How to be mentioned first by AI, and what to fix on your site
Write ten to thirty questions in your customers' own words, naming your category and city but never your business. Put every one of them to ChatGPT, Gemini, Claude and Perplexity from logged-out sessions, and record three things each time: which businesses were mentioned, which sources were cited, and the total number of answers you asked for. How often you are mentioned — and how often you are mentioned first — is what you are measuring. Then fix the two things that decide it: what your own website plainly says about what you do and where, and whether the sources those engines already quote mention you at all.
The five steps
- Write buyer questions, never brand questions. Cover the whole journey: who is best, who is cheapest, who is open now, who handles one specific need.
- Ask every engine from a clean session, logged out or with memory off, so the answer is not personalized toward you.
- Record names, cited sources and the total asked, so what you end up with is a rate and not an impression.
- Check what your own site gives them to read: robots.txt, structured data, category, location, services, hours and contact details.
- Re-run the identical set later and compare. The change between runs is the finding, not the raw number in either one.
What makes a measurement trustworthy
- A denominator. 'ChatGPT mentioned us' means nothing without how many times it was asked and did not.
- Repeats of the same question. Engines are probabilistic; asking each question more than once separates a stable position from a coin flip.
- Merged names. The same business appears as a trading name, a legal name and a branch; counted separately, every competitor looks smaller than they are.
- Geography that holds. An answer naming a business in another country is not a competitor, and quietly inflates the list it appears on.
- The raw answers kept. If a figure cannot be traced back to the answer and timestamp it came from, it cannot be checked.
What to fix on your own site
Absence is rarely a penalty. It is usually a lack of quotable, corroborated text: the engine has nothing plain to read about what you do and where, and nobody independent has written it down either. So the fixes are ordinary and specific — say your category, city and services in plain text on your own pages, keep structured data, hours and contact details accurate, stop blocking assistant crawlers, and get mentioned on the directories, review sites and local pages the engines already quote for your category.
A measured report gives you both halves at once: the sources behind the answers in your market, and a readiness check on your own site with the specific change to make in each place. Nobody can guarantee an AI position — but you can make yourself far easier to mention first.
Common questions
What is AI visibility?
How often AI answer engines name your business when someone asks them for a recommendation in your category and area. It is not a ranking on a page of links — the engine returns a short list of names, so you are either in it or you are not.
How do I monitor AI search visibility over time?
Fix your question set and never change it, then re-run the identical set at a regular interval — monthly is plenty for most businesses. Record the mention rate, the businesses named ahead of you, and the sources cited each time. The comparison between runs is the signal; a single run is a snapshot of something that moves.
How often should I re-check?
Monthly or quarterly. Checking daily mostly measures the models' own randomness, and any change you make to your site or your presence on third-party sources takes weeks to show up in what the engines quote.
How do I improve my AI searchability?
Make your business easy to describe and easy to corroborate. State category, city and services plainly in text on your own site, keep structured data and opening hours accurate, don't block assistant crawlers in robots.txt, and get named on the independent sources the engines already cite for your category — directories, review sites, local press and forum threads.
Does blocking AI crawlers hurt my visibility?
It can, badly. A robots.txt rule that blocks assistant crawlers stops the engines reading your site directly, so anything they say about you comes only from third parties. Check what your robots.txt actually blocks before assuming it is fine.
Read a complete report, then decide.
A full released run is free to read — no sign-in, nothing withheld.