Short, working definitions — written for marketers who have to explain the difference between being mentioned and being cited.
Whether an AI answer engine names a brand when a buyer asks a category question, and whether it points to the brand’s own pages as the source. It is measured separately from search rankings because the answer engine can name a brand without citing it.
The assistant layer that returns a written answer instead of a list of links — for example ChatGPT, Perplexity, Gemini, Copilot. For marketing measurement the relevant output is the answer text, the named brands and the cited sources.
The work of making a brand likely to be named and cited in AI answers: publishing pages that answer real buyer questions, structuring the content so it can be quoted, and earning mentions on the pages the engine already trusts.
Overlaps with AEO; usually refers to optimizing for generative engines specifically (the multi-step retrieval-and-generation stack), including how sources are retrieved and how passages are quoted.
The AI answer names the brand in its text. A mention is what most “AI visibility” screenshots show. It does not mean the brand owns the evidence behind the claim.
The AI answer points to a page as the supporting source. Citation ownership is the harder and more durable metric: when the engine cites the brand’s own site, the brand controls the evidence.
The share of answers for a question set where the brand’s own domain is cited, compared with three named competitors. Reported as a fraction, e.g. 1/3 answers.
A list of the pages that won the citation for each buyer question, traced to host and page. It is the working document for content decisions: it shows which comparison pages, docs pages or review pages the engine trusts.
A single logged-out run of a question set, used as a sample. It is not a verdict: answers vary between sessions, which is why the paid diagnostic repeats each question three times across three days.