AI search field guide
AI visibility metrics: formulas, denominators and failure handling
A metric is useful only when you can name the observations included in it. Define mention rate, citation rate, competitive share and position before comparing results.
Mention rate
Mention rate is eligible answers naming the tracked brand divided by eligible answers, multiplied by 100. With 12 brand mentions in 48 eligible answers, the rate is 25%. Preserve brand aliases and the question cohort. A repeated brand name within one answer does not need to count as multiple answer-level mentions. Report failed checks separately from absence.
Citation rate
Citation rate uses answers with complete citation evidence. If the owned domain appears in six of 30 citation-complete eligible answers, the rate is 20%. Partial or unavailable citation extraction cannot establish that the brand was not cited. InfiniteRanks keeps citation completeness explicit. Domain and URL grouping answer different questions; a subdomain may need separate ownership review.
Competitive share and position
The competitive share calculation uses the tracked-brand mention set as its denominator. It is affected by which competitors are included and is not total market share. Average position uses answers with a detected position; an absent brand does not have a rank of zero. Inspect the actual list or wording before treating textual order as a recommendation hierarchy.
Publish a measurement note
State the prompt count, successful observations, dates, country, language, model, collection method and exclusions. Show absolute counts beside rates. Keep a stable cohort and disclose changes. A chart without this note can hide model changes, prompt edits or missing collection, making a precise-looking percentage less informative than the underlying answer records. Check the API-versus-browser guide before combining observations from different surfaces.
Sources and further reading
- Google Search Central: AI features and your website
- Google: creating helpful, reliable, people-first content
Sources explain the underlying standards. They do not endorse InfiniteRanks or guarantee a ranking outcome.