Recorded prompt
“How should a new website measure citations in ChatGPT and Perplexity?”
- Product
- Perplexity
- Surface
- Perplexity Search with web search enabled
- Model
- Perplexity automatic model (model version not exposed)
- Locale
- en-US (English answer; zh-CN interface)
- Observed answer
- Open session ↗
- Observed
- 15 November 2026, 17:11 CST
- Fresh conversation
- Yes
What the answer said
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.
10 recorded sources
What this observation cannot show
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.