AI Answer Stability | How often AI engines name your brand

Mentions by run

One column per answer, grouped by engine. The number is the brand’s position in that answer; blank means it was not named. Arrow keys move between cells.

Mention rate with 95% interval

Per engine, for your brand. A wide band means more runs are needed before the number is worth reporting.

Cited sources

How many answers cited each site. A site cited in most runs is a stable source; one cited once is noise.

How this is measured

Each engine gets the same question several times. A run is one answer. Mention rate is the share of answers that name the brand; named first is the share where it is the first option offered.

The 95% interval is a Wilson score interval on the mention rate. With 10 runs a rate of 50% carries an interval of roughly 24% to 76%, which is why a single-run visibility score cannot be compared week to week. An interval wider than 30 points is flagged as too wide to report.

Stability is how much the brands offered overlap from one answer to the next within an engine (mean pairwise Jaccard, 0 to 1). A run that returned no answer is left out of every figure rather than counted as not naming the brand.

Brands you enter are matched directly in the answer text. Other brands are those an answer offers as options for the question, as judged by a language model. Brands mentioned only in passing, and brands named in the question itself, are left out.

These are API answers, collected through each engine's API with web search on. They can differ from what someone sees in the ChatGPT, Gemini, Perplexity or Claude apps. Cited sources are those found by the web search run for these API answers. Google AI Mode is its results page. Runs are set to the UK, except Gemini, which cannot be pinned to a country this way.

Answers are shared for 24 hours: if the same question was checked today, you see those runs and the time they were collected.