Buyers ask ChatGPT for recommendations every day. We run the actual probe and show you who gets named in your market.
For twenty years the question was "where do we rank". A buyer typed something into Google, scanned ten blue links and picked one. That habit is splitting in two. A growing share of people now describe their problem to an assistant and accept the two or three names it hands back, and those answers appear inside Google's own results, in Bing and Copilot, in the ChatGPT app, and increasingly inside the tools people already have open at work.
AI visibility is whether a model names your business when someone asks it for a recommendation in your category and city. It is a harsher outcome than ranking. Position eight on Google still gets seen by someone scrolling. Being absent from a three-name answer is complete invisibility, and the buyer never learns you existed.
This is the part that surprises owners who have paid for SEO for years. Ranking and being recommended run on different fuel.
Google decides positions largely from links and on-page relevance: who points at you, with what words, and whether your page answers the query. A language model has no live ranking table. It answers from how businesses in your category have been described in text across the sources it learned from and the pages it can cite live. If nobody has written a sentence describing what you do, in a place worth citing, the model has nothing to repeat, however well your own site is optimised.
The common reasons a business never gets named:
None of this is a trick, and there is no tag you can add to your homepage that makes a model like you. The work is to make your business easy to describe and easy to verify.
Decide the exact phrasing that should come back: what you do, who for, where, and one concrete differentiator. Then put that sentence, near enough word for word, on your homepage, your about page, your Google Business Profile description, and every directory profile you control. You are giving the model a consistent line to quote instead of asking it to invent one.
Find the pages a model would reasonably lean on for your category and city: the local business journal, the chamber listing, the "best X in Y" round-ups that already rank, your trade association's member directory, an active local subreddit thread. Being described on those pages moves the needle far more than another backlink from a generic directory. One paragraph in a local round-up is worth more than fifty low-quality links.
Ask satisfied customers to name the service they came for and the city, in their own words. "Great service" is invisible. "They did my daughter's Invisalign in Austin and explained every step" is a sentence a model can use, and it is the kind of specificity that shows up in an answer.
LocalBusiness schema with your name, address, phone, opening hours and services helps machines resolve you into a single confident entity. It is genuinely worth doing once and it is table stakes rather than an advantage. It will not, by itself, get you named.
Pages that answer a real question plainly, with prices, timeframes and specifics, are the pages models quote. A page called "How much does Invisalign cost in Austin" that gives an actual range beats a page called "Invisalign services" that gives none.
Answers move. Re-run the probe monthly, and run it on your two strongest competitors as well: knowing who currently owns the answer tells you which sources to go and get described on.
We take what you do and where you are and turn it into the question a buyer would actually ask, for example "Who is the best dentist in Austin?". That question goes to a live AI model. We read back the businesses it names, then check whether your domain or business name appears among them. You see the real answer, not a score we invented.
Two honest caveats. Model answers vary between runs and between providers, so treat a single result as a strong signal rather than a verdict, and re-check over time. And a market where the model refuses to name anyone is not a failure: it means nobody owns that answer yet, which is the easiest possible position to attack.
| Google ranking | AI recommendation | |
|---|---|---|
| Decided by | Links, relevance, technical health | How consistently you are described across sources |
| Result shape | Ten positions, scrollable | Two or three names, no scroll |
| Being eighth | Some traffic | No mention at all |
| Fastest lever | Fix technical issues, earn links | Get described on pages that already get cited |
| Feedback speed | Weeks to months | Answers can shift within weeks in thin local markets |
Yes, constantly. Ask it for the best dentist, plumber or law firm in a city and it will name specific businesses. Those answers now sit inside Google search results, Bing, Copilot and the ChatGPT app itself, so being named is closer to being ranked than most owners realise.
We send the actual buying question a customer would ask, for example "Who is the best dentist in Austin?", to a live AI model, then read the businesses it names back. There is no simulation and no scoring formula: you see the real answer and whether your domain or business name appears in it.
Ranking and being recommended are different problems. Models lean on how often a business is described in text across the web, in reviews, directories, local press and forum threads, rather than on the link signals that move Google positions. A site can sit at position three and still never be described anywhere the model learned from.
It overlaps but it is not identical. Classic SEO optimises a page for a query. AI visibility optimises how your business is described across the web so a model has something concrete to repeat. The work that helps most, earning mentions on pages that already get cited, is useful for both.
Faster than most people expect for a local business, because local markets are thin. When a model has only a handful of sources describing dentists in one city, a few new well structured mentions can change the answer. National and competitive categories take much longer.
Yes, three checks a day with no account. We email you a copy of the result and you can share the link or download the card. Nothing is charged and no card is asked for.