We ask the assistants your customers' questions, several times each, and report the share of answers that name you — with the interval that sample size honestly supports, per platform and per market.
Measure my domain — free Run the AI-readability check
First measurement free · no card · results in minutes · the method
In our published audit one platform gave a different citation on around 40% of repeated identical questions. Ask once and you are reading the platform's mood, not your visibility.
The same store, the same week, measured 59.5% cited on a category question list and 40.5% on its own Search Console queries. A tool that picks the questions for you is choosing your score.
An answer engine does not rank ten links; it names two or three sources inside a paragraph. What you need is the share of answers that name you, and how wide the uncertainty around that share is.
Twelve to sixty buying questions per market, written natively in the language — or, for the own-question measurement, drawn from your Search Console with brand terms removed.
ChatGPT, Google AI Overviews, Google AI Mode, Gemini and Perplexity, each question asked two or three times. We record who is cited, the full answer, and the source pages.
Per platform and per market: the share of answers citing you, drawn on a track with its interval. Wide band, volatile platform. Two bands that overlap are not a gap, and the report says so.
Every domain cited across the answer pool, ranked, with your position; the worst-performing questions first; every observation exportable.
Five assistants, each with its own rate and interval, because they do not behave alike — the audit found one citing on nine answers in ten and another on one in three.
Up to three of six markets — English, German, Spanish, French, Italian, Dutch — with questions written in each language, not translated.
The competitors and publishers named when you are not, ranked by frequency. The list your next quarter of pages comes from.
The URLs behind every answer. Getting onto those pages, or replacing them, is what moves the rate.
Monthly re-measurement judged by a two-sided Fisher exact test at the sample size you actually have. No trend line through noise.
One CSV row per question, platform and repeat, and a client-ready print layout. Nothing behind the figure is hidden from you.
Method and data are open: the measurement study · the audit · the provenance study.