",
          "position": 5
        },
        {
          "title": "How to Verify AI-Generated Answers and Citations - AOFIRS",
          "host": "aofirs",
          "url": "https://aofirs.org/articles/how-to-verify-ai-generated-answers-and-citations/",
          "position": 6
        },
        {
          "title": "How to Check if an AI Citation Is Real and Reliable — WorkWithAI.expert",
          "host": "workwithai",
          "url": "https://workwithai.expert/read/how-to-check-if-ai-citation-is-real-relevant-up-to-date",
          "position": 7
        },
        {
          "title": "Why AI Answers Have No Citations and How to Investigate",
          "host": "cooperaiq",
          "url": "https://cooperaiq.com/news/guides/why-ai-answers-have-no-citations",
          "position": 8
        },
        {
          "title": "How to check LLM citations against their sources | Rillor",
          "host": "rillor",
          "url": "https://rillor.com/insights/checking-llm-citations-against-sources",
          "position": 9
        },
        {
          "title": "What Is AI Citation Accuracy? Why Getting Cited Isn't Enough",
          "host": "thepromptinsider",
          "url": "https://thepromptinsider.com/aeo/ai-citation-accuracy/",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-official-documentation-preference",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-official-documentation-preference",
      "published_at": "2026-11-15T12:58:00.000Z",
      "updated_at": "2026-11-15T12:58:00.000Z",
      "revision": 1,
      "slug": "perplexity-official-documentation-preference",
      "panelId": "official-documentation-preference",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-official-documentation-preference",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "Do AI search engines prefer official documentation?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 20:58 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/e50eb91d-478e-4e29-8013-f19dafca64fa",
      "mentions": "Cited",
      "summary": "Often, but it depends on the question: AI search engines tend to favor official documentation for product, technical, or other authoritative facts, while comparison and recommendation queries are more likely to draw on independent sources. There’s no universal preference; clarity, relevance, and freshness also matter.",
      "answerSnapshot": "9秒\n\nOften, but it depends on the question: AI search engines tend to favor official documentation for product, technical, or other authoritative facts, while comparison and recommendation queries are more likely to draw on independent sources. There’s no universal preference; clarity, relevance, and freshness also matter.\nbetterdocs\n+2",
      "citations": [
        {
          "title": "Does AI Search Prefer Third-Party Sources or Brand Websites?",
          "host": "neuraladx",
          "url": "https://neuraladx.com/ai-search-source-preference-third-party-vs-brand-websites/",
          "position": 1
        },
        {
          "title": "Documentation SEO: How to Make Your Docs the Source AI Cites | RankZero Guides",
          "host": "rankzero",
          "url": "https://www.rankzero.io/guides/documentation-seo",
          "position": 2
        },
        {
          "title": "How AI Search Is Changing the Way Users Find Help And ...",
          "host": "betterdocs",
          "url": "https://betterdocs.co/how-ai-search-is-changing-documentation/",
          "position": 3
        },
        {
          "title": "Latest Google Search Documentation Updates | Google Search Central | What's new | Google for Developers",
          "host": "developers.google",
          "url": "https://developers.google.com/search/updates?authuser=1&visit_id=639157400745103073-1443175028&rd=1",
          "position": 4
        },
        {
          "title": "Why AI Search Prefers Product Pages Over Blogs",
          "host": "arobis",
          "url": "https://arobis.ai/blog/why-ai-search-engines-prefer-your-product-and-landing-pages",
          "position": 5
        },
        {
          "title": "How AI Search Engines Choose Sources to Cite",
          "host": "rankray",
          "url": "https://rankray.com/blog/how-chatgpt-ai-search-engines-choose-sources/",
          "position": 6
        },
        {
          "title": "Best Practices: Optimize Product Docs for AI Search & User Queries",
          "host": "geneo",
          "url": "https://geneo.app/blog/optimize-product-documentation-for-ai-search-best-practices/",
          "position": 7
        },
        {
          "title": "qiaomu-seo/references/ai-search.md at main · joeseesun/qiaomu-seo",
          "host": "github",
          "url": "https://github.com/joeseesun/qiaomu-seo/blob/main/references/ai-search.md",
          "position": 8
        },
        {
          "title": "Technical SEO for Docs That AI Can Discover",
          "host": "skayle",
          "url": "https://skayle.ai/blog/technical-seo-for-ai-discovery",
          "position": 9
        },
        {
          "title": "Your docs are now read by machines",
          "host": "joelolawanle",
          "url": "https://joelolawanle.com/blog/ai-readable-documentation",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-official-documentation-preference-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-official-documentation-preference-repeat-2",
      "published_at": "2026-11-15T12:58:00.000Z",
      "updated_at": "2026-11-15T12:58:00.000Z",
      "revision": 1,
      "slug": "perplexity-official-documentation-preference-repeat-2",
      "panelId": "official-documentation-preference",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-official-documentation-preference",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "Do AI search engines prefer official documentation?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 20:58 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/3e1329d8-ee9f-412e-ad2b-2b14a6992139",
      "mentions": "Cited",
      "summary": "Sometimes. AI search engines tend to favor official documentation for factual questions about a product, platform, or standard, but they’re more likely to rely on independent sources for comparisons, recommendations, and reputation questions. There’s no universal preference; it depends on the query.",
      "answerSnapshot": "8秒\n\nSometimes. AI search engines tend to favor official documentation for factual questions about a product, platform, or standard, but they’re more likely to rely on independent sources for comparisons, recommendations, and reputation questions. There’s no universal preference; it depends on the query.\nneuraladx\n+1",
      "citations": [
        {
          "title": "Does AI Search Prefer Third-Party Sources or Brand Websites?",
          "host": "neuraladx",
          "url": "https://neuraladx.com/ai-search-source-preference-third-party-vs-brand-websites/",
          "position": 1
        },
        {
          "title": "Documentation SEO: How to Make Your Docs the Source AI Cites | RankZero Guides",
          "host": "rankzero",
          "url": "https://www.rankzero.io/guides/documentation-seo",
          "position": 2
        },
        {
          "title": "Latest Google Search Documentation Updates | Google Search Central | What's new | Google for Developers",
          "host": "developers.google",
          "url": "https://developers.google.com/search/updates?authuser=1&visit_id=639157400745103073-1443175028&rd=1",
          "position": 3
        },
        {
          "title": "How AI Search Is Changing the Way Users Find Help And ...",
          "host": "betterdocs",
          "url": "https://betterdocs.co/how-ai-search-is-changing-documentation/",
          "position": 4
        },
        {
          "title": "Why AI Search Prefers Product Pages Over Blogs",
          "host": "arobis",
          "url": "https://arobis.ai/blog/why-ai-search-engines-prefer-your-product-and-landing-pages",
          "position": 5
        },
        {
          "title": "How AI Search Engines Choose Sources to Cite",
          "host": "rankray",
          "url": "https://rankray.com/blog/how-chatgpt-ai-search-engines-choose-sources/",
          "position": 6
        },
        {
          "title": "Best Practices: Optimize Product Docs for AI Search & User Queries",
          "host": "geneo",
          "url": "https://geneo.app/blog/optimize-product-documentation-for-ai-search-best-practices/",
          "position": 7
        },
        {
          "title": "qiaomu-seo/references/ai-search.md at main · joeseesun/qiaomu-seo",
          "host": "github",
          "url": "https://github.com/joeseesun/qiaomu-seo/blob/main/references/ai-search.md",
          "position": 8
        },
        {
          "title": "Technical SEO for Docs That AI Can Discover",
          "host": "skayle",
          "url": "https://skayle.ai/blog/technical-seo-for-ai-discovery",
          "position": 9
        },
        {
          "title": "Your docs are now read by machines",
          "host": "joelolawanle",
          "url": "https://joelolawanle.com/blog/ai-readable-documentation",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-official-documentation-preference-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-official-documentation-preference-repeat-3",
      "published_at": "2026-11-15T12:58:00.000Z",
      "updated_at": "2026-11-15T12:58:00.000Z",
      "revision": 1,
      "slug": "perplexity-official-documentation-preference-repeat-3",
      "panelId": "official-documentation-preference",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-official-documentation-preference",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "Do AI search engines prefer official documentation?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 20:58 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/e6111b32-4abe-415c-9f29-1c48df6eabb3",
      "mentions": "Cited",
      "summary": "Sometimes. AI search engines tend to rely on official documentation for factual questions about how a specific product, platform, or standard works, but use independent sources more often for comparisons, recommendations, and reputation questions. There’s no universal preference; it depends on what the user asks.",
      "answerSnapshot": "9秒\n\nSometimes. AI search engines tend to rely on official documentation for factual questions about how a specific product, platform, or standard works, but use independent sources more often for comparisons, recommendations, and reputation questions. There’s no universal preference; it depends on what the user asks.\nneuraladx\n+1",
      "citations": [
        {
          "title": "Does AI Search Prefer Third-Party Sources or Brand Websites?",
          "host": "neuraladx",
          "url": "https://neuraladx.com/ai-search-source-preference-third-party-vs-brand-websites/",
          "position": 1
        },
        {
          "title": "Documentation SEO: How to Make Your Docs the Source AI Cites | RankZero Guides",
          "host": "rankzero",
          "url": "https://www.rankzero.io/guides/documentation-seo",
          "position": 2
        },
        {
          "title": "Latest Google Search Documentation Updates | Google Search Central | What's new | Google for Developers",
          "host": "developers.google",
          "url": "https://developers.google.com/search/updates?authuser=1&visit_id=639157400745103073-1443175028&rd=1",
          "position": 3
        },
        {
          "title": "How AI Search Is Changing the Way Users Find Help And ...",
          "host": "betterdocs",
          "url": "https://betterdocs.co/how-ai-search-is-changing-documentation/",
          "position": 4
        },
        {
          "title": "Why AI Search Prefers Product Pages Over Blogs",
          "host": "arobis",
          "url": "https://arobis.ai/blog/why-ai-search-engines-prefer-your-product-and-landing-pages",
          "position": 5
        },
        {
          "title": "How AI Search Engines Choose Sources to Cite",
          "host": "rankray",
          "url": "https://rankray.com/blog/how-chatgpt-ai-search-engines-choose-sources/",
          "position": 6
        },
        {
          "title": "Best Practices: Optimize Product Docs for AI Search & User Queries",
          "host": "geneo",
          "url": "https://geneo.app/blog/optimize-product-documentation-for-ai-search-best-practices/",
          "position": 7
        },
        {
          "title": "qiaomu-seo/references/ai-search.md at main · joeseesun/qiaomu-seo",
          "host": "github",
          "url": "https://github.com/joeseesun/qiaomu-seo/blob/main/references/ai-search.md",
          "position": 8
        },
        {
          "title": "Technical SEO for Docs That AI Can Discover",
          "host": "skayle",
          "url": "https://skayle.ai/blog/technical-seo-for-ai-discovery",
          "position": 9
        },
        {
          "title": "Your docs are now read by machines",
          "host": "joelolawanle",
          "url": "https://joelolawanle.com/blog/ai-readable-documentation",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-source-freshness",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-source-freshness",
      "published_at": "2026-11-15T12:58:00.000Z",
      "updated_at": "2026-11-15T12:58:00.000Z",
      "revision": 1,
      "slug": "perplexity-source-freshness",
      "panelId": "source-freshness",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-source-freshness",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "How do AI search engines decide whether information is fresh enough?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 20:58 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/20b108ac-07f4-422b-82d3-eb526af038bf",
      "mentions": "Cited",
      "summary": "AI search engines judge freshness by combining when a page was published or updated, whether its facts and terminology still match current reality, and whether the engine has recently crawled and indexed it. A new date alone isn’t proof the content is current—or that the engine has seen the update.",
      "answerSnapshot": "10秒\n\nAI search engines judge freshness by combining when a page was published or updated, whether its facts and terminology still match current reality, and whether the engine has recently crawled and indexed it. A new date alone isn’t proof the content is current—or that the engine has seen the update.\nsearchable\n+2\n\nHow much freshness matters depends on the question: current prices or “latest” updates call for recent sources, while stable topics can rely on older material.\nkime\n+1",
      "citations": [
        {
          "title": "Content Freshness Signals: How AI Determines If Your ... - Searchable",
          "host": "searchable",
          "url": "https://www.searchable.com/blog/content-freshness-signals",
          "position": 1
        },
        {
          "title": "When Content Freshness Affects AI Search Citations",
          "host": "deepsmith",
          "url": "https://deepsmith.ai/blog/content-freshness-ai-search-citations",
          "position": 2
        },
        {
          "title": "Maintain content freshness for AI search",
          "host": "siteimprove",
          "url": "https://www.siteimprove.com/blog/maintaining-content-freshness-for-ai-search/",
          "position": 3
        },
        {
          "title": "What Is Content Freshness? Why It Matters for AI | KIME",
          "host": "kime",
          "url": "https://kime.ai/glossary/content-freshness",
          "position": 4
        },
        {
          "title": "How Fresh Is the Data Returned by Search APIs?",
          "host": "olostep",
          "url": "https://www.olostep.com/glossary/agentic-search-apis/how-fresh-data-search-apis",
          "position": 5
        },
        {
          "title": "Source Freshness: Check Dates in a Live Web Search ...",
          "host": "tinyfish",
          "url": "https://www.tinyfish.ai/blog/source-freshness-live-web-search",
          "position": 6
        },
        {
          "title": "Does Content Freshness Influence AI Search? [Data Insights] - OMNIUS",
          "host": "omnius",
          "url": "https://www.omnius.so/blog/does-content-freshness-influence-ai-search",
          "position": 7
        },
        {
          "title": "Content freshness signals for generative AI systems",
          "host": "spawned",
          "url": "https://spawned.com/learn/content-freshness-signals-for-generative-ai-systems",
          "position": 8
        },
        {
          "title": "Content Freshness Signals That AI Search Engines Look For | AY ...",
          "host": "ayrank",
          "url": "https://www.ayrank.com/blog/content-freshness-signals-ai-search",
          "position": 9
        },
        {
          "title": "What's the role of freshness in keeping information visible in AI ...",
          "host": "senso",
          "url": "https://www.senso.ai/prompts-content/whats-the-role-of-freshness-in-keeping-information-visible-in-ai-search",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-chatgpt-search-source-selection-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-chatgpt-search-source-selection-repeat-3",
      "published_at": "2026-11-15T12:57:00.000Z",
      "updated_at": "2026-11-15T12:57:00.000Z",
      "revision": 1,
      "slug": "perplexity-chatgpt-search-source-selection-repeat-3",
      "panelId": "chatgpt-search-source-selection",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-chatgpt-search-source-selection",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "How does ChatGPT Search choose the sources it cites?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 20:57 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/cd3fc61a-75b6-4c52-bfca-a892f5bcd7b9",
      "mentions": "Cited",
      "summary": "ChatGPT Search rewrites your question into one or more web searches, retrieves relevant pages, and may run follow-up searches. It then cites sources that support specific parts of its answer; OpenAI notes it may use other search providers, but doesn’t publish a detailed ranking formula.",
      "answerSnapshot": "9秒\n\nChatGPT Search rewrites your question into one or more web searches, retrieves relevant pages, and may run follow-up searches. It then cites sources that support specific parts of its answer; OpenAI notes it may use other search providers, but doesn’t publish a detailed ranking formula.\nhelp.openai",
      "citations": [
        {
          "title": "Searching the web with ChatGPT | OpenAI Help Center",
          "host": "help.openai",
          "url": "https://help.openai.com/en/articles/9237897-publishers-and-developers-faq",
          "position": 1
        },
        {
          "title": "help.openai.com › en › articlesSearching the web with ChatGPT - OpenAI Help Center",
          "host": "help.openai",
          "url": "https://help.openai.com/en/articles/9237897-conducting-your-searches-on-searchgpt%253F.doc",
          "position": 2
        },
        {
          "title": "help.openai.com · en · articlesSearching the web with ChatGPT | OpenAI Help Center",
          "host": "help.openai",
          "url": "https://help.openai.com/en/articles/9237897-chatgpt-search%25253F.gz",
          "position": 3
        },
        {
          "title": "Searching the web with ChatGPT | OpenAI Help Center",
          "host": "help.openai",
          "url": "https://help.openai.com/articles/9237897-chatgpt-search",
          "position": 4
        },
        {
          "title": "ChatGPT search - OpenAI Help Center",
          "host": "help.openai",
          "url": "https://help.openai.com/en/articles/9237897-chatgpt-search%25252525252525252525252525252525252525252523.pict",
          "position": 5
        },
        {
          "title": "How ChatGPT Actually Chooses Sources: A Technical Breakdown",
          "host": "neuronwriter",
          "url": "https://neuronwriter.com/how-chatgpt-chooses-sources/",
          "position": 6
        },
        {
          "title": "How ChatGPT Chooses Sources and Brands to Cite (2026)",
          "host": "createandgrow",
          "url": "https://createandgrow.com/how-chatgpt-chooses-sources/",
          "position": 7
        },
        {
          "title": "How does ChatGPT choose its sources when browsing the ...",
          "host": "zapier",
          "url": "https://zapier.com/blog/chatgpt-sources/",
          "position": 8
        },
        {
          "title": "Inside the source choices behind ChatGPT search results",
          "host": "webiano",
          "url": "https://webiano.digital/inside-the-source-choices-behind-chatgpt-search-results/",
          "position": 9
        },
        {
          "title": "How ChatGPT Chooses Its Sources for Citations - vryse.co",
          "host": "vryse",
          "url": "https://vryse.co/blog/how-chatgpt-chooses-its-sources",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-google-ai-mode-source-selection",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-google-ai-mode-source-selection",
      "published_at": "2026-11-15T12:57:00.000Z",
      "updated_at": "2026-11-15T12:57:00.000Z",
      "revision": 1,
      "slug": "perplexity-google-ai-mode-source-selection",
      "panelId": "google-ai-mode-source-selection",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-google-ai-mode-source-selection",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "How does Google AI Mode choose the sources it cites?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 20:57 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/e0872219-47e2-4494-8198-7e8707a753eb",
      "mentions": "Cited",
      "summary": "Google AI Mode expands your question into related searches across its subtopics (“query fan-out”), then uses results from Google’s Search index to build an answer and link to pages that support it. A page generally needs to be indexed and eligible for a search snippet to be cited; Google hasn’t published a separate formula for choosing citations.",
      "answerSnapshot": "7秒\n\nGoogle AI Mode expands your question into related searches across its subtopics (“query fan-out”), then uses results from Google’s Search index to build an answer and link to pages that support it. A page generally needs to be indexed and eligible for a search snippet to be cited; Google hasn’t published a separate formula for choosing citations.\ntrovance\n+1",
      "citations": [
        {
          "title": "How Google AI Mode Picks Its Sources",
          "host": "trovance",
          "url": "https://trovance.ai/resources/google-ai-mode-sources/",
          "position": 1
        },
        {
          "title": "How AI Overviews Work: How Google Picks Sources",
          "host": "lemniscategrowth",
          "url": "https://lemniscategrowth.com/blogs/how-ai-overviews-work.html",
          "position": 2
        },
        {
          "title": "How Google AI Overviews Choose Sources (2026)",
          "host": "createandgrow",
          "url": "https://createandgrow.com/how-google-ai-overviews-choose-sources/",
          "position": 3
        },
        {
          "title": "How AI Overviews Choose Sources: What Google's Patents Say",
          "host": "omnius",
          "url": "https://www.omnius.so/blog/google-ai-search-patents",
          "position": 4
        },
        {
          "title": "How Does AI Search Work? AI Citation Mechanics Explained",
          "host": "straightnorth",
          "url": "https://www.straightnorth.com/blog/how-do-ai-search-engines-choose-which-sources-to-cite/",
          "position": 5
        },
        {
          "title": "Google AI Mode: what it changes for your site | José Galán",
          "host": "josegalan",
          "url": "https://josegalan.dev/en/news/google-ai-mode",
          "position": 6
        },
        {
          "title": "How AI Engines Choose Sources to Cite: 2026 Guide",
          "host": "searchcounselco",
          "url": "https://searchcounselco.com/how-ai-engines-choose-sources/",
          "position": 7
        },
        {
          "title": "How Google AI Mode Chooses Passages: The Signals Behind Its Citations",
          "host": "szymaniakdigital.co",
          "url": "https://szymaniakdigital.co.uk/how-google-ai-mode-chooses-passages/",
          "position": 8
        },
        {
          "title": "Google AI Mode: How It Chooses Which Sites to Cite (2026)",
          "host": "aimode",
          "url": "https://aimode.my/how-google-ai-mode-chooses-citations/",
          "position": 9
        },
        {
          "title": "How Google AI Overviews and AI Mode choose sources",
          "host": "elementera",
          "url": "https://www.elementera.com/blog/google-ai-overviews-ai-mode-sources",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-google-ai-mode-source-selection-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-google-ai-mode-source-selection-repeat-2",
      "published_at": "2026-11-15T12:57:00.000Z",
      "updated_at": "2026-11-15T12:57:00.000Z",
      "revision": 1,
      "slug": "perplexity-google-ai-mode-source-selection-repeat-2",
      "panelId": "google-ai-mode-source-selection",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-google-ai-mode-source-selection",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "How does Google AI Mode choose the sources it cites?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 20:57 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/2336da31-83be-45a7-b699-33041e610d78",
      "mentions": "Cited",
      "summary": "Google AI Mode uses Google Search systems and “query fan-out”: it breaks your question into related searches, retrieves relevant pages, then cites pages that support parts of its answer. A page generally needs to be indexed and eligible for a search snippet; Google hasn’t published a separate, exact formula for choosing citations.",
      "answerSnapshot": "9秒\n\nGoogle AI Mode uses Google Search systems and “query fan-out”: it breaks your question into related searches, retrieves relevant pages, then cites pages that support parts of its answer. A page generally needs to be indexed and eligible for a search snippet; Google hasn’t published a separate, exact formula for choosing citations.\ntrovance\n+1",
      "citations": [
        {
          "title": "How Google AI Mode Picks Its Sources",
          "host": "trovance",
          "url": "https://trovance.ai/resources/google-ai-mode-sources/",
          "position": 1
        },
        {
          "title": "How AI Overviews Work: How Google Picks Sources",
          "host": "lemniscategrowth",
          "url": "https://lemniscategrowth.com/blogs/how-ai-overviews-work.html",
          "position": 2
        },
        {
          "title": "How Google AI Overviews Choose Sources (2026)",
          "host": "createandgrow",
          "url": "https://createandgrow.com/how-google-ai-overviews-choose-sources/",
          "position": 3
        },
        {
          "title": "How AI Overviews Choose Sources: What Google's Patents Say",
          "host": "omnius",
          "url": "https://www.omnius.so/blog/google-ai-search-patents",
          "position": 4
        },
        {
          "title": "How Does AI Search Work? AI Citation Mechanics Explained",
          "host": "straightnorth",
          "url": "https://www.straightnorth.com/blog/how-do-ai-search-engines-choose-which-sources-to-cite/",
          "position": 5
        },
        {
          "title": "Google AI Mode: what it changes for your site | José Galán",
          "host": "josegalan",
          "url": "https://josegalan.dev/en/news/google-ai-mode",
          "position": 6
        },
        {
          "title": "How AI Engines Choose Sources to Cite: 2026 Guide",
          "host": "searchcounselco",
          "url": "https://searchcounselco.com/how-ai-engines-choose-sources/",
          "position": 7
        },
        {
          "title": "How Google AI Mode Chooses Passages: The Signals Behind Its Citations",
          "host": "szymaniakdigital.co",
          "url": "https://szymaniakdigital.co.uk/how-google-ai-mode-chooses-passages/",
          "position": 8
        },
        {
          "title": "Google AI Mode: How It Chooses Which Sites to Cite (2026)",
          "host": "aimode",
          "url": "https://aimode.my/how-google-ai-mode-chooses-citations/",
          "position": 9
        },
        {
          "title": "How Google AI Overviews and AI Mode choose sources",
          "host": "elementera",
          "url": "https://www.elementera.com/blog/google-ai-overviews-ai-mode-sources",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-google-ai-mode-source-selection-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-google-ai-mode-source-selection-repeat-3",
      "published_at": "2026-11-15T12:57:00.000Z",
      "updated_at": "2026-11-15T12:57:00.000Z",
      "revision": 1,
      "slug": "perplexity-google-ai-mode-source-selection-repeat-3",
      "panelId": "google-ai-mode-source-selection",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-google-ai-mode-source-selection",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "How does Google AI Mode choose the sources it cites?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 20:57 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/8be8f21e-dce3-47bd-a9b9-48ca69efea09",
      "mentions": "Cited",
      "summary": "Google AI Mode breaks your question into related subtopics and runs multiple searches (“query fan-out”), then uses the results to build an answer and link to pages that support it.",
      "answerSnapshot": "7秒\n\nGoogle AI Mode breaks your question into related subtopics and runs multiple searches (“query fan-out”), then uses the results to build an answer and link to pages that support it. A page generally needs to be indexed and eligible to appear with a Search snippet; Google hasn’t published a separate formula for choosing citations, so being highly ranked doesn’t guarantee a citation.\ntrovance\n+1",
      "citations": [
        {
          "title": "How Google AI Mode Picks Its Sources",
          "host": "trovance",
          "url": "https://trovance.ai/resources/google-ai-mode-sources/",
          "position": 1
        },
        {
          "title": "How AI Overviews Work: How Google Picks Sources",
          "host": "lemniscategrowth",
          "url": "https://lemniscategrowth.com/blogs/how-ai-overviews-work.html",
          "position": 2
        },
        {
          "title": "How Google AI Overviews Choose Sources (2026)",
          "host": "createandgrow",
          "url": "https://createandgrow.com/how-google-ai-overviews-choose-sources/",
          "position": 3
        },
        {
          "title": "How AI Overviews Choose Sources: What Google's Patents Say",
          "host": "omnius",
          "url": "https://www.omnius.so/blog/google-ai-search-patents",
          "position": 4
        },
        {
          "title": "How Does AI Search Work? AI Citation Mechanics Explained",
          "host": "straightnorth",
          "url": "https://www.straightnorth.com/blog/how-do-ai-search-engines-choose-which-sources-to-cite/",
          "position": 5
        },
        {
          "title": "Google AI Mode: what it changes for your site | José Galán",
          "host": "josegalan",
          "url": "https://josegalan.dev/en/news/google-ai-mode",
          "position": 6
        },
        {
          "title": "How AI Engines Choose Sources to Cite: 2026 Guide",
          "host": "searchcounselco",
          "url": "https://searchcounselco.com/how-ai-engines-choose-sources/",
          "position": 7
        },
        {
          "title": "How Google AI Mode Chooses Passages: The Signals Behind Its Citations",
          "host": "szymaniakdigital.co",
          "url": "https://szymaniakdigital.co.uk/how-google-ai-mode-chooses-passages/",
          "position": 8
        },
        {
          "title": "Google AI Mode: How It Chooses Which Sites to Cite (2026)",
          "host": "aimode",
          "url": "https://aimode.my/how-google-ai-mode-chooses-citations/",
          "position": 9
        },
        {
          "title": "How Google AI Overviews and AI Mode choose sources",
          "host": "elementera",
          "url": "https://www.elementera.com/blog/google-ai-overviews-ai-mode-sources",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-chatgpt-search-source-selection",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-chatgpt-search-source-selection",
      "published_at": "2026-11-15T12:56:00.000Z",
      "updated_at": "2026-11-15T12:56:00.000Z",
      "revision": 1,
      "slug": "perplexity-chatgpt-search-source-selection",
      "panelId": "chatgpt-search-source-selection",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-chatgpt-search-source-selection",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "How does ChatGPT Search choose the sources it cites?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 20:56 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/701d5bca-f9fb-4ee0-8e05-3b0b0e6698c1",
      "mentions": "Cited",
      "summary": "ChatGPT Search rewrites your question into one or more searches, retrieves and reviews relevant pages, and may run follow-up searches before composing an answer. It cites sources that support specific claims in the answer; the exact ranking and selection criteria aren’t fully detailed in OpenAI’s public explanation.",
      "answerSnapshot": "10秒\n\nChatGPT Search rewrites your question into one or more searches, retrieves and reviews relevant pages, and may run follow-up searches before composing an answer. It cites sources that support specific claims in the answer; the exact ranking and selection criteria aren’t fully detailed in OpenAI’s public explanation.\nhelp.openai",
      "citations": [
        {
          "title": "Searching the web with ChatGPT | OpenAI Help Center",
          "host": "help.openai",
          "url": "https://help.openai.com/en/articles/9237897-publishers-and-developers-faq",
          "position": 1
        },
        {
          "title": "help.openai.com › en › articlesSearching the web with ChatGPT - OpenAI Help Center",
          "host": "help.openai",
          "url": "https://help.openai.com/en/articles/9237897-conducting-your-searches-on-searchgpt%253F.doc",
          "position": 2
        },
        {
          "title": "help.openai.com · en · articlesSearching the web with ChatGPT | OpenAI Help Center",
          "host": "help.openai",
          "url": "https://help.openai.com/en/articles/9237897-chatgpt-search%25253F.gz",
          "position": 3
        },
        {
          "title": "Searching the web with ChatGPT | OpenAI Help Center",
          "host": "help.openai",
          "url": "https://help.openai.com/articles/9237897-chatgpt-search",
          "position": 4
        },
        {
          "title": "ChatGPT search - OpenAI Help Center",
          "host": "help.openai",
          "url": "https://help.openai.com/en/articles/9237897-chatgpt-search%25252525252525252525252525252525252525252523.pict",
          "position": 5
        },
        {
          "title": "How ChatGPT Actually Chooses Sources: A Technical Breakdown",
          "host": "neuronwriter",
          "url": "https://neuronwriter.com/how-chatgpt-chooses-sources/",
          "position": 6
        },
        {
          "title": "How ChatGPT Chooses Sources and Brands to Cite (2026)",
          "host": "createandgrow",
          "url": "https://createandgrow.com/how-chatgpt-chooses-sources/",
          "position": 7
        },
        {
          "title": "How does ChatGPT choose its sources when browsing the ...",
          "host": "zapier",
          "url": "https://zapier.com/blog/chatgpt-sources/",
          "position": 8
        },
        {
          "title": "Inside the source choices behind ChatGPT search results",
          "host": "webiano",
          "url": "https://webiano.digital/inside-the-source-choices-behind-chatgpt-search-results/",
          "position": 9
        },
        {
          "title": "How ChatGPT Chooses Its Sources for Citations - vryse.co",
          "host": "vryse",
          "url": "https://vryse.co/blog/how-chatgpt-chooses-its-sources",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-chatgpt-search-source-selection-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-chatgpt-search-source-selection-repeat-2",
      "published_at": "2026-11-15T12:56:00.000Z",
      "updated_at": "2026-11-15T12:56:00.000Z",
      "revision": 1,
      "slug": "perplexity-chatgpt-search-source-selection-repeat-2",
      "panelId": "chatgpt-search-source-selection",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-chatgpt-search-source-selection",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "How does ChatGPT Search choose the sources it cites?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 20:56 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/24dd78a8-0f3a-4cd0-9dfb-222d59f88974",
      "mentions": "Cited",
      "summary": "ChatGPT Search turns your question into one or more web searches, reviews relevant results, and may run follow-up searches. It cites sources that support specific parts of its answer; a citation is meant to let you check where that information came from.",
      "answerSnapshot": "9秒\n\nChatGPT Search turns your question into one or more web searches, reviews relevant results, and may run follow-up searches. It cites sources that support specific parts of its answer; a citation is meant to let you check where that information came from.\nhelp.openai",
      "citations": [
        {
          "title": "Searching the web with ChatGPT | OpenAI Help Center",
          "host": "help.openai",
          "url": "https://help.openai.com/en/articles/9237897-publishers-and-developers-faq",
          "position": 1
        },
        {
          "title": "help.openai.com › en › articlesSearching the web with ChatGPT - OpenAI Help Center",
          "host": "help.openai",
          "url": "https://help.openai.com/en/articles/9237897-conducting-your-searches-on-searchgpt%253F.doc",
          "position": 2
        },
        {
          "title": "help.openai.com · en · articlesSearching the web with ChatGPT | OpenAI Help Center",
          "host": "help.openai",
          "url": "https://help.openai.com/en/articles/9237897-chatgpt-search%25253F.gz",
          "position": 3
        },
        {
          "title": "Searching the web with ChatGPT | OpenAI Help Center",
          "host": "help.openai",
          "url": "https://help.openai.com/articles/9237897-chatgpt-search",
          "position": 4
        },
        {
          "title": "ChatGPT search - OpenAI Help Center",
          "host": "help.openai",
          "url": "https://help.openai.com/en/articles/9237897-chatgpt-search%25252525252525252525252525252525252525252523.pict",
          "position": 5
        },
        {
          "title": "How ChatGPT Actually Chooses Sources: A Technical Breakdown",
          "host": "neuronwriter",
          "url": "https://neuronwriter.com/how-chatgpt-chooses-sources/",
          "position": 6
        },
        {
          "title": "How ChatGPT Chooses Sources and Brands to Cite (2026)",
          "host": "createandgrow",
          "url": "https://createandgrow.com/how-chatgpt-chooses-sources/",
          "position": 7
        },
        {
          "title": "How does ChatGPT choose its sources when browsing the ...",
          "host": "zapier",
          "url": "https://zapier.com/blog/chatgpt-sources/",
          "position": 8
        },
        {
          "title": "Inside the source choices behind ChatGPT search results",
          "host": "webiano",
          "url": "https://webiano.digital/inside-the-source-choices-behind-chatgpt-search-results/",
          "position": 9
        },
        {
          "title": "How ChatGPT Chooses Its Sources for Citations - vryse.co",
          "host": "vryse",
          "url": "https://vryse.co/blog/how-chatgpt-chooses-its-sources",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-get-cited-perplexity",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-get-cited-perplexity",
      "published_at": "2026-11-15T12:56:00.000Z",
      "updated_at": "2026-11-15T12:56:00.000Z",
      "revision": 1,
      "slug": "perplexity-get-cited-perplexity",
      "panelId": "get-cited-perplexity",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-get-cited-perplexity",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "How can a new website get cited by Perplexity?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 20:56 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/f05f2ab1-3794-477f-9af5-1aa3586d12f7",
      "mentions": "Cited",
      "summary": "A new site can get cited by Perplexity, but there is no submission form or guaranteed method. The practical formula is: make the site crawlable, make individual pages easy to quote, establish trust, and earn mentions on sources Perplexity already uses.",
      "answerSnapshot": "A new site can get cited by Perplexity, but there is no submission form or guaranteed method. The practical formula is: make the site crawlable, make individual pages easy to quote, establish trust, and earn mentions on sources Perplexity already uses.\ncrawltoclick\n+1\n\nMake it crawlable\n\nEnsure robots.txt does not block PerplexityBot (and Perplexity-User, where relevant); also check Cloudflare, WAF, bot-management, and firewall settings, since a robots.txt rule alone does not guarantee crawler access.\nanobee\n+2\n\nKeep an accurate XML sitemap with valid lastmod dates, and use IndexNow / Bing Webmaster Tools to help discovery of new URLs.\ncrawltoclick\n+2\n\nPut essential content in server-rendered HTML: prices, specs, FAQs, tables, and dates should not live only in JavaScript, images, or PDFs.\ncrawltoclick\n+1\n\nVerify that key pages return a 200 response and readable HTML when fetched with the PerplexityBot user agent.\npavadotech\n+1\n\nBuild quotable pages\n\nPerplexity tends to cite pages that directly and cleanly answer a specific question. Lead each important page with a concise answer in the first 100 words, then support it with headings, short paragraphs, lists, tables, and named sources.\npavadotech\n+2\n\nFor example, instead of a vague “Why choose us” page, create a page titled “What does [service] cost in [location]?” that opens with a direct answer, shows a pricing table, explains assumptions, and links to primary evidence.\n\nEstablish trust\n\nUse consistent brand, product, author, and business information across your site, directories, LinkedIn, and profiles; add appropriate Organization/Person/Product schema where relevant.\nsearchmaxxed\n+1\n\nPublish original data, research, comparisons, or useful tools, and cite external sources for factual claims.\ncrawltoclick\n+1\n\nShow an honest “last updated” date and refresh cornerstone pages when facts change.\ncrawltoclick\n+1\n\nBuild conventional SEO foundations—clear titles, internal links, topical clusters, and backlinks—because strong Google/Bing visibility increases the chance Perplexity’s retrieval finds and selects your page.\nairanked\n+1\n\nEarn external mentions\n\nYour own domain is only part of the equation. Perplexity often draws on third-party sources such as reviews, directories, forums, YouTube, press coverage, and community discussions, so being mentioned there can make your brand easier to retrieve and validate.\ncrawltoclick\n+2\n\nGet listed in relevant, reputable directories and “best of” roundups.\n\nEncourage genuine customer reviews.\n\nContribute useful, non-promotional answers in relevant communities.\n\nPublish short explanatory videos, case studies, and PR-worthy original research.\npavadotech\n+1\n\nMeasure it\n\nCreate a fixed list of 20–50 buyer questions, ask them in Perplexity periodically, and log whether your site appears in the numbered sources, which page is cited, and which competitors are shown. This tells you which topics need better pages, fresher evidence, or more external validation.\npavadotech\n+1\n\nThe fastest realistic path for a new site is usually a narrow, high-intent niche: publish a genuinely useful comparison or answer page, make it crawlable and evidence-rich, then earn a few credible mentions around it.\nanobee\n+1",
      "citations": [
        {
          "title": "How to Get Your Website Cited by Perplexity | Crawl To Click",
          "host": "crawltoclick",
          "url": "https://crawltoclick.com/perplexity-seo/",
          "position": 1
        },
        {
          "title": "How to rank in Perplexity and get cited in its answers",
          "host": "airanked",
          "url": "https://airanked.ai/guides/how-to-rank-in-perplexity",
          "position": 2
        },
        {
          "title": "7 Perplexity Seo Steps To...",
          "host": "anobee",
          "url": "https://anobee.com/seo/perplexity-seo",
          "position": 3
        },
        {
          "title": "How to Get Cited by Perplexity (2026) | Searchmaxxed",
          "host": "searchmaxxed",
          "url": "https://www.searchmaxxed.com/research/how-to-get-cited-by-perplexity",
          "position": 4
        },
        {
          "title": "How to Get Cited by Perplexity AI: 3 Gates",
          "host": "pavadotech",
          "url": "https://www.pavadotech.com/blog/how-to-get-cited-by-perplexity-ai",
          "position": 5
        },
        {
          "title": "Perplexity SEO: How to Get Cited by Perplexity 2026",
          "host": "quillly",
          "url": "https://quillly.com/blogs/perplexity-seo-get-cited",
          "position": 6
        },
        {
          "title": "How to rank on Perplexity: a citation checklist | Snoika",
          "host": "snoika",
          "url": "https://snoika.com/blog/how-to-rank-on-perplexity",
          "position": 7
        },
        {
          "title": "Perplexity SEO: how to get cited in Perplexity (2026) | AnswerRadar",
          "host": "answerradar",
          "url": "https://answerradar.net/guides/perplexity-seo",
          "position": 8
        },
        {
          "title": "How to Rank in Perplexity: Get Cited as a Source in 2026",
          "host": "sorank",
          "url": "https://www.sorank.com/blog-geo-seo/how-to-rank-in-perplexity",
          "position": 9
        },
        {
          "title": "How to get your website cited by ChatGPT, Perplexity and Google AI Mode",
          "host": "dev",
          "url": "https://dev.to/vishall/how-to-get-your-website-cited-by-chatgpt-perplexity-and-google-ai-mode-p82",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-get-cited-perplexity-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-get-cited-perplexity-repeat-2",
      "published_at": "2026-11-15T12:56:00.000Z",
      "updated_at": "2026-11-15T12:56:00.000Z",
      "revision": 1,
      "slug": "perplexity-get-cited-perplexity-repeat-2",
      "panelId": "get-cited-perplexity",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-get-cited-perplexity",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "How can a new website get cited by Perplexity?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 20:56 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/82065e53-d3c8-4c8c-989e-7a541937cd47",
      "mentions": "Cited",
      "summary": "A new website can earn Perplexity citations by making its pages crawlable, publishing clear and specific answers, and building corroborating mentions across the web. There is no direct “submit my site” button; visibility comes from strong retrieval signals plus content that Perplexity can confidently quote.",
      "answerSnapshot": "1秒\n\nA new website can earn Perplexity citations by making its pages crawlable, publishing clear and specific answers, and building corroborating mentions across the web. There is no direct “submit my site” button; visibility comes from strong retrieval signals plus content that Perplexity can confidently quote.\nairanked\n+1\n\nMake the site crawlable\n\nEnsure robots.txt does not block PerplexityBot, and check that firewalls, CDN rules, or bot-management tools are not blocking it. Cloudflare’s default AI-crawler settings can block new domains, so verify the CDN layer as well as robots.txt.\nanobee\n+1\n\nSubmit an accurate XML sitemap with current lastmod dates, and keep important pages available as server-rendered HTML rather than content hidden behind JavaScript, images, or PDFs.\ncrawltoclick\n+1\n\nUse clean URLs, descriptive titles, internal links, and structured data that matches the visible page content.\nsnoika\n+1\n\nWrite citation-ready pages\n\nPerplexity tends to cite pages that directly answer a narrow question in a self-contained passage. Put the direct answer near the top, then support it with specifics.\ncrawltoclick\n+2\n\nTarget real customer questions, such as “What is the best CRM for a 10-person agency?” or “How much does X cost in the UK?”\n\nBegin each page with a 1–2 sentence answer, then use question-shaped headings, short paragraphs, lists, and tables.\nsnoika\n+1\n\nInclude original data, named sources, prices, specifications, comparisons, and a visible “last updated” date.\ncrawltoclick\n+1\n\nAvoid vague “thought leadership” pages; create distinct pages for category, comparison, how-to, and product questions.\nanswerradar\n\nBuild external trust\n\nPerplexity’s answers often overlap with established search rankings and rely on third-party sources, so a new site should not rely only on its own domain.\nanobee\n+2\n\nEarn coverage or inclusion in relevant “best of” lists, directories, review sites, trade publications, and comparison pages that already appear for your topic.\nshila\n+1\n\nKeep your company description, product details, pricing, and positioning consistent across your website, LinkedIn, directories, and profiles.\nanswerradar\n+1\n\nContribute genuinely useful answers in relevant communities such as Reddit, Quora, YouTube, GitHub, or specialist forums—without turning them into advertisements.\nshila\n+1\n\nPursue reviews, customer references, and press mentions; these help establish the entity and corroborate claims.\nshila\n+1\n\nMeasure and iterate\n\nCreate a list of 20–50 questions your audience would ask, run them in Perplexity, and record whether your site is cited, which page was used, and which competitors or third-party sources appear. Re-check every few weeks, since results can vary between sessions.\nshila\n+1\n\nFor a brand-new site, prioritise a handful of highly specific pages over many thin articles. A focused page with a clear answer, verifiable evidence, current dates, and a few credible external mentions has a much better chance of being cited than a broad homepage.\nsnoika\n+2",
      "citations": [
        {
          "title": "How to rank in Perplexity and get cited in its answers",
          "host": "airanked",
          "url": "https://airanked.ai/guides/how-to-rank-in-perplexity",
          "position": 1
        },
        {
          "title": "Perplexity SEO: 8 Steps to Get Cited by Perplexity",
          "host": "shila",
          "url": "https://shila.so/blog/perplexity-seo",
          "position": 2
        },
        {
          "title": "7 Perplexity Seo Steps To...",
          "host": "anobee",
          "url": "https://anobee.com/seo/perplexity-seo",
          "position": 3
        },
        {
          "title": "How to Get Your Website Cited by Perplexity | Crawl To Click",
          "host": "crawltoclick",
          "url": "https://crawltoclick.com/perplexity-seo/",
          "position": 4
        },
        {
          "title": "How to rank on Perplexity: a citation checklist | Snoika",
          "host": "snoika",
          "url": "https://snoika.com/blog/how-to-rank-on-perplexity",
          "position": 5
        },
        {
          "title": "Perplexity SEO: how to get cited in Perplexity (2026) | AnswerRadar",
          "host": "answerradar",
          "url": "https://answerradar.net/guides/perplexity-seo",
          "position": 6
        },
        {
          "title": "Perplexity SEO: How to Get Cited by Perplexity 2026",
          "host": "quillly",
          "url": "https://quillly.com/blogs/perplexity-seo-get-cited",
          "position": 7
        },
        {
          "title": "How to Get Cited by Perplexity (2026) | Searchmaxxed",
          "host": "searchmaxxed",
          "url": "https://www.searchmaxxed.com/research/how-to-get-cited-by-perplexity",
          "position": 8
        },
        {
          "title": "How to get your website cited by ChatGPT, Perplexity and Google AI Mode",
          "host": "dev",
          "url": "https://dev.to/vishall/how-to-get-your-website-cited-by-chatgpt-perplexity-and-google-ai-mode-p82",
          "position": 9
        },
        {
          "title": "How to Rank in Perplexity: Get Cited as a Source in 2026",
          "host": "sorank",
          "url": "https://www.sorank.com/blog-geo-seo/how-to-rank-in-perplexity",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-get-cited-perplexity-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-get-cited-perplexity-repeat-3",
      "published_at": "2026-11-15T12:56:00.000Z",
      "updated_at": "2026-11-15T12:56:00.000Z",
      "revision": 1,
      "slug": "perplexity-get-cited-perplexity-repeat-3",
      "panelId": "get-cited-perplexity",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-get-cited-perplexity",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "How can a new website get cited by Perplexity?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 20:56 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/e9449018-f045-4483-8084-de7daa6d2700",
      "mentions": "Cited",
      "summary": "A new website can get cited by Perplexity by making its pages crawlable, publishing clear answer-first content, and earning mentions on sources Perplexity already trusts. There is no direct URL-submission or webmaster console, so visibility comes from crawl access, strong search presence, and third-party corroboration.",
      "answerSnapshot": "1秒\n\nA new website can get cited by Perplexity by making its pages crawlable, publishing clear answer-first content, and earning mentions on sources Perplexity already trusts. There is no direct URL-submission or webmaster console, so visibility comes from crawl access, strong search presence, and third-party corroboration.\nairanked\n+1\n\nMake the site crawlable\n\nEnsure PerplexityBot is allowed in robots.txt; also check firewall, CDN, and bot-management settings so it receives a normal 200 response and full HTML rather than a challenge or JavaScript-only page.\nshila\n+2\n\nKeep an accurate XML sitemap with current lastmod dates, and make core information available in server-rendered HTML rather than images, PDFs, or click-to-load components.\ncrawltoclick\n+1\n\nIf you use Cloudflare, verify that AI crawlers are not blocked by default; new domains may have this setting enabled.\nanobee\n\nPublish quotable pages\n\nCreate individual pages for the specific questions your audience asks—for example, “best CRM for small agencies,” “how much does X cost,” or “how does Y compare with Z.” Put a direct, self-contained answer in the opening section, then support it with headings, short paragraphs, tables, FAQs, and named sources.\ncrawltoclick\n+2\n\nLead with the answer: state the recommendation, definition, price, or process in the first 100 words.\n\nUse question-style headings that mirror real searches.\n\nInclude original data, transparent methodology, author credentials, and a visible update date.\n\nKeep facts consistent across your site, business profiles, directories, and product pages.\nsnoika\n+2\n\nBuild external trust\n\nPerplexity’s retrieval overlaps strongly with high-ranking Google results, so conventional SEO remains foundational: indexable pages, useful internal links, quality backlinks, and rankings for relevant queries.\nanobee\n+2\n\nAlso pursue citations and mentions on the pages that already appear for your target questions:\n\nGet listed in relevant directories and “best of” roundups, with consistent company descriptions and genuine reviews.\nshila\n+2\n\nContribute useful, non-promotional answers in Reddit, Quora, Stack Overflow, and niche communities where appropriate.\nshila\n+1\n\nPublish helpful public material on YouTube, GitHub, LinkedIn, or industry publications that can be discovered and referenced.\ndev\n\nPitch publishers with a concrete angle—an original dataset, useful comparison, local insight, or missing use case—rather than asking simply to be added.\nshila\n\nMeasure and iterate\n\nMake a list of 20–50 real customer questions, run them in Perplexity from a fresh session, and log whether your site is cited, which page is used, its position among sources, and which competitors or third-party pages appear. Recheck every two to four weeks, since answers and cited sources can vary between runs.\nshila\n+1\n\nFor a new site, prioritize a small set of narrowly useful pages over broad generic posts: a precise comparison, a practical guide, or an original data-backed resource is more likely to be selected than a thin overview. Fresh, specific content can begin appearing in citations after crawling, but off-site trust typically takes weeks to months to build.\nshila\n+1",
      "citations": [
        {
          "title": "How to rank in Perplexity and get cited in its answers",
          "host": "airanked",
          "url": "https://airanked.ai/guides/how-to-rank-in-perplexity",
          "position": 1
        },
        {
          "title": "Perplexity SEO: 8 Steps to Get Cited by Perplexity",
          "host": "shila",
          "url": "https://shila.so/blog/perplexity-seo",
          "position": 2
        },
        {
          "title": "7 Perplexity Seo Steps To...",
          "host": "anobee",
          "url": "https://anobee.com/seo/perplexity-seo",
          "position": 3
        },
        {
          "title": "How to Get Your Website Cited by Perplexity | Crawl To Click",
          "host": "crawltoclick",
          "url": "https://crawltoclick.com/perplexity-seo/",
          "position": 4
        },
        {
          "title": "How to rank on Perplexity: a citation checklist | Snoika",
          "host": "snoika",
          "url": "https://snoika.com/blog/how-to-rank-on-perplexity",
          "position": 5
        },
        {
          "title": "How to Get Cited by Perplexity (2026) | Searchmaxxed",
          "host": "searchmaxxed",
          "url": "https://www.searchmaxxed.com/research/how-to-get-cited-by-perplexity",
          "position": 6
        },
        {
          "title": "Perplexity SEO: how to get cited in Perplexity (2026) | AnswerRadar",
          "host": "answerradar",
          "url": "https://answerradar.net/guides/perplexity-seo",
          "position": 7
        },
        {
          "title": "How to Rank in Perplexity: Get Cited as a Source in 2026",
          "host": "sorank",
          "url": "https://www.sorank.com/blog-geo-seo/how-to-rank-in-perplexity",
          "position": 8
        },
        {
          "title": "How to get your website cited by ChatGPT, Perplexity and Google AI Mode",
          "host": "dev",
          "url": "https://dev.to/vishall/how-to-get-your-website-cited-by-chatgpt-perplexity-and-google-ai-mode-p82",
          "position": 9
        },
        {
          "title": "Perplexity SEO: How to Get Cited by Perplexity 2026",
          "host": "quillly",
          "url": "https://quillly.com/blogs/perplexity-seo-get-cited",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-get-cited-chatgpt",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-get-cited-chatgpt",
      "published_at": "2026-11-15T12:55:00.000Z",
      "updated_at": "2026-11-15T12:55:00.000Z",
      "revision": 1,
      "slug": "perplexity-get-cited-chatgpt",
      "panelId": "get-cited-chatgpt",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-get-cited-chatgpt",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "How can a new website get cited by ChatGPT?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 20:55 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/4c67cdca-e157-4fbf-a18c-bd462ed7c575",
      "mentions": "Cited",
      "summary": "A new website can get cited by ChatGPT by making its pages easy for AI search systems to access, index, and verify—then earning independent mentions on the sites ChatGPT already uses. There is no submission form or guaranteed citation; the goal is to become a clear, trustworthy source for specific questions.",
      "answerSnapshot": "A new website can get cited by ChatGPT by making its pages easy for AI search systems to access, index, and verify—then earning independent mentions on the sites ChatGPT already uses. There is no submission form or guaranteed citation; the goal is to become a clear, trustworthy source for specific questions.\nhostinger\n+2\n\nMake the site crawlable\n\nAllow OpenAI’s search crawler, OAI-SearchBot, in robots.txt; also ensure your firewall, CDN, or bot-protection service does not block OpenAI’s published crawler IP ranges.\nsearchable\n+1\n\nSubmit an XML sitemap to Google Search Console and Bing Webmaster Tools, then resolve indexing errors and remove any noindex tags from pages you want discovered.\nhostinger\n+1\n\nPut the important information in server-rendered HTML rather than only in JavaScript, images, PDFs, or gated content.\nhostinger\n+2\n\nPublish answer-first pages\n\nCreate one focused page for each real customer question—for example, “best marketing automation tools for small UK businesses” or “how to calculate customer acquisition cost.” Use a direct question as the H2, state the answer in the first 100 words, and support it with original data, named authors, publication/update dates, and links to primary sources.\nseowind\n+2\n\nFor a marketing site, useful citation-worthy assets include:\n\nOriginal benchmark data, surveys, or case-study results\n\nClear definitions and step-by-step guides\n\nTransparent comparison pages with explicit evaluation criteria\n\nDetailed service pages that explain who the service is for, what it includes, pricing logic, and proof\n\nFAQ sections that answer precise buyer questions\n\nAvoid generic, thin “AI-written” content; ChatGPT is more likely to cite a page that adds evidence or a distinctive, verifiable point of view.\nseowind\n+1\n\nBuild external proof\n\nChatGPT often cites sources that are corroborated elsewhere, so off-site visibility matters as much as on-site content. Identify the pages ChatGPT cites for your target prompts, then pursue relevant inclusion through directories, industry roundups, comparison articles, expert commentary, guest posts, partnerships, and review platforms.\nhostinger\n+2\n\nKeep your brand description consistent across your website, LinkedIn, Google Business Profile, directories, and press mentions. This helps systems associate your brand with the right topic and location.\nsatelo\n+1\n\nUse structured data\n\nAdd valid schema where it accurately describes the page: Organization for the business, Article for editorial content, FAQPage for genuine FAQs, and product/service schema where appropriate. Structured data does not guarantee a citation, but it can make entities, authorship, dates, and page purpose easier to understand.\nseomods\n+1\n\nMeasure and improve\n\nRun 15–20 prompts your ideal customers would realistically use in ChatGPT with web search enabled, and record which URLs are cited. Recheck monthly because citations can vary by prompt wording, location, and individual response; also monitor analytics for referrals from chatgpt.com and server logs for OAI-SearchBot.\nsearchable\n+1\n\nPriority\tWhat to do\tWhy it helps\nAccess\tAllow OAI-SearchBot; submit sitemap\tChatGPT must be able to find and read the page. \nhostinger\n+1\n\nContent\tAnswer one question directly, with evidence\tMakes the page easy to extract and cite. \nseowind\n+1\n\nTrust\tAdd author, dates, sources, and original data\tIncreases source credibility. \nseowind\n+1\n\nReputation\tEarn mentions on relevant third-party sites\tChatGPT often cites sources already recognized across the web. \nhostinger\n+1\n\nIteration\tTrack prompts and refresh important pages\tReveals which pages win citations and keeps facts current. \nsearchable\n+1\n\nFor a brand-new site, start with 5–10 high-intent questions, publish genuinely useful pages for them, get indexed, and spend the remaining effort on credible third-party coverage rather than trying to manipulate AI answers.\nhostinger\n+2",
      "citations": [
        {
          "title": "How to rank in ChatGPT in 6 steps",
          "host": "hostinger",
          "url": "https://www.hostinger.com/tutorials/how-to-rank-in-chatgpt/",
          "position": 1
        },
        {
          "title": "How to Get Cited by ChatGPT: 5 Steps to More Citations",
          "host": "searchable",
          "url": "https://www.searchable.com/pt/blog/how-to-get-cited-by-chatgpt",
          "position": 2
        },
        {
          "title": "How to Get Cited by ChatGPT: A Guide for Small Sites",
          "host": "satelo",
          "url": "https://satelo.io/blog/how-to-get-cited-by-chatgpt/",
          "position": 3
        },
        {
          "title": "How to Get Cited by ChatGPT: A Practical Checklist",
          "host": "seowind",
          "url": "https://seowind.io/how-to-get-cited-by-chatgpt/",
          "position": 4
        },
        {
          "title": "How to Appear in ChatGPT Search Results (2026)",
          "host": "seorav",
          "url": "https://www.seorav.com/blog/how-to-get-your-website-into-chatgpt-search-results",
          "position": 5
        },
        {
          "title": "How to Get Your Website Cited by ChatGPT (2026 Guide) — SeoMods",
          "host": "seomods",
          "url": "https://seomods.com/blog/how-to-get-cited-by-chatgpt",
          "position": 6
        },
        {
          "title": "How ChatGPT decides which websites to cite | SearchScore",
          "host": "searchscore",
          "url": "https://searchscore.io/guides/how-chatgpt-cites-websites/",
          "position": 7
        },
        {
          "title": "ChatGPT SEO: How to Get Your Website Cited by ChatGPT: 2026 ...",
          "host": "promptwatch",
          "url": "https://promptwatch.com/blog/chatgpt-seo",
          "position": 8
        },
        {
          "title": "How to get cited by ChatGPT: what it takes to rank in it, The Internet Ninja",
          "host": "theinternetninja",
          "url": "https://theinternetninja.com/guides/how-to-get-cited-by-chatgpt/",
          "position": 9
        },
        {
          "title": "7 Steps to Get Your Website Cited by ChatGPT in 2026",
          "host": "websiteaeogeochecker",
          "url": "https://websiteaeogeochecker.com/blog/how-to-get-website-cited-by-chatgpt",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-get-cited-chatgpt-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-get-cited-chatgpt-repeat-2",
      "published_at": "2026-11-15T12:55:00.000Z",
      "updated_at": "2026-11-15T12:55:00.000Z",
      "revision": 1,
      "slug": "perplexity-get-cited-chatgpt-repeat-2",
      "panelId": "get-cited-chatgpt",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-get-cited-chatgpt",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "How can a new website get cited by ChatGPT?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 20:55 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/440886c7-ca1c-459c-b964-be681d453d2b",
      "mentions": "Cited",
      "summary": "A new website can be cited by ChatGPT when it is technically crawlable, clearly answers a specific question, and is corroborated by independent sources. There is no submission form or guaranteed “citation tag,” so the goal is to make your site easy to discover, verify, and quote.",
      "answerSnapshot": "A new website can be cited by ChatGPT when it is technically crawlable, clearly answers a specific question, and is corroborated by independent sources. There is no submission form or guaranteed “citation tag,” so the goal is to make your site easy to discover, verify, and quote.\nsearchable\n+1\n\nMake it crawlable\n\nAllow OpenAI’s search crawler, OAI-SearchBot, in robots.txt; also ensure your CDN, firewall, or bot-protection service does not block it.\nsearchable\n+1\n\nRemove noindex tags from important pages, submit an XML sitemap, and verify the site in Google Search Console and Bing Webmaster Tools.\nhostinger\n+1\n\nPut the essential information in server-rendered HTML rather than only in JavaScript, images, gated content, or interactive widgets.\nseowind\n+1\n\nAnswer one question well\n\nCreate pages that directly answer the questions your target customers would ask ChatGPT—for example, “What is the best CRM for a small UK marketing agency?” or “How much does X cost in Maidenhead?”\n\nPut the direct answer in the opening section, then support it with evidence, examples, and sources.\nsearchable\n+1\n\nUse clear headings that mirror real questions, and keep each section focused on one question.\nsatelo\n+1\n\nPublish original value: a useful comparison, a practical guide, local market data, a calculator, or a well-sourced study. Generic content is less likely to earn a citation.\nsatelo\n+1\n\nLink every key statistic to a primary source, and state the date, geography, and methodology where relevant.\nseowind\n\nBuild external trust\n\nChatGPT often cites pages that are already referenced or corroborated across the web, especially authoritative industry, government, news, review, and comparison sources.\nhostinger\n+2\n\nIdentify the sites ChatGPT already cites for your target prompts, then pursue relevant coverage there: directories, review platforms, expert roundups, trade associations, guest articles, and journalist outreach.\nhostinger\n+1\n\nKeep your business name, category, location, and one-line description consistent across those third-party profiles.\nsatelo\n\nEarn genuine mentions and backlinks from reputable sites rather than buying links or creating low-value AI-generated pages.\nseorav\n\nTrack what works\n\nRun the same customer questions in ChatGPT with web search enabled, record which URLs are cited, and monitor whether your pages appear. Because answers and sources can vary by prompt, location, and run, track this repeatedly rather than treating one result as definitive.\nsearchable\n+1\n\nPriority\tAction\tWhy it matters\n1\tAllow OAI-SearchBot and submit sitemaps\tChatGPT cannot cite pages it cannot access. \nsearchable\n+1\n\n2\tCreate answer-first pages for real questions\tMakes the content easy to retrieve and quote. \nsearchable\n+1\n\n3\tAdd original evidence and clear sourcing\tIncreases credibility and quotability. \nseowind\n+1\n\n4\tEarn third-party coverage\tPlaces your brand in the sources ChatGPT already uses. \nhostinger\n+1\n\n5\tRefresh and monitor key pages\tKeeps facts current and reveals which pages actually earn citations. \nhostinger\n+1\n\nFor a new site, focus first on 10–20 high-intent questions, make each answer genuinely useful, and pursue a few high-quality mentions on sites already appearing in ChatGPT’s answers. That is more effective than publishing many thin pages or trying to optimize for a single “AI ranking” hack.\nsatelo\n+1",
      "citations": [
        {
          "title": "How to Get Cited by ChatGPT: 5 Steps to More Citations",
          "host": "searchable",
          "url": "https://www.searchable.com/pt/blog/how-to-get-cited-by-chatgpt",
          "position": 1
        },
        {
          "title": "How to Get Cited by ChatGPT: A Guide for Small Sites",
          "host": "satelo",
          "url": "https://satelo.io/blog/how-to-get-cited-by-chatgpt/",
          "position": 2
        },
        {
          "title": "How to Get Cited by ChatGPT: A Practical Checklist",
          "host": "seowind",
          "url": "https://seowind.io/how-to-get-cited-by-chatgpt/",
          "position": 3
        },
        {
          "title": "How to rank in ChatGPT in 6 steps",
          "host": "hostinger",
          "url": "https://www.hostinger.com/tutorials/how-to-rank-in-chatgpt/",
          "position": 4
        },
        {
          "title": "How ChatGPT decides which websites to cite | SearchScore",
          "host": "searchscore",
          "url": "https://searchscore.io/guides/how-chatgpt-cites-websites/",
          "position": 5
        },
        {
          "title": "How to Appear in ChatGPT Search Results (2026)",
          "host": "seorav",
          "url": "https://www.seorav.com/blog/how-to-get-your-website-into-chatgpt-search-results",
          "position": 6
        },
        {
          "title": "How to get cited by ChatGPT: what it takes to rank in it, The Internet Ninja",
          "host": "theinternetninja",
          "url": "https://theinternetninja.com/guides/how-to-get-cited-by-chatgpt/",
          "position": 7
        },
        {
          "title": "7 Steps to Get Your Website Cited by ChatGPT in 2026",
          "host": "websiteaeogeochecker",
          "url": "https://websiteaeogeochecker.com/blog/how-to-get-website-cited-by-chatgpt",
          "position": 8
        },
        {
          "title": "ChatGPT SEO: How to Get Your Website Cited by ChatGPT: 2026 ...",
          "host": "promptwatch",
          "url": "https://promptwatch.com/blog/chatgpt-seo",
          "position": 9
        },
        {
          "title": "How to Get Your Website Cited by ChatGPT (2026 Guide) — SeoMods",
          "host": "seomods",
          "url": "https://seomods.com/blog/how-to-get-cited-by-chatgpt",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-get-cited-chatgpt-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-get-cited-chatgpt-repeat-3",
      "published_at": "2026-11-15T12:55:00.000Z",
      "updated_at": "2026-11-15T12:55:00.000Z",
      "revision": 1,
      "slug": "perplexity-get-cited-chatgpt-repeat-3",
      "panelId": "get-cited-chatgpt",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-get-cited-chatgpt",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "How can a new website get cited by ChatGPT?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 20:55 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/e1dd7001-0d92-4622-a314-58f193bda509",
      "mentions": "Cited",
      "summary": "A new website can get cited by ChatGPT, but there is no submission form or guaranteed method. The practical goal is to make your pages discoverable, accessible, clearly answerful, and independently corroborated—then monitor the specific prompts where you want to appear.",
      "answerSnapshot": "A new website can get cited by ChatGPT, but there is no submission form or guaranteed method. The practical goal is to make your pages discoverable, accessible, clearly answerful, and independently corroborated—then monitor the specific prompts where you want to appear.\nhostinger\n+2\n\nMake the site crawlable\n\nAllow OpenAI’s search crawler, OAI-SearchBot, in robots.txt; also ensure your firewall, CDN, or bot-protection service does not block OpenAI’s published crawler IP ranges.\nhostinger\n+2\n\nAvoid noindex tags, login walls, and form gates on the pages you want cited.\nhostinger\n+1\n\nPut the essential answer in server-rendered HTML—not only in JavaScript, images, PDFs, or interactive components—so crawlers can read it.\nhostinger\n+1\n\nSubmit an XML sitemap to Google Search Console and Bing Webmaster Tools; Bing coverage is especially relevant because ChatGPT search can draw on search-engine results.\nhostinger\n+1\n\nBuild pages worth citing\n\nCreate one focused page for each important customer question, rather than burying the answer in a broad service page. A useful structure is:\n\nA direct answer in the opening paragraph.\n\nA clear H2/H3 that reflects the question.\n\nSpecific facts, definitions, comparisons, pricing or process details where relevant.\n\nOriginal data, case studies, expert commentary, or a genuinely useful comparison.\n\nA named author, publication/update date, and sources for every important claim.\nsearchable\n+2\n\nFor example, if you sell marketing automation, publish a page titled “How much does marketing automation cost for small UK businesses?” with transparent assumptions, a current price range, and a comparison table—not just “contact us for a quote.”\n\nEarn external proof\n\nChatGPT often cites sources that other credible sites already discuss, so third-party mentions matter—particularly for a new domain.\nhostinger\n+2\n\nGet listed in relevant industry directories and review platforms.\n\nPitch useful data, expert commentary, or corrections to trade publications and niche roundups.\n\nPursue guest articles, partnerships, and mentions from respected organisations in your field.\n\nKeep your company description, service area, and key claims consistent across your site, LinkedIn, directories, and profiles.\nhostinger\n+1\n\nTarget the right prompts\n\nIdentify 15–20 questions your ideal customers actually ask, then run them in ChatGPT with web search enabled. Record which sites are cited, especially comparison articles, directories, forums, and publications.\nhostinger\n+2\n\nThen work to earn a presence on those sources: complete directory profiles, contribute to relevant discussions, offer a useful data point to comparison-article authors, and create a better, more transparent version of the content they currently cite.\nhostinger\n+1\n\nMeasure and refresh\n\nThere is no stable “ranking” to check: ChatGPT can produce different answers and citations depending on the prompt, location, and run.\nsatelo\n\nRe-run your priority prompts monthly and log cited URLs.\n\nCheck analytics for referral traffic from chatgpt.com.\n\nReview server logs for OAI-SearchBot visits; if it never appears, revisit crawling and firewall settings.\nhostinger\n+1\n\nGenuinely update high-value pages when facts change—new figures, examples, and a visible updated date—rather than superficially republishing them.\nseomods\n+1\n\nWhat not to do\n\nDo not stuff pages with “AI-friendly” keywords, fabricate statistics, or create thin pages merely to target prompts. Clear structure, accurate sourcing, original value, and credible external mentions are more durable than tricks; llms.txt is optional and is not documented by OpenAI as a ranking input.\nseowind\n+2\n\nFor a new site, the fastest realistic path is usually: make key pages crawlable and indexable, answer a narrow question exceptionally well, then get cited or mentioned by the sites ChatGPT already cites for that question.\nhostinger\n+2",
      "citations": [
        {
          "title": "How to rank in ChatGPT in 6 steps",
          "host": "hostinger",
          "url": "https://www.hostinger.com/tutorials/how-to-rank-in-chatgpt/",
          "position": 1
        },
        {
          "title": "How to Get Cited by ChatGPT: 5 Steps to More Citations",
          "host": "searchable",
          "url": "https://www.searchable.com/pt/blog/how-to-get-cited-by-chatgpt",
          "position": 2
        },
        {
          "title": "How to Get Cited by ChatGPT: A Guide for Small Sites",
          "host": "satelo",
          "url": "https://satelo.io/blog/how-to-get-cited-by-chatgpt/",
          "position": 3
        },
        {
          "title": "How to Get Cited by ChatGPT: A Practical Checklist",
          "host": "seowind",
          "url": "https://seowind.io/how-to-get-cited-by-chatgpt/",
          "position": 4
        },
        {
          "title": "How to Appear in ChatGPT Search Results (2026)",
          "host": "seorav",
          "url": "https://www.seorav.com/blog/how-to-get-your-website-into-chatgpt-search-results",
          "position": 5
        },
        {
          "title": "How to Get Your Website Cited by ChatGPT (2026 Guide) — SeoMods",
          "host": "seomods",
          "url": "https://seomods.com/blog/how-to-get-cited-by-chatgpt",
          "position": 6
        },
        {
          "title": "How to get cited by ChatGPT: what it takes to rank in it, The Internet Ninja",
          "host": "theinternetninja",
          "url": "https://theinternetninja.com/guides/how-to-get-cited-by-chatgpt/",
          "position": 7
        },
        {
          "title": "ChatGPT SEO: How to Get Your Website Cited by ChatGPT: 2026 ...",
          "host": "promptwatch",
          "url": "https://promptwatch.com/blog/chatgpt-seo",
          "position": 8
        },
        {
          "title": "How ChatGPT decides which websites to cite | SearchScore",
          "host": "searchscore",
          "url": "https://searchscore.io/guides/how-chatgpt-cites-websites/",
          "position": 9
        },
        {
          "title": "7 Steps to Get Your Website Cited by ChatGPT in 2026",
          "host": "websiteaeogeochecker",
          "url": "https://websiteaeogeochecker.com/blog/how-to-get-website-cited-by-chatgpt",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-source-types-cited-most-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-source-types-cited-most-repeat-3",
      "published_at": "2026-11-15T12:55:00.000Z",
      "updated_at": "2026-11-15T12:55:00.000Z",
      "revision": 1,
      "slug": "perplexity-source-types-cited-most-repeat-3",
      "panelId": "source-types-cited-most",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-source-types-cited-most",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "Which source types do AI search engines cite most often?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 20:55 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/0983a15d-9a41-49a3-ac36-ae704c559ad1",
      "mentions": "Cited",
      "summary": "AI search engines most often cite official/brand-owned sites, news media, and editorial or niche expert content; user-generated content (UGC) such as Reddit, YouTube, and review platforms is also prominent—especially for Google’s AI surfaces. The exact mix varies substantially by engine and query type.",
      "answerSnapshot": "AI search engines most often cite official/brand-owned sites, news media, and editorial or niche expert content; user-generated content (UGC) such as Reddit, YouTube, and review platforms is also prominent—especially for Google’s AI surfaces. The exact mix varies substantially by engine and query type.\nsearchenginejournal\n+2\n\nMost-cited source types\nSource type\tTypical role\tRelative prominence\nOfficial / brand-owned sites\tProduct pages, documentation, pricing, policies, company information\tOften the largest single category; one controlled study found 34% for ChatGPT, 46% for Google AI surfaces, and 44% for Perplexity. \nbavior\n\nNews media\tCurrent events, reporting, research coverage, credibility signals\tMajor category across engines; ChatGPT tends to cite it especially heavily. \nsearchengineland\n+1\n\nEditorial, niche, and comparison content\tReviews, trade press, comparison pages, specialist guides, finance data\tA consistently important layer; BrightEdge found commercial/editorial sources were the broadest visibility lever across five AI engines. \nbrightedge\n\nUGC: Reddit, YouTube, LinkedIn, forums\tReal-world experience, discussion, how-to help, recommendations\tImportant but uneven: Google AI Overviews and Gemini use it more than ChatGPT; one large analysis found Reddit was the most-cited domain overall. \nsearchengineland\n+1\n\nReference / institutional sources\tWikipedia, government, academic, and standards bodies\tFrequently cited for factual and educational queries, though not uniformly dominant. \nsearchengineland\n+1\n\nReview and directory platforms\tG2, Yelp, Trustpilot, retailer listings, directories\tParticularly visible in recommendation, comparison, and purchase-related queries. \nsearchengineland\n+1\nDifferences by engine\n\nChatGPT leans toward high-authority factual sources—Wikipedia, established news outlets, and expert/editorial sites—and uses UGC sparingly.\nsearchengineland\n\nGoogle AI Overviews and Gemini draw from a wider mix, with notable use of blogs, YouTube, Reddit, and other community or social content.\nsearchengineland\n+1\n\nPerplexity commonly combines official sites, news, niche publications, and community sources; studies find it favors BBC among news domains and uses Reddit, LinkedIn, and G2 for some B2B queries.\narxiv\n+1\n\nClaude has been observed favoring brand domains and compliance-grade institutional sources rather than YouTube, Wikipedia, or Reddit in the cited study.\nconductor\n\nWhat this means\n\nFor visibility in AI search, the strongest source mix is usually: clear first-party content + credible third-party editorial coverage + accurate listings/reviews + selective community presence. Structured, verifiable information—such as official specifications, pricing, documentation, and directory data—appears especially influential because engines can retrieve and cite it directly.\nyext\n+1\n\nFor a marketing team, prioritize authoritative product/service pages, comparison and review coverage, trade press, and well-maintained profiles on relevant review platforms; for Google-oriented visibility, add useful YouTube and community content.\nsearchenginejournal\n+1",
      "citations": [
        {
          "title": "Data Shows AI Citation Patterns Reveal Strategic SEO ...",
          "host": "searchenginejournal",
          "url": "https://www.searchenginejournal.com/comparison-of-ai-citation-patterns-offers-strategic-seo-insights/573327/",
          "position": 1
        },
        {
          "title": "What Sources Get Cited by AI Search Engines — Bavior",
          "host": "bavior",
          "url": "https://bavior.com/learn/what-sources-get-cited",
          "position": 2
        },
        {
          "title": "Why AI Engines Cite Different Sources but Recommend ...",
          "host": "brightedge",
          "url": "https://www.brightedge.com/resources/weekly-ai-search-insights/ai-search-same-brands-different-sources",
          "position": 3
        },
        {
          "title": "How to get cited by AI: SEO insights from 8000 AI citations",
          "host": "searchengineland",
          "url": "https://searchengineland.com/how-to-get-cited-by-ai-seo-insights-from-8000-ai-citations-455284",
          "position": 4
        },
        {
          "title": "AI search engines cite Reddit, YouTube, and LinkedIn most",
          "host": "searchengineland",
          "url": "https://searchengineland.com/ai-search-engines-cite-reddit-youtube-and-linkedin-most-study-473138",
          "position": 5
        },
        {
          "title": "How AI Search Engines Choose Sources and Citations",
          "host": "conductor",
          "url": "https://www.conductor.com/academy/how-ai-citations-differ/",
          "position": 6
        },
        {
          "title": "How ChatGPT, Perplexity, Gemini, and Claude Actually Decide ...",
          "host": "yext",
          "url": "https://www.yext.com/blog/how-chatgpt-perplexity-gemini-claude-decide-what-to-cite",
          "position": 7
        },
        {
          "title": "News Source Citing Patterns in AI Search Systems",
          "host": "arxiv",
          "url": "https://arxiv.org/html/2507.05301v1",
          "position": 8
        },
        {
          "title": "AI search visibility report: October 2026 findings",
          "host": "meltwater",
          "url": "https://www.meltwater.com/en/blog/ai-search-visibility-report-october-2026",
          "position": 9
        },
        {
          "title": "What sources do AI Search Engines cite? Analysis of 40k responses …",
          "host": "xfunnel",
          "url": "https://www.xfunnel.ai/blog/what-sources-do-ai-search-engines-choose",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-source-types-cited-most",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-source-types-cited-most",
      "published_at": "2026-11-15T12:54:00.000Z",
      "updated_at": "2026-11-15T12:54:00.000Z",
      "revision": 1,
      "slug": "perplexity-source-types-cited-most",
      "panelId": "source-types-cited-most",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-source-types-cited-most",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "Which source types do AI search engines cite most often?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 20:54 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/ba0e5dec-7cff-4fa8-9f34-0b8bc34b281e",
      "mentions": "Cited",
      "summary": "AI search engines most often cite third-party, editorial/commercial sources—especially news, reviews, comparison sites, and industry publications—followed by official/brand sites. User-generated content (Reddit, YouTube, forums) is cited too, but its share varies sharply by engine.",
      "answerSnapshot": "AI search engines most often cite third-party, editorial/commercial sources—especially news, reviews, comparison sites, and industry publications—followed by official/brand sites. User-generated content (Reddit, YouTube, forums) is cited too, but its share varies sharply by engine.\nsearchenginejournal\n+2\n\nMost-cited source types\nSource type\tTypical role\tCitation pattern\nCommercial & editorial\tReviews, comparison sites, trade press, retailer listings, finance data\tThe largest shared category across major AI surfaces: about 37% of Gemini citations to 51% of AI Overviews citations in one 2026 analysis. \nsearchenginejournal\n+1\n\nOfficial / institutional\tGovernment, academic, major-brand and industry-leader pages\tA major source layer, roughly 10%–26% of citations depending on the engine; official pages can be the single largest category in some studies. \nsearchenginejournal\n+1\n\nNews\tMajor outlets and trade media\tFrequently cited, particularly for timely or factual questions; one dataset put news at 31% of ChatGPT citations but 16%–19% for Google and Perplexity. \nbavior\n\nUser-generated content\tReddit, YouTube, Quora, forums, social posts, reviews\tUsed by all engines, but much less consistently: from 0.2% to 18% of citations in the BrightEdge analysis. \nsearchenginejournal\n\nBlogs / expert content\tSpecialist publishers, how-to articles, vendor blogs\tImportant especially for product, how-to, and recommendation queries; one study found blogs accounted for roughly 21%–46% of citations depending on the engine. \nsearchengineland\nWhat differs by engine\n\nChatGPT and Perplexity tend to lean more heavily on high-authority factual sources such as Wikipedia, established news outlets, and specialist expert/review sites; Perplexity also often favors industry-specific review sources.\nsearchengineland\n\nGoogle AI Overviews and Gemini cast a wider net, with relatively stronger use of YouTube, Reddit, blogs, social content, and vendor material.\nsearchenginejournal\n+1\n\nClaude appears more selective in the cited research, favoring brand domains and institutional/compliance-grade sources rather than YouTube, Wikipedia, or Reddit.\nconductor\n\nEarned third-party coverage—coverage, affiliate/editorial sites, and independent reviews—was the most frequent citation category across ChatGPT, Gemini, and Perplexity in a separate 40,000-response study.\nxfunnel\n\nImportant nuance\n\n“Most cited” does not mean “most influential.” In one academic dataset, news media had a high citation share but relatively low influence on the final answer, while encyclopedia sources had the highest measured influence; engines may cite a timely news page while relying more on an explanatory reference page.\nbavior\n\nFor practical visibility, the strongest mix is usually: accurate official/brand pages + credible third-party editorial coverage and reviews + a small, high-quality UGC presence—rather than relying on one source type alone.\nsearchenginejournal\n+1",
      "citations": [
        {
          "title": "Data Shows AI Citation Patterns Reveal Strategic SEO ...",
          "host": "searchenginejournal",
          "url": "https://www.searchenginejournal.com/comparison-of-ai-citation-patterns-offers-strategic-seo-insights/573327/",
          "position": 1
        },
        {
          "title": "AI Search Optimization: Data Finds Brand Mentions Improve Visibility",
          "host": "searchenginejournal",
          "url": "https://www.searchenginejournal.com/ai-search-engines-often-cite-third-party-content-study-finds/540692/",
          "position": 2
        },
        {
          "title": "Why AI Engines Cite Different Sources but Recommend ...",
          "host": "brightedge",
          "url": "https://www.brightedge.com/resources/weekly-ai-search-insights/ai-search-same-brands-different-sources",
          "position": 3
        },
        {
          "title": "What Sources Get Cited by AI Search Engines — Bavior",
          "host": "bavior",
          "url": "https://bavior.com/learn/what-sources-get-cited",
          "position": 4
        },
        {
          "title": "How to get cited by AI: SEO insights from 8000 AI citations",
          "host": "searchengineland",
          "url": "https://searchengineland.com/how-to-get-cited-by-ai-seo-insights-from-8000-ai-citations-455284",
          "position": 5
        },
        {
          "title": "How AI Search Engines Choose Sources and Citations",
          "host": "conductor",
          "url": "https://www.conductor.com/academy/how-ai-citations-differ/",
          "position": 6
        },
        {
          "title": "What sources do AI Search Engines cite? Analysis of 40k responses …",
          "host": "xfunnel",
          "url": "https://www.xfunnel.ai/blog/what-sources-do-ai-search-engines-choose",
          "position": 7
        },
        {
          "title": "News Source Citing Patterns in AI Search Systems",
          "host": "arxiv",
          "url": "https://arxiv.org/html/2507.05301v1",
          "position": 8
        },
        {
          "title": "How ChatGPT, Perplexity, Gemini, and Claude Actually Decide ...",
          "host": "yext",
          "url": "https://www.yext.com/blog/how-chatgpt-perplexity-gemini-claude-decide-what-to-cite",
          "position": 9
        },
        {
          "title": "AI search visibility report: October 2026 findings",
          "host": "meltwater",
          "url": "https://www.meltwater.com/en/blog/ai-search-visibility-report-october-2026",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-source-types-cited-most-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-source-types-cited-most-repeat-2",
      "published_at": "2026-11-15T12:54:00.000Z",
      "updated_at": "2026-11-15T12:54:00.000Z",
      "revision": 1,
      "slug": "perplexity-source-types-cited-most-repeat-2",
      "panelId": "source-types-cited-most",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-source-types-cited-most",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "Which source types do AI search engines cite most often?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 20:54 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/cb7a63bb-5336-4198-90b5-150a5d2fa20f",
      "mentions": "Cited",
      "summary": "AI search engines most often cite official/brand-owned websites, news outlets, and niche editorial or review sites—but the exact mix varies substantially by engine and query type.",
      "answerSnapshot": "AI search engines most often cite official/brand-owned websites, news outlets, and niche editorial or review sites—but the exact mix varies substantially by engine and query type. Across several recent studies, commercial/editorial sources and official domains together usually account for the largest share of citations, while user-generated content (UGC) is much more prominent on Google’s AI surfaces.\nsearchenginejournal\n+2\n\nMost-cited source types\nSource type\tTypical role\tRelative citation strength\nOfficial / brand-owned sites\tProduct pages, documentation, company sites, directories\tOften the largest single category; one controlled study found 34.2% for ChatGPT, 46.4% for Google AI surfaces, and 44.1% for Perplexity. \nbavior\n\nNews and trade press\tCurrent events, industry reporting, authoritative context\tA major source everywhere, especially ChatGPT; news accounted for 31.2% of ChatGPT citations in that study. \nbavior\n\nNiche / vertical publications\tReviews, comparisons, specialist guidance, finance and product data\tFrequently cited, particularly for purchase, comparison, and recommendation queries. \nsearchenginejournal\n+2\n\nUGC\tReddit, YouTube, Quora, LinkedIn, review communities\tImportant for Google AI Overviews and Gemini; less used by ChatGPT and Perplexity. \nsearchenginejournal\n+2\n\nEncyclopedic / reference sources\tWikipedia and similar factual reference pages\tStrong for ChatGPT, especially education and recommendation queries, but less dominant elsewhere. \nsearchengineland\n+1\n\nStructured listings / directories\tBusiness listings, review platforms, product catalogs\tHighly represented in one citation study—54.5% of distinct cited sources—because they provide verifiable, structured facts. \nyext\nDifferences by engine\n\nChatGPT tends to favor established, factual sources: Wikipedia, major news outlets, and authoritative editorial sites; it uses UGC sparingly.\nsearchengineland\n+1\n\nGoogle AI Overviews and Gemini use the broadest mix, with substantial reliance on blogs, YouTube, Reddit, and other community content.\nsearchengineland\n+2\n\nPerplexity commonly blends official sites, editorial/blog content, and selective UGC; it has been observed favoring sources such as BBC, Reddit, LinkedIn, and G2 depending on the query.\narxiv\n+1\n\nClaude has been observed to rely more on brand domains and compliance-grade institutional sources, with little use of YouTube, Wikipedia, or Reddit in one two-month tracking study.\nconductor\n\nWhat this means practically\n\nFor a brand or publisher, the highest-leverage citation sources are usually: a clear official site, credible third-party editorial coverage, relevant review/comparison sites, accurate listings, and a presence in the UGC platforms relevant to the category. Review sites, comparison content, trade press, retailer listings, and finance data are specifically identified as sources AI systems frequently reach for.\nsearchenginejournal\n+1\n\nFor a local or recommendation-focused query, UGC and review platforms can matter disproportionately; for factual, news, or technical queries, official documentation, established news coverage, and specialist publications tend to carry more weight.\nsearchengineland\n+2",
      "citations": [
        {
          "title": "Data Shows AI Citation Patterns Reveal Strategic SEO ...",
          "host": "searchenginejournal",
          "url": "https://www.searchenginejournal.com/comparison-of-ai-citation-patterns-offers-strategic-seo-insights/573327/",
          "position": 1
        },
        {
          "title": "What Sources Get Cited by AI Search Engines — Bavior",
          "host": "bavior",
          "url": "https://bavior.com/learn/what-sources-get-cited",
          "position": 2
        },
        {
          "title": "Why AI Engines Cite Different Sources but Recommend ...",
          "host": "brightedge",
          "url": "https://www.brightedge.com/resources/weekly-ai-search-insights/ai-search-same-brands-different-sources",
          "position": 3
        },
        {
          "title": "How to get cited by AI: SEO insights from 8000 AI citations",
          "host": "searchengineland",
          "url": "https://searchengineland.com/how-to-get-cited-by-ai-seo-insights-from-8000-ai-citations-455284",
          "position": 4
        },
        {
          "title": "AI search engines cite Reddit, YouTube, and LinkedIn most",
          "host": "searchengineland",
          "url": "https://searchengineland.com/ai-search-engines-cite-reddit-youtube-and-linkedin-most-study-473138",
          "position": 5
        },
        {
          "title": "How AI Search Engines Choose Sources and Citations",
          "host": "conductor",
          "url": "https://www.conductor.com/academy/how-ai-citations-differ/",
          "position": 6
        },
        {
          "title": "How ChatGPT, Perplexity, Gemini, and Claude Actually Decide ...",
          "host": "yext",
          "url": "https://www.yext.com/blog/how-chatgpt-perplexity-gemini-claude-decide-what-to-cite",
          "position": 7
        },
        {
          "title": "News Source Citing Patterns in AI Search Systems",
          "host": "arxiv",
          "url": "https://arxiv.org/html/2507.05301v1",
          "position": 8
        },
        {
          "title": "AI search visibility report: October 2026 findings",
          "host": "meltwater",
          "url": "https://www.meltwater.com/en/blog/ai-search-visibility-report-october-2026",
          "position": 9
        },
        {
          "title": "What sources do AI Search Engines cite? Analysis of 40k responses …",
          "host": "xfunnel",
          "url": "https://www.xfunnel.ai/blog/what-sources-do-ai-search-engines-choose",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account Perplexity browser run in a fresh conversation. The product did not expose its complete retrieval configuration. The source panel reported 10 sources and 10 unique source records were captured. Results can vary by session, locale, account state and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/copilot-citation-worthy-pages",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/copilot-citation-worthy-pages",
      "published_at": "2026-11-15T09:45:22.000Z",
      "updated_at": "2026-11-15T10:38:10.000Z",
      "revision": 3,
      "slug": "copilot-citation-worthy-pages",
      "panelId": "citation-worthy-pages",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "copilot-citation-worthy-pages",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "What makes a web page more likely to be cited by AI search?",
      "product": "Microsoft Copilot",
      "observedAt": "15 November 2026",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Microsoft Copilot web chat with web search",
      "model": "Copilot Auto (model version not exposed)",
      "responseUrl": "https://copilot.com/chat/conversation/ca57c3ae-ca7d-4e59-ab02-8b45a5016c3a",
      "mentions": "Cited",
      "summary": "Copilot emphasized demonstrated expertise, original information, direct answers, people-first content, clear structure, technical accessibility and current evidence.",
      "answerSnapshot": "Copilot said AI search systems tend to cite pages that are easy to trust, easy to understand and clearly answer specific questions. It listed eight recurring factors: demonstrated expertise and trustworthiness, original information, direct answers to specific questions, people-first content, clear structure, strong technical SEO, accurate and current information, and topical authority. It recommended visible authorship and contact information, primary evidence, concise early answers, descriptive headings, crawlable content, internal links, review dates and a coherent cluster of related expertise. It cautioned that thin, mass-produced or keyword-stuffed pages are less defensible.",
      "citations": [
        {
          "title": "Creating Helpful, Reliable, People-First Content",
          "host": "developers.google.com",
          "url": "https://developers.google.com/search/docs/fundamentals/creating-helpful-content",
          "position": 1
        },
        {
          "title": "Google Search's Guidance on Generative AI Content",
          "host": "developers.google.com",
          "url": "https://developers.google.com/search/docs/fundamentals/using-gen-ai-content",
          "position": 2
        },
        {
          "title": "Google helpful content guidance",
          "host": "whitepress.com",
          "url": "https://www.whitepress.com/en/knowledge-base/2227/google-helpful-content",
          "position": 3
        }
      ],
      "limitations": "One authenticated Microsoft Copilot web-chat run. The answer is general guidance rather than causal evidence that changing a page will produce a citation.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/copilot-citation-worthy-pages-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/copilot-citation-worthy-pages-repeat-2",
      "published_at": "2026-11-15T10:38:09.000Z",
      "updated_at": "2026-11-15T10:38:10.000Z",
      "revision": 2,
      "slug": "copilot-citation-worthy-pages-repeat-2",
      "panelId": "citation-worthy-pages",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "copilot-citation-worthy-pages",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "What makes a web page more likely to be cited by AI search?",
      "product": "Microsoft Copilot",
      "observedAt": "15 November 2026, 18:31 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Microsoft Copilot web chat with web search",
      "model": "Copilot Auto (model version not exposed)",
      "responseUrl": "https://copilot.com/chat/conversation/30339954-a19c-489a-9468-1b29d5d5b6d3",
      "mentions": "Cited",
      "summary": "Copilot emphasized answer-first passages, clear structure, named expertise, specific entities and facts, structured data, topic completeness and site-level authority.",
      "answerSnapshot": "Copilot said pages are more citable when they are easy to understand, trustworthy and easy for machines to extract. It recommended a clear answer early, question-based headings, short paragraphs, steps, lists, comparison tables and FAQs; visible authors, company context, original research and cited evidence; named entities, dates, statistics and definitions; common Schema.org types; comprehensive coverage of follow-up questions; and site-level credibility supported by reputable links, consistent expertise and brand recognition. It presented these as recurring industry patterns rather than disclosed ranking weights.",
      "citations": [
        {
          "title": "How to Structure Content for AI Citations and LLM Visibility in 2026",
          "host": "writesonic.com",
          "position": 1
        },
        {
          "title": "The top content formats & types that earn AI search citations",
          "host": "blog.hubspot.com",
          "position": 2
        },
        {
          "title": "Structured Data for AI Citations: The Complete 2026 Guide",
          "host": "aiadvantageagency.com",
          "position": 3
        },
        {
          "title": "Structured Data for AI Search: October 2026",
          "host": "maintouch.com",
          "position": 4
        },
        {
          "title": "How to Increase AI Search Citations: A Prioritised Framework",
          "host": "searchable.com",
          "position": 5
        },
        {
          "title": "Structured Data for AI Search: 40% More Citations",
          "host": "xseek.io",
          "position": 6
        }
      ],
      "limitations": "Authenticated Microsoft Copilot web-chat run in a fresh conversation. The answer summarizes third-party recommendations and does not establish that any one change causes a citation.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/copilot-citation-worthy-pages-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/copilot-citation-worthy-pages-repeat-3",
      "published_at": "2026-11-15T10:38:09.000Z",
      "updated_at": "2026-11-15T10:38:10.000Z",
      "revision": 2,
      "slug": "copilot-citation-worthy-pages-repeat-3",
      "panelId": "citation-worthy-pages",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "copilot-citation-worthy-pages",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "What makes a web page more likely to be cited by AI search?",
      "product": "Microsoft Copilot",
      "observedAt": "15 November 2026, 18:32 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Microsoft Copilot web chat with web search",
      "model": "Copilot Auto (model version not exposed)",
      "responseUrl": "https://copilot.com/chat/conversation/7a86567c-4c1d-492b-9696-71cbe94bcfd2",
      "mentions": "Cited",
      "summary": "Copilot framed citation-worthiness around demonstrated expertise, original information, trust signals, structured entities, answer-first formatting, external authority and maintenance.",
      "answerSnapshot": "Copilot said citable pages are easy to trust, parse and distinguish through unique information. It prioritized visible authorship and credentials, first-hand data and case studies, sourcing and contact information, transparent editorial standards, secure operation, Schema.org entity markup, direct answer-first formatting, reputable external mentions and links, and visible publish or update dates with regular revision. It summarized the shift as traditional SEO rewarding pages that rank well while AI search increasingly rewards sources that are trusted and useful to quote.",
      "citations": [
        {
          "title": "E-E-A-T for AI Search: Building Trust for LLM Citations",
          "host": "writesonic.com",
          "position": 1
        },
        {
          "title": "E-E-A-T and AI Search: Why Author Credentials Matter",
          "host": "contently.com",
          "position": 2
        },
        {
          "title": "E-E-A-T for AI Search: How It Affects AI Citations",
          "host": "bluejar.ai",
          "position": 3
        },
        {
          "title": "E-E-A-T for AI Search: Does It Still Matter in 2026?",
          "host": "supergeo.io",
          "position": 4
        },
        {
          "title": "E-E-A-T for AI Search: Author Entity Signals 2026",
          "host": "nadiamohamed.me",
          "position": 5
        },
        {
          "title": "E-E-A-T for AI Search: How to Build Authority That Gets Cited by AI",
          "host": "ziptie.ai",
          "position": 6
        }
      ],
      "limitations": "Authenticated Microsoft Copilot web-chat run in a fresh conversation. The answer is general guidance and does not prove that any listed content tactic will cause a citation.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/copilot-how-ai-search-selects-sources",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/copilot-how-ai-search-selects-sources",
      "published_at": "2026-11-15T09:45:22.000Z",
      "updated_at": "2026-11-15T10:38:10.000Z",
      "revision": 3,
      "slug": "copilot-how-ai-search-selects-sources",
      "panelId": "how-ai-search-selects-sources",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "copilot-how-ai-search-selects-sources",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "How do AI search engines choose which sources to cite?",
      "product": "Microsoft Copilot",
      "observedAt": "15 November 2026",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Microsoft Copilot web chat with web search",
      "model": "Copilot Auto (model version not exposed)",
      "responseUrl": "https://copilot.com/chat/conversation/463caa51-5cc6-4f52-814a-22251878a967",
      "mentions": "Cited",
      "summary": "Copilot described retrieval-augmented generation, passage relevance, source trust, freshness, domain reputation and selective attribution. It said traditional ranking and AI citation are not equivalent.",
      "answerSnapshot": "Copilot said most AI search engines use retrieval-augmented generation: search for relevant information, retrieve candidate documents, rank passages for question relevance, evaluate quality and trust signals, then generate an answer and attribute sources. It listed topical expertise, credibility, content quality, freshness and domain reputation as common trust signals. Copilot emphasized that systems select passages rather than only pages, that strong traditional ranking does not guarantee an AI citation, and that different engines can have low citation overlap. It summarized citation-worthy sources as directly relevant, factually detailed, trustworthy, clearly written, authoritative and useful for a specific sub-question.",
      "citations": [
        {
          "title": "AI Citations Explained: Google, ChatGPT, Perplexity",
          "host": "analyticsinsight.net",
          "url": "https://www.analyticsinsight.net/seo/how-ai-search-engines-select-sources-to-cite-what-publishers-need-to-know",
          "position": 1
        },
        {
          "title": "AI Source Selection: How AI Chooses Which Sources to Cite",
          "host": "amicited.com",
          "url": "https://www.amicited.com/glossary/ai-source-selection/",
          "position": 2
        },
        {
          "title": "How AI Search Selects Citations",
          "host": "apiserpent.com",
          "url": "https://apiserpent.com/blog/how-ai-search-selects-citations",
          "position": 3
        },
        {
          "title": "AI Search Citation Decision Factors",
          "host": "machinerelations.ai",
          "url": "https://machinerelations.ai/research/ai-search-citation-decision-factors-comparison-2026",
          "position": 4
        }
      ],
      "limitations": "One authenticated Microsoft Copilot web-chat run. Copilot Auto did not expose the underlying model version or retrieval index. Results can vary by account, locale, model selection and session.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/copilot-how-ai-search-selects-sources-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/copilot-how-ai-search-selects-sources-repeat-2",
      "published_at": "2026-11-15T10:38:09.000Z",
      "updated_at": "2026-11-15T10:38:10.000Z",
      "revision": 2,
      "slug": "copilot-how-ai-search-selects-sources-repeat-2",
      "panelId": "how-ai-search-selects-sources",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "copilot-how-ai-search-selects-sources",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "How do AI search engines choose which sources to cite?",
      "product": "Microsoft Copilot",
      "observedAt": "15 November 2026, 18:29 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Microsoft Copilot web chat with web search",
      "model": "Copilot Auto (model version not exposed)",
      "responseUrl": "https://copilot.com/chat/conversation/630a237d-5a2e-4ff2-9ab9-7ccedc08b92b",
      "mentions": "Cited",
      "summary": "Copilot described retrieval and grounding as the common sequence: interpret the query, retrieve candidate passages, filter for relevance and trust, generate an answer, then attach supporting citations.",
      "answerSnapshot": "Copilot said the citations usually come from material retrieved at query time rather than model memory. It outlined five stages: analyze intent and generate internal queries, retrieve candidate sources from an index or corpus, rank and filter at passage level for relevance, authority and freshness, generate the answer from the retrieved context, then map claims back to documents and display supporting links. It said direct answers, clear factual passages, trust and extractability make citation more likely, and that strong traditional ranking does not guarantee AI citation.",
      "citations": [
        {
          "title": "What Is Source Grounding in AI? Benefits, Citations, and RAG",
          "host": "parallel.ai",
          "position": 1
        },
        {
          "title": "How LLMs Retrieve and Cite Sources: RAG, Grounding, and Attribution",
          "host": "digital-identity-architects.com",
          "position": 2
        },
        {
          "title": "What is source grounding?",
          "host": "airops.com",
          "position": 3
        },
        {
          "title": "What Is Grounding in AI? How LLMs Tie Answers to Sources",
          "host": "searchable.com",
          "position": 4
        },
        {
          "title": "Grounding in AI Search: How Engines Tie Answers to Real Sources",
          "host": "aiseocompany.com",
          "position": 5
        },
        {
          "title": "AI Search Grounding: How AI Citation Works in ChatGPT, Perplexity, and More",
          "host": "aiplusautomation.com",
          "position": 6
        }
      ],
      "limitations": "Authenticated Microsoft Copilot web-chat run in a fresh conversation. Copilot Auto did not expose model or retrieval-index details, and the source panel exposed domains and titles but not stable destination URLs.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/copilot-how-ai-search-selects-sources-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/copilot-how-ai-search-selects-sources-repeat-3",
      "published_at": "2026-11-15T10:38:09.000Z",
      "updated_at": "2026-11-15T10:38:10.000Z",
      "revision": 2,
      "slug": "copilot-how-ai-search-selects-sources-repeat-3",
      "panelId": "how-ai-search-selects-sources",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "copilot-how-ai-search-selects-sources",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "How do AI search engines choose which sources to cite?",
      "product": "Microsoft Copilot",
      "observedAt": "15 November 2026, 18:30 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Microsoft Copilot web chat with web search",
      "model": "Copilot Auto (model version not exposed)",
      "responseUrl": "https://copilot.com/chat/conversation/8c6e3927-3354-438c-b82a-a2486f9b06c6",
      "mentions": "Cited",
      "summary": "Copilot emphasized query rewriting, candidate retrieval, passage ranking, quality signals and claim-level attribution, while noting low citation overlap between AI products.",
      "answerSnapshot": "Copilot described a retrieval-and-ranking process before answer generation: understand and rewrite the question, retrieve candidate pages or passages, rank passages rather than whole pages, evaluate relevance, expertise, extractability, corroboration, freshness and verifiable facts, then attach citations to sources that support specific answer statements. It said a top Google position does not guarantee citation, different AI engines often have low source overlap, and consensus across trusted sources can reduce factual risk.",
      "citations": [
        {
          "title": "How AI Engines Choose Sources to Cite: 2026 Guide",
          "host": "searchcounselco.com",
          "position": 1
        },
        {
          "title": "AI Search Source Selection: How Engines Pick Sources",
          "host": "ollagraph.com",
          "position": 2
        },
        {
          "title": "How AI Search Engines Choose Sources for Citation",
          "host": "authoritystack.ai",
          "position": 3
        },
        {
          "title": "How Six AI Engines Choose Sources | MR Research",
          "host": "machinerelations.ai",
          "position": 4
        },
        {
          "title": "How AI Search Engines Choose Sources | MR Research",
          "host": "machinerelations.ai",
          "position": 5
        },
        {
          "title": "GitHub - abouchard11/ai-citation-patterns",
          "host": "github.com",
          "position": 6
        }
      ],
      "limitations": "Authenticated Microsoft Copilot web-chat run in a fresh conversation. The answer is one observation within a small repeated sample and does not expose Copilot's proprietary ranking weights.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/copilot-measure-chatgpt-citations",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/copilot-measure-chatgpt-citations",
      "published_at": "2026-11-15T09:45:22.000Z",
      "updated_at": "2026-11-15T10:38:10.000Z",
      "revision": 3,
      "slug": "copilot-measure-chatgpt-citations",
      "panelId": "measure-chatgpt-citations",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "copilot-measure-chatgpt-citations",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "How should a new website measure citations in ChatGPT and Perplexity?",
      "product": "Microsoft Copilot",
      "observedAt": "15 November 2026",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Microsoft Copilot web chat with web search",
      "model": "Copilot Auto (model version not exposed)",
      "responseUrl": "https://copilot.com/chat/conversation/ca0b6c28-9029-4dcc-a6c3-1c0ec4f7f77b",
      "mentions": "Cited",
      "summary": "Copilot recommended a fixed prompt panel and five recurring metrics: citation rate, competitor citation share, AI referral traffic, citation quality and citation consistency.",
      "answerSnapshot": "Copilot recommended treating AI citations as an acquisition and visibility channel rather than only referral traffic. It proposed a fixed panel of 20 to 50 customer prompts run weekly in ChatGPT and Perplexity. Its five metrics were citation rate, share of citations versus competitors, AI referral traffic in analytics, citation quality and citation consistency. It advised recording whether the site was the primary source, one of several sources, only mentioned or linked directly, and reviewing monthly trends rather than single-day counts or one viral mention. It defined citation rate as the number of prompt runs citing the site divided by total prompt runs.",
      "citations": [
        {
          "title": "How to Track AI Citations for Your Website",
          "host": "thecssagency.com",
          "url": "https://www.thecssagency.com/blog/how-to-track-ai-citations",
          "position": 1
        },
        {
          "title": "How to Measure AI Citations",
          "host": "optimaize.app",
          "url": "https://optimaize.app/guides/measure-ai-citations",
          "position": 2
        },
        {
          "title": "How to Measure AI Visibility and Citations",
          "host": "authoritystack.ai",
          "url": "https://authoritystack.ai/ai-search/how-to-measure-ai-visibility-and-citations",
          "position": 3
        },
        {
          "title": "AI Citation Tracking Guide 2026",
          "host": "loudpixel.ai",
          "url": "https://loudpixel.ai/blog/ai-citation-tracking-guide-2026/",
          "position": 4
        }
      ],
      "limitations": "One authenticated Microsoft Copilot web-chat run. The answer describes a measurement design; this record does not establish that the cited third-party methods are independently validated.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/copilot-measure-chatgpt-citations-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/copilot-measure-chatgpt-citations-repeat-2",
      "published_at": "2026-11-15T10:38:09.000Z",
      "updated_at": "2026-11-15T10:38:10.000Z",
      "revision": 2,
      "slug": "copilot-measure-chatgpt-citations-repeat-2",
      "panelId": "measure-chatgpt-citations",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "copilot-measure-chatgpt-citations",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "How should a new website measure citations in ChatGPT and Perplexity?",
      "product": "Microsoft Copilot",
      "observedAt": "15 November 2026, 18:33 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Microsoft Copilot web chat with web search",
      "model": "Copilot Auto (model version not exposed)",
      "responseUrl": "https://copilot.com/chat/conversation/f33a955e-ef13-435f-9d8d-819caf902c36",
      "mentions": "Cited",
      "summary": "Copilot recommended platform-specific citation rate, citation share, prompt coverage, citation position and AI referral outcomes, measured against a fixed recurring benchmark set.",
      "answerSnapshot": "Copilot advised treating AI citations as an acquisition channel while measuring ChatGPT and Perplexity separately. It proposed five metrics: citation rate, share of visible citations, prompt coverage by query category, citation position and referral sessions or conversions. It recommended running the same benchmark prompt set weekly and reporting a compact dashboard rather than random questions or single answers. It summarized the workflow as retrieval, citation and click-through, while warning that a citation does not guarantee a visit or revenue.",
      "citations": [
        {
          "title": "AI Citations per Answer: ChatGPT Cites 3.7, Perplexity 20",
          "host": "suffdigital.com",
          "position": 1
        },
        {
          "title": "Perplexity vs ChatGPT Citations | How Each Picks",
          "host": "chudi.dev",
          "position": 2
        },
        {
          "title": "How Citation Display in AI Answers Began",
          "host": "note.com",
          "position": 3
        },
        {
          "title": "AI Platforms Like ChatGPT and Perplexity Decide What to Cite",
          "host": "authoritytech.io",
          "position": 4
        },
        {
          "title": "AI Search Engine Check Procedure",
          "host": "note.com",
          "position": 5
        },
        {
          "title": "The 10 Best AI Search Engines in 2026",
          "host": "memeburn.com",
          "position": 6
        },
        {
          "title": "How AI Engines Cite — ChatGPT, Perplexity, Claude, Gemini",
          "host": "agentpatterns.ai",
          "position": 7
        },
        {
          "title": "ChatGPT vs Perplexity: How AI Citation Logic Really Works",
          "host": "scribble.network",
          "position": 8
        },
        {
          "title": "How ChatGPT, Perplexity & Gemini Choose What to Cite",
          "host": "citationradar.ai",
          "position": 9
        }
      ],
      "limitations": "Authenticated Microsoft Copilot web-chat run in a fresh conversation. The answer is a proposed measurement framework; the example dashboard values are illustrative rather than observed benchmark data.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/copilot-measure-chatgpt-citations-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/copilot-measure-chatgpt-citations-repeat-3",
      "published_at": "2026-11-15T10:38:09.000Z",
      "updated_at": "2026-11-15T10:38:10.000Z",
      "revision": 2,
      "slug": "copilot-measure-chatgpt-citations-repeat-3",
      "panelId": "measure-chatgpt-citations",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "copilot-measure-chatgpt-citations",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "How should a new website measure citations in ChatGPT and Perplexity?",
      "product": "Microsoft Copilot",
      "observedAt": "15 November 2026, 18:34 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Microsoft Copilot web chat with web search",
      "model": "Copilot Auto (model version not exposed)",
      "responseUrl": "https://copilot.com/chat/conversation/362df8fc-5c73-4923-8198-a73e9ee81a54",
      "mentions": "Cited",
      "summary": "Copilot recommended a fixed prompt panel, citation share of voice, AI referral outcomes and separate tracking for mentions, citations and clicks.",
      "answerSnapshot": "Copilot proposed three measurement layers: citation share of voice across a stable target-query set, AI referral traffic and conversions in analytics, and separate counts for brand mentions, source citations and click-throughs. It recommended a compact monthly dashboard covering citation count, share of voice, prompt coverage, referral sessions, conversion rate, top cited pages and prompt-level results. It warned that AI referral attribution can be incomplete and said traditional rankings or domain authority should not be treated as direct citation metrics.",
      "citations": [
        {
          "title": "AI citation tracking: How to track and grow AI engine citations",
          "host": "blog.hubspot.com",
          "position": 1
        },
        {
          "title": "AI Visibility Analytics | Track Brand Citations Across ChatGPT",
          "host": "citemetrix.com",
          "position": 2
        },
        {
          "title": "How to Measure AI Search Traffic from ChatGPT, Perplexity & Gemini",
          "host": "optiseon.com",
          "position": 3
        },
        {
          "title": "AI Referral Traffic: How to Track and Increase It",
          "host": "marketinglad.io",
          "position": 4
        },
        {
          "title": "AI Citations Explained: Google, ChatGPT, Perplexity",
          "host": "analyticsinsight.net",
          "position": 5
        },
        {
          "title": "ChatGPT vs Perplexity for AI Visibility in 2026",
          "host": "quickseo.ai",
          "position": 6
        }
      ],
      "limitations": "Authenticated Microsoft Copilot web-chat run in a fresh conversation. The response proposes a monitoring design and does not establish that the cited third-party benchmarks are independently validated.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-citation-worthy-pages",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-citation-worthy-pages",
      "published_at": "2026-11-15T09:12:02.000Z",
      "updated_at": "2026-11-15T10:38:10.000Z",
      "revision": 3,
      "slug": "perplexity-citation-worthy-pages",
      "panelId": "citation-worthy-pages",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-citation-worthy-pages",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "What makes a web page more likely to be cited by AI search?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 17:11 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/ae41d10c-e357-406a-abc8-25daa03208ba",
      "mentions": "Cited",
      "summary": "Perplexity said citation likelihood depends on retrievability, direct query relevance, extractable passages, trust, original information and corroboration. It did not claim that any one tactic guarantees a citation.",
      "answerSnapshot": "Perplexity stated that a page is more likely to be cited when it is easy to retrieve, directly answers a real question, appears trustworthy and contains a clean, self-contained passage that can be extracted and attributed. It organized the factors into retrievability, query relevance, extractability, authority and trust, original information gain, corroboration and freshness, and clean structure. It warned that long generic pages, keyword stuffing and list-heavy content do not reliably increase citations. It recommended a direct early answer followed by definitions, comparison tables, original evidence, source attribution and clearly headed related questions.",
      "citations": [
        {
          "title": "What Drives AI Citations? Content, Retrieval and Model Effects",
          "host": "arxiv.org",
          "url": "https://arxiv.org/html/2610.15033",
          "position": 1
        },
        {
          "title": "AI search visibility: Why your web content gets skipped",
          "host": "experienceleague.adobe.com",
          "url": "https://experienceleague.adobe.com/en/perspectives/ai-search-visibility-why-web-content-gets-skipped",
          "position": 2
        },
        {
          "title": "How to get cited in AI search: The complete guide",
          "host": "tryprofound.com",
          "url": "https://www.tryprofound.com/articles/how-to-get-cited-in-ai-search",
          "position": 3
        },
        {
          "title": "What high-citation brands do differently in AI search",
          "host": "blog.hubspot.com",
          "url": "https://blog.hubspot.com/marketing/brands-with-high-ai-search-citations",
          "position": 4
        },
        {
          "title": "AI search and LLM optimization tactics",
          "host": "searchengineland.com",
          "url": "https://searchengineland.com/guide/ai-and-llm-optimization",
          "position": 5
        },
        {
          "title": "What Makes Websites Get Cited by AI?",
          "host": "silktide.com",
          "url": "https://silktide.com/blog/how-websites-get-cited-in-ai/",
          "position": 6
        },
        {
          "title": "How Do Technical SEO Factors Impact AI Search?",
          "host": "semrush.com",
          "url": "https://www.semrush.com/blog/technical-seo-impact-on-ai-search-study/",
          "position": 7
        },
        {
          "title": "Which Pages Get Cited by AI?",
          "host": "vemetric.com",
          "url": "https://www.vemetric.com/blog/ai-citations-tracking",
          "position": 8
        },
        {
          "title": "How to Optimize Content for AI Search Engines",
          "host": "searchable.com",
          "url": "https://www.searchable.com/blog/optimize-content-for-ai-search",
          "position": 9
        },
        {
          "title": "AI Search Optimization: Tips and Best Practices",
          "host": "aisearch.similarweb.com",
          "url": "https://aisearch.similarweb.com/blog/ai-search-optimization-best-practices/",
          "position": 10
        }
      ],
      "limitations": "One authenticated free-account run on 15 November 2026. The answer describes general factors and is not causal evidence that changing any one page will produce a citation.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-citation-worthy-pages-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-citation-worthy-pages-repeat-2",
      "published_at": "2026-11-15T10:38:09.000Z",
      "updated_at": "2026-11-15T10:38:10.000Z",
      "revision": 2,
      "slug": "perplexity-citation-worthy-pages-repeat-2",
      "panelId": "citation-worthy-pages",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-citation-worthy-pages",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "What makes a web page more likely to be cited by AI search?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 18:09 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/38d80297-a9af-4b59-b86f-b009e55c1852",
      "mentions": "Cited",
      "summary": "Perplexity again prioritized direct question-to-page relevance, retrievability, original evidence and clear conditions, while separating observed associations from guaranteed citation factors.",
      "answerSnapshot": "Perplexity said discoverability, direct question matching and reference-worthy evidence matter more than special AI-friendly markup. It summarized an October 2026 observational study of 158,099 answers and 468,037 citations in which better question-to-paragraph and question-to-title matches were associated with more citations, while FAQ blocks, schema and summary boxes did not show the same robust within-site relationship. It recommended crawler access, internal links, readable text, unique first-hand information and passages with enough context to quote accurately. It explicitly framed these as observed associations rather than a guaranteed ranking formula.",
      "citations": [
        {
          "title": "AI Features and Your Website | Google Search Central",
          "host": "developers.google.com",
          "url": "https://developers.google.com/search/docs/appearance/ai-features",
          "position": 1
        },
        {
          "title": "What Drives AI Citations? Content, Retrieval and Model Effects",
          "host": "arxiv.org",
          "url": "https://arxiv.org/html/2610.15033",
          "position": 2
        },
        {
          "title": "Top ways to ensure your content performs well in Google's AI search",
          "host": "developers.google.com",
          "url": "https://developers.google.com/search/blog/2025/05/succeeding-in-ai-search",
          "position": 3
        },
        {
          "title": "What Makes Websites Get Cited by AI? Meta-Analysis Results",
          "host": "silktide.com",
          "url": "https://silktide.com/blog/how-websites-get-cited-in-ai/",
          "position": 4
        },
        {
          "title": "Why Some Websites Get Cited by AI Search Engines",
          "host": "callidoratechnology.com",
          "url": "https://callidoratechnology.com/blog/why-some-websites-get-cited-by-ai-search-engines",
          "position": 5
        },
        {
          "title": "How Websites Get Mentioned in AI Search Results",
          "host": "readyartwork.com",
          "url": "https://readyartwork.com/what-makes-a-business-website-more-likely-to-be-mentioned-in-ai-search-results/",
          "position": 6
        },
        {
          "title": "What Makes a Web Page Worth Citing in AI Search?",
          "host": "linkedin.com",
          "url": "https://linkedin.com/pulse/what-makes-web-page-worth-citing-ai-search-three29-yql3c",
          "position": 7
        },
        {
          "title": "What's actually helping websites get cited in AI search?",
          "host": "reddit.com",
          "url": "https://reddit.com/r/GenerativeSEOstrategy/comments/1wil17y/whats_actually_helping_websites_get_cited_in_ai/",
          "position": 8
        },
        {
          "title": "AI Citation Tracker: Discover Why Some Websites Get Cited by AI",
          "host": "my.wdt.edu",
          "url": "https://my.wdt.edu/ICS/icsfs/mm/ai-citation.pdf?target=c5fa0aa7-c58c-4338-9c87-15be3c62ac3e",
          "position": 9
        },
        {
          "title": "What Does It Mean When An AI Cites Your Website?",
          "host": "seo-theory.com",
          "url": "https://seo-theory.com/what-does-it-mean-when-an-ai-cites-your-website/",
          "position": 10
        },
        {
          "title": "What Makes a Website More Likely to Be Cited by AI Tools?",
          "host": "digital-marketing-agency.us",
          "url": "https://digital-marketing-agency.us/blog/what-makes-a-website-more-likely-to-be-cited-by-ai-tools/",
          "position": 11
        },
        {
          "title": "How Do AI Search Engines Decide Which Sources to Cite?",
          "host": "digitalstrategyforce.com",
          "url": "https://digitalstrategyforce.com/journal/how-do-ai-search-engines-decide-which-sources-to-cite/",
          "position": 12
        },
        {
          "title": "Google's Guide to Optimizing for Generative AI Features on Search",
          "host": "developers.google.com",
          "url": "https://developers.google.com/search/docs/fundamentals/ai-optimization-guide",
          "position": 13
        },
        {
          "title": "A new resource for optimizing for generative AI in Search",
          "host": "developers.google.com",
          "url": "https://developers.google.com/search/blog/2026/05/a-new-resource-for-optimizing",
          "position": 14
        },
        {
          "title": "Latest Google Search Documentation Updates",
          "host": "developers.google.com",
          "url": "https://developers.google.com/search/updates",
          "position": 15
        },
        {
          "title": "Google AI Mode - a new way to search",
          "host": "developers.google.com",
          "url": "https://developers.google.com/search/docs/crawling-indexing/google-agent-search",
          "position": 16
        },
        {
          "title": "Google Search's guidance on using generative AI content",
          "host": "developers.google.com",
          "url": "https://developers.google.com/search/docs/fundamentals/using-gen-ai-content",
          "position": 17
        },
        {
          "title": "Google Search's guidance about AI-generated content",
          "host": "developers.google.com",
          "url": "https://developers.google.com/search/blog/2023/02/google-search-and-ai-content",
          "position": 18
        },
        {
          "title": "Introduction to structured data markup in Google Search",
          "host": "developers.google.com",
          "url": "https://developers.google.com/search/docs/appearance/structured-data/intro-structured-data",
          "position": 19
        },
        {
          "title": "Introducing Search Generative AI performance reports in Search Console",
          "host": "developers.google.com",
          "url": "https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports?hl=en",
          "position": 20
        }
      ],
      "limitations": "Authenticated free-account run in a fresh Perplexity conversation. The answer is one observation within a small three-run sample and does not establish causal effects for any content change.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-citation-worthy-pages-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-citation-worthy-pages-repeat-3",
      "published_at": "2026-11-15T10:38:09.000Z",
      "updated_at": "2026-11-15T10:38:10.000Z",
      "revision": 2,
      "slug": "perplexity-citation-worthy-pages-repeat-3",
      "panelId": "citation-worthy-pages",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "partial",
      "series": "perplexity-citation-worthy-pages",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "What makes a web page more likely to be cited by AI search?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 18:09 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/9b0e26f9-56a2-4568-9110-d54c54299faf",
      "mentions": "Cited",
      "summary": "Perplexity emphasized question-to-passage relevance, crawlability and unique evidence, while warning that AI-specific markup and content formatting alone are not proven citation levers.",
      "answerSnapshot": "Perplexity said the strongest practical priority is relevance: the page should be discoverable, directly match the question and provide evidence worth referencing. It cited a 2026 observational study of 158,099 answers and 468,037 citations that associated stronger question-to-paragraph and question-to-title matches with more citations, while finding no comparably robust effect for FAQ blocks, schema, headings, tables, lists or word count. It recommended intentional crawler access, internal links, text-based explanations, unique first-hand evidence and enough context for a passage to be quoted accurately. It also noted that Google says special AI schema and llms.txt are not required.",
      "citations": [
        {
          "title": "What Drives AI Citations? Content, Retrieval and Model Effects",
          "host": "arxiv.org",
          "url": "https://arxiv.org/html/2610.15033",
          "position": 1
        },
        {
          "title": "AI Features and Your Website",
          "host": "developers.google.com",
          "url": "https://developers.google.com/search/docs/appearance/ai-features",
          "position": 2
        },
        {
          "title": "Top ways to ensure your content performs well in Google's AI search",
          "host": "developers.google.com",
          "url": "https://developers.google.com/search/blog/2025/05/succeeding-in-ai-search",
          "position": 3
        }
      ],
      "limitations": "Authenticated free-account run in a fresh Perplexity conversation. The answer exposed three linked sources in the rendered answer; the consolidated source panel did not populate during capture.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-how-ai-search-selects-sources",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-how-ai-search-selects-sources",
      "published_at": "2026-11-15T09:12:02.000Z",
      "updated_at": "2026-11-15T10:38:10.000Z",
      "revision": 3,
      "slug": "perplexity-how-ai-search-selects-sources",
      "panelId": "how-ai-search-selects-sources",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-how-ai-search-selects-sources",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "How do AI search engines choose which sources to cite?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 17:10 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/835e912d-b508-416b-82a8-074051c1382f",
      "mentions": "Cited",
      "summary": "Perplexity described source selection as passage-level retrieval and re-ranking, followed by answer generation with attribution. It emphasized question relevance, extractability, credibility, corroboration and freshness.",
      "answerSnapshot": "Perplexity answered that AI search engines generally cite sources whose specific passages best answer the question, while weighing credibility, freshness and traceability. It described a four-stage process: understand the query, retrieve candidate pages, re-rank passages, then generate an answer with citations. It listed direct relevance, an extractable answer, specific evidence, trust, corroboration and freshness as factors. It also cautioned that a citation shows grounding in a retrieved source; it does not prove that every sentence was independently verified.",
      "citations": [
        {
          "title": "How Does AI Search Work? AI Citation Mechanics Explained",
          "host": "straightnorth.com",
          "url": "https://straightnorth.com/blog/how-do-ai-search-engines-choose-which-sources-to-cite/",
          "position": 1
        },
        {
          "title": "How Do AI Engines Choose Which Sources to Cite?",
          "host": "gracker.ai",
          "url": "https://gracker.ai/questions/how-do-ai-engines-choose-which-sources-to-cite",
          "position": 2
        },
        {
          "title": "How AI Engines Choose Sources to Cite",
          "host": "attrifast.com",
          "url": "https://attrifast.com/blog/how-ai-engines-choose-sources",
          "position": 3
        },
        {
          "title": "Why AI citation is retrieval then ranking",
          "host": "lyntonweb.com",
          "url": "https://lyntonweb.com/library/how-llms-decide-what-to-cite",
          "position": 4
        },
        {
          "title": "How to Get Cited by AI: 9 Ways to Earn LLM Citations in 2026",
          "host": "surfer.ai",
          "url": "https://surfer.ai/blog/llm-citations/",
          "position": 5
        },
        {
          "title": "How AI decides which sources to cite",
          "host": "searchtides.com",
          "url": "https://searchtides.com/ai-visibility-engineering/how-ai-citation-works/",
          "position": 6
        },
        {
          "title": "How AI models choose which sources to cite",
          "host": "searchscore.io",
          "url": "https://searchscore.io/guides/how-ai-chooses-citations/",
          "position": 7
        },
        {
          "title": "How AI Search Engines Choose Sources and Citations",
          "host": "conductor.com",
          "url": "https://www.conductor.com/academy/how-ai-citations-differ/",
          "position": 8
        },
        {
          "title": "How Do AI Engines Decide Which Sources to Cite?",
          "host": "devenup.com",
          "url": "https://devenup.com/blog/how-do-ai-engines-decide-which-sources-to-cite",
          "position": 9
        },
        {
          "title": "How Do AI Search Engines Choose Sources?",
          "host": "leadtoconversion.com",
          "url": "https://leadtoconversion.com/blog/how-do-ai-search-engines-choose-sources/",
          "position": 10
        }
      ],
      "limitations": "One authenticated free-account run on 15 November 2026. Perplexity did not expose its automatic model version or retrieval index. The answer can change by session, language, account state, model selection and date.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-how-ai-search-selects-sources-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-how-ai-search-selects-sources-repeat-2",
      "published_at": "2026-11-15T10:38:09.000Z",
      "updated_at": "2026-11-15T10:38:10.000Z",
      "revision": 2,
      "slug": "perplexity-how-ai-search-selects-sources-repeat-2",
      "panelId": "how-ai-search-selects-sources",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "partial",
      "series": "perplexity-how-ai-search-selects-sources",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "How do AI search engines choose which sources to cite?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 18:08 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/50a5f71a-21c8-4681-bcb0-ff70f892b3a7",
      "mentions": "Cited",
      "summary": "Perplexity framed citation selection as candidate discovery, passage retrieval, relevance and usefulness assessment, then answer generation with supporting links; ranking is relevant but is not a guarantee.",
      "answerSnapshot": "Perplexity described four stages: discover candidate sources through related searches or query fan-out, retrieve useful evidence including specific passages, assess relevance, credibility and freshness, then generate the answer and attach supporting links. It highlighted exact-question relevance, reliability, clear evidence, appropriate freshness, accessibility and eligibility, and distinctive value. It said traditional ranking still matters but does not guarantee citation because an answer needs evidence for each claim and may search related subtopics.",
      "citations": [
        {
          "title": "AI Features and Your Website | Google Search Central",
          "host": "developers.google.com",
          "url": "https://developers.google.com/search/docs/appearance/ai-features",
          "position": 1
        },
        {
          "title": "How Do AI Search Engines Choose Which Sources to Cite?",
          "host": "straightnorth.com",
          "url": "https://straightnorth.com/blog/how-do-ai-search-engines-choose-which-sources-to-cite/",
          "position": 2
        },
        {
          "title": "Google's Guide to Optimizing for Generative AI Features on Search",
          "host": "developers.google.com",
          "url": "https://developers.google.com/search/docs/fundamentals/ai-optimization-guide",
          "position": 3
        }
      ],
      "limitations": "Authenticated free-account run in a fresh Perplexity conversation. The rendered answer exposed three linked source URLs; the consolidated source panel did not populate during capture.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-how-ai-search-selects-sources-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-how-ai-search-selects-sources-repeat-3",
      "published_at": "2026-11-15T10:38:09.000Z",
      "updated_at": "2026-11-15T10:38:10.000Z",
      "revision": 2,
      "slug": "perplexity-how-ai-search-selects-sources-repeat-3",
      "panelId": "how-ai-search-selects-sources",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "partial",
      "series": "perplexity-how-ai-search-selects-sources",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "How do AI search engines choose which sources to cite?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 18:08 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/a21d12ac-ead7-4a88-8bf2-4d7eee9d4076",
      "mentions": "Cited",
      "summary": "Perplexity separated retrieval from citation: engines first find and rank candidate sources, then cite passages that support specific claims, without publishing a universal weighting formula.",
      "answerSnapshot": "Perplexity described source selection as query interpretation and search fan-out, retrieval and ranking, identification of passages that support the answer, then attribution. It said citation selection is related to search ranking but is not merely the ordinary results list. It highlighted exact-question relevance, passage-level support, reliability, accessibility and eligibility, freshness and extractable content, while noting that platforms do not publish a complete weighting formula. It warned that a retrieved page is not necessarily cited.",
      "citations": [
        {
          "title": "AI Features and Your Website | Google Search Central",
          "host": "developers.google.com",
          "url": "https://developers.google.com/search/docs/appearance/ai-features",
          "position": 1
        },
        {
          "title": "Searching the web with ChatGPT",
          "host": "help.openai.com",
          "url": "https://help.openai.com/en/articles/9237897-searching-the-web-with-chatgpt",
          "position": 2
        },
        {
          "title": "Google's Guide to Optimizing for Generative AI Features on Search",
          "host": "developers.google.com",
          "url": "https://developers.google.com/search/docs/fundamentals/ai-optimization-guide",
          "position": 3
        }
      ],
      "limitations": "Authenticated free-account run in a fresh Perplexity conversation. The rendered answer exposed three linked source URLs; the consolidated source panel did not populate during capture.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-measure-chatgpt-citations",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-measure-chatgpt-citations",
      "published_at": "2026-11-15T09:12:02.000Z",
      "updated_at": "2026-11-15T10:38:10.000Z",
      "revision": 3,
      "slug": "perplexity-measure-chatgpt-citations",
      "panelId": "measure-chatgpt-citations",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-measure-chatgpt-citations",
      "repeatRun": 1,
      "repeatTotal": 3,
      "question": "How should a new website measure citations in ChatGPT and Perplexity?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 17:11 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/575d36ec-07a9-43a5-b358-cff30b3cffc2",
      "mentions": "Cited",
      "summary": "Perplexity recommended a repeatable prompt-panel experiment with stable wording, repeated runs and separate measures for citation, mention, cited URL and competitor presence.",
      "answerSnapshot": "Perplexity recommended treating citation measurement as a repeatable prompt-panel experiment rather than a one-off search. It suggested 20 to 30 buyer questions across brand, category, comparison and problem-solving intents; keeping wording stable; using fresh ChatGPT and Perplexity sessions with the same country, language and visible mode; and running each prompt at least three times. It defined citation rate as prompt runs where the domain is cited at least once divided by total prompt runs. It also recommended recording the exact cited URL, position and competitor domains, and measuring referral traffic and crawler activity separately. It warned not to count a brand mention as a citation or a crawler hit as inclusion in an answer.",
      "citations": [
        {
          "title": "How to Measure Your AI Citation Rate Across ChatGPT, Perplexity, and Claude",
          "host": "freecodecamp.org",
          "url": "https://www.freecodecamp.org/news/how-to-measure-your-ai-citation-rate-across-chatgpt-perplexity-and-claude/",
          "position": 1
        },
        {
          "title": "How to Measure If ChatGPT and AI Overviews Cite You",
          "host": "denverseo.io",
          "url": "https://denverseo.io/blog/measure-ai-citations/",
          "position": 2
        },
        {
          "title": "How to Get Cited in ChatGPT, Perplexity and Google AI Overviews",
          "host": "growth100x.com",
          "url": "https://growth100x.com/insights/how-to-get-cited-chatgpt-perplexity-google-ai-overviews/",
          "position": 3
        },
        {
          "title": "AI Citation Tracking Guide",
          "host": "useomnia.com",
          "url": "https://useomnia.com/blog/ai-citation-tracking",
          "position": 4
        },
        {
          "title": "What Are AI Citations?",
          "host": "searchable.com",
          "url": "https://www.searchable.com/fr/glossary/ai-citations",
          "position": 5
        },
        {
          "title": "How to Get Your Website Cited by ChatGPT and Perplexity",
          "host": "automatelab.tech",
          "url": "https://automatelab.tech/blog/al-products/get-website-cited-chatgpt-perplexity-2026-checklist/",
          "position": 6
        },
        {
          "title": "How to Get Your Site Cited in ChatGPT and Perplexity Answers",
          "host": "geniusfirms.com",
          "url": "https://geniusfirms.com/insight/how-to-get-your-site-cited-in-chatgpt-and-perplexity-answers/",
          "position": 7
        },
        {
          "title": "How to track when AI systems cite your site",
          "host": "docs.google.com",
          "url": "https://docs.google.com/document/d/1x-ablJZOWH-t3NRd-gOSxPAQm3pzgetlbaO5C8eGuyQ/mobilebasic",
          "position": 8
        },
        {
          "title": "Is ChatGPT citing your site? A 30 minute way to check",
          "host": "dev.to",
          "url": "https://dev.to/robin_geo/is-chatgpt-citing-your-site-a-30-minute-way-to-check-and-what-makes-a-page-citable-1p4k",
          "position": 9
        },
        {
          "title": "AI citation tracking: How to track and grow AI engine citations",
          "host": "blog.hubspot.com",
          "url": "https://blog.hubspot.com/marketing/ai-citation-tracking",
          "position": 10
        }
      ],
      "limitations": "One authenticated free-account run on 15 November 2026. Perplexity's answer describes a measurement design; no ChatGPT or Perplexity citation panel was run for this record.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-measure-chatgpt-citations-repeat-2",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-measure-chatgpt-citations-repeat-2",
      "published_at": "2026-11-15T10:38:09.000Z",
      "updated_at": "2026-11-15T10:38:10.000Z",
      "revision": 2,
      "slug": "perplexity-measure-chatgpt-citations-repeat-2",
      "panelId": "measure-chatgpt-citations",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-measure-chatgpt-citations",
      "repeatRun": 2,
      "repeatTotal": 3,
      "question": "How should a new website measure citations in ChatGPT and Perplexity?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 18:09 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/6409e748-56f5-4d52-a415-ff2dfa752513",
      "mentions": "Cited",
      "summary": "Perplexity recommended a 20–30 question panel, platform-specific scorecards and a clear separation between mentions, citations, crawler activity and referral outcomes.",
      "answerSnapshot": "Perplexity proposed a fixed 20–30 prompt panel split across unbranded information, long-tail problems, category buying questions and branded evaluation. It recommended fresh chats with consistent language, location, account state and search mode, repeating each prompt and reporting ChatGPT and Perplexity separately. Its scorecard included citation rate, brand mention rate, prompt coverage, distinct cited pages, competitor inclusion, referral sessions and conversions. It stressed that a mention is not a link, a crawler request is not a citation, and a referral measures clicks rather than all exposure.",
      "citations": [
        {
          "title": "How to Measure Your AI Citation Rate Across ChatGPT, Perplexity, and Claude",
          "host": "freecodecamp.org",
          "url": "https://freecodecamp.org/news/how-to-measure-your-ai-citation-rate-across-chatgpt-perplexity-and-claude/",
          "position": 1
        },
        {
          "title": "Track Sources Cited by ChatGPT and Perplexity: A Practical Guide",
          "host": "maxaeo.ai",
          "url": "https://maxaeo.ai/blog/track-sources-chatgpt-perplexity/",
          "position": 2
        },
        {
          "title": "How to Measure If ChatGPT and AI Overviews Cite You",
          "host": "denverseo.io",
          "url": "https://denverseo.io/blog/measure-ai-citations/",
          "position": 3
        },
        {
          "title": "How to Get Cited by ChatGPT and Perplexity (and Track It)",
          "host": "websitevikreta.com",
          "url": "https://websitevikreta.com/blog/educational/cited-chatgpt-perplexity-strategies",
          "position": 4
        },
        {
          "title": "The New Organic Growth Stack: Search, Social, Communities & AI",
          "host": "searchenginejournal.com",
          "url": "https://searchenginejournal.com/new-organic-growth-stack-search-social-communities-ai/594542/",
          "position": 5
        },
        {
          "title": "Trustpoint Xposure AEO study release",
          "host": "theglobeandmail.com",
          "url": "https://theglobeandmail.com/investing/markets/markets-news/newswire/8507270/trustpoint-xposure-the-first-aeo-certified-pr-agency-in-the-united-states-releases-findings-showing-that-strong-google-rankings-frequently-do-not-translate-into-ai-search-citations-and-explains-why-seo-and-aeo-require-different-strategies/",
          "position": 6
        },
        {
          "title": "How are you measuring brand visibility in ChatGPT and Perplexity for clients?",
          "host": "reddit.com",
          "url": "https://reddit.com/r/aeo/comments/1thjsyo/how_are_you_measuring_brand_visibility_in_chatgpt/",
          "position": 7
        },
        {
          "title": "How to Optimize a Website to Get Cited by ChatGPT and Perplexity",
          "host": "localblitz.com",
          "url": "https://localblitz.com/how-to-optimize-a-website-to-get-cited-by-chatgpt-and-perplexity-6-step-guide/",
          "position": 8
        },
        {
          "title": "I checked 30 top websites for AI-crawler readiness",
          "host": "dev.to",
          "url": "https://dev.to/dongdongdev/i-checked-30-top-websites-for-ai-crawler-readiness-the-ones-chatgpt-cites-most-scored-worst-327",
          "position": 9
        },
        {
          "title": "How to Get Your Site Cited in ChatGPT and Perplexity Answers?",
          "host": "geniusfirms.com",
          "url": "https://geniusfirms.com/insight/how-to-get-your-site-cited-in-chatgpt-and-perplexity-answers/",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account run in a fresh Perplexity conversation. The observed source ranking and answer details changed from repeat 3 despite the same question and visible mode.",
      "industryVertical": "ai-search"
    },
    {
      "id": "evidence/perplexity-measure-chatgpt-citations-repeat-3",
      "content_type": "evidence",
      "canonical_url": "https://seogeo.org/evidence/perplexity-measure-chatgpt-citations-repeat-3",
      "published_at": "2026-11-15T10:38:09.000Z",
      "updated_at": "2026-11-15T10:38:10.000Z",
      "revision": 2,
      "slug": "perplexity-measure-chatgpt-citations-repeat-3",
      "panelId": "measure-chatgpt-citations",
      "panelVersion": "1.0",
      "collectionRound": "2026-11-15",
      "captureCompleteness": "complete",
      "series": "perplexity-measure-chatgpt-citations",
      "repeatRun": 3,
      "repeatTotal": 3,
      "question": "How should a new website measure citations in ChatGPT and Perplexity?",
      "product": "Perplexity",
      "observedAt": "15 November 2026, 18:10 CST",
      "locale": "en-US (English answer; zh-CN interface)",
      "surface": "Perplexity Search with web search enabled",
      "model": "Perplexity automatic model (model version not exposed)",
      "responseUrl": "https://www.perplexity.ai/session/c9e27a41-9a8d-48ec-9153-b0a3e9761335",
      "mentions": "Cited",
      "summary": "Perplexity again recommended a fixed prompt panel, three runs per prompt and separate citation, mention, referral and conversion measures.",
      "answerSnapshot": "For a new site, Perplexity recommended a fixed set of 20–30 customer questions grouped by unbranded discovery, long-tail questions and branded evaluation. It advised fresh conversations with consistent mode, language and location, followed by three runs per prompt and a recorded answer, mention, direct citation, exact cited URL, competitor and accuracy fields. It defined citation rate as answers linking to the domain divided by valid tested answers and distinguished that from brand mentions, distinct cited pages, AI referral sessions and referral conversions. It also said crawler activity is diagnostic rather than proof of citation and recommended preserving a baseline before changing the prompt panel.",
      "citations": [
        {
          "title": "How to Measure Your AI Citation Rate Across ChatGPT, Perplexity, and Claude",
          "host": "freecodecamp.org",
          "url": "https://freecodecamp.org/news/how-to-measure-your-ai-citation-rate-across-chatgpt-perplexity-and-claude/",
          "position": 1
        },
        {
          "title": "How to Measure If ChatGPT and AI Overviews Cite You",
          "host": "denverseo.io",
          "url": "https://denverseo.io/blog/measure-ai-citations/",
          "position": 2
        },
        {
          "title": "How to Get Cited by ChatGPT and Perplexity (and Track It)",
          "host": "websitevikreta.com",
          "url": "https://websitevikreta.com/blog/educational/cited-chatgpt-perplexity-strategies",
          "position": 3
        },
        {
          "title": "Track Sources Cited by ChatGPT and Perplexity: A Practical Guide",
          "host": "maxaeo.ai",
          "url": "https://maxaeo.ai/blog/track-sources-chatgpt-perplexity/",
          "position": 4
        },
        {
          "title": "Trustpoint Xposure AEO study release",
          "host": "theglobeandmail.com",
          "url": "https://theglobeandmail.com/investing/markets/markets-news/newswire/8507270/trustpoint-xposure-the-first-aeo-certified-pr-agency-in-the-united-states-releases-findings-showing-that-strong-google-rankings-frequently-do-not-translate-into-ai-search-citations-and-explains-why-seo-and-aeo-require-different-strategies/",
          "position": 5
        },
        {
          "title": "The New Organic Growth Stack: Search, Social, Communities & AI",
          "host": "searchenginejournal.com",
          "url": "https://searchenginejournal.com/new-organic-growth-stack-search-social-communities-ai/594542/",
          "position": 6
        },
        {
          "title": "How are you measuring brand visibility in ChatGPT and Perplexity for clients?",
          "host": "reddit.com",
          "url": "https://reddit.com/r/aeo/comments/1thjsyo/how_are_you_measuring_brand_visibility_in_chatgpt/",
          "position": 7
        },
        {
          "title": "How to Optimize a Website to Get Cited by ChatGPT and Perplexity",
          "host": "localblitz.com",
          "url": "https://localblitz.com/how-to-optimize-a-website-to-get-cited-by-chatgpt-and-perplexity-6-step-guide/",
          "position": 8
        },
        {
          "title": "I checked 30 top websites for AI-crawler readiness",
          "host": "dev.to",
          "url": "https://dev.to/dongdongdev/i-checked-30-top-websites-for-ai-crawler-readiness-the-ones-chatgpt-cites-most-scored-worst-327",
          "position": 9
        },
        {
          "title": "How to Get Your Site Cited in ChatGPT and Perplexity Answers?",
          "host": "geniusfirms.com",
          "url": "https://geniusfirms.com/insight/how-to-get-your-site-cited-in-chatgpt-and-perplexity-answers/",
          "position": 10
        }
      ],
      "limitations": "Authenticated free-account run in a fresh Perplexity conversation. Perplexity did not expose its automatic model version or retrieval index; the answer and source set can vary by session, account state and date.",
      "industryVertical": "ai-search"
    }
  ]
}
