How AI answers divide their sources
A preview framework for comparing source types across questions, products and repeated observations.
10 questions · 3 products · 30 observationsSource mix · Citation behavior · MethodRead reportResearch
Explore studies of source mix, AI citation behavior and answer changes. Every report connects its findings to a defined sample, method version and visible limitations.
A preview framework for comparing source types across questions, products and repeated observations.
10 questions · 3 products · 30 observationsSource mix · Citation behavior · MethodRead reportWhy dates, prompts and source context determine what a single citation observation can support.
Method note · 4 examplesEvidence · Measurement · LimitationsRead reportGEO research questions
Short answers with links to the evidence, definitions and limits behind them.
A defensible visibility study can measure observed mentions, linked citations, source types and changes across repeated questions. It should not present a single opaque score or imply that visibility is the same as search ranking.
A source mix describes the proportions of source types cited across a defined set of observations. It is meaningful only when the questions, products, locale and method version remain comparable.
No. Research can show what happened in a sampled context and identify patterns worth testing. It cannot guarantee how a live AI product will answer a future question.