Shared definitions prevent teams from drawing different conclusions from the same chart.
Discovery and retrieval
AI search is a useful umbrella term for experiences that answer questions while incorporating retrieved information. An answer engine is the product or experience providing the answer. Retrieval means obtaining material used in a task; it is different from the model’s previously learned information.
Query fan-out describes expanding a question into related searches or subquestions. A citation is a source link shown with an answer. A visible citation may help investigate a claim, but the displayed links are not necessarily a complete record of every influence on the answer.
Brand measurement
A mention names a qualifying brand or product under your detection rules. Visibility is a defined proportion of observed answers containing such mentions. Share of voice compares your qualifying mentions with a stated total, often within a selected competitor set. Always state the denominator.
Position describes order where the answer actually provides an ordered recommendation. Sentiment describes evaluative language; accuracy asks whether checkable claims are correct. A positive description and a correct description are different dimensions.
Traffic and commercial outcomes
A referral is an identifiable visit arriving from another source under your analytics rules. A conversion is a defined event, such as a completed purchase or qualified lead. Attribution assigns credit using a chosen model; it does not automatically establish causality.
Incrementality asks what changed because of an intervention compared with what would otherwise have happened. ROI compares a defined return with a defined investment. Visibility is not revenue, and attributed revenue is not necessarily profit or incremental revenue.
Research and interpretation
A prompt panel is a selected set of questions. A baseline is the reference observation used for comparison. A hypothesis is a proposition to test, not a result. A controlled experiment uses an appropriate comparison design; a before-and-after observation has weaker protection against alternative explanations.
A citation gap is an observed absence of relevant source representation relative to a chosen comparison. A brand gap is an observed omission of a qualifying brand. Neither term identifies the cause by itself. Keep these working definitions in the report and adapt them explicitly when your method changes.
