AI visibility
Which engines are tested, how appearance is measured across reruns, and what is deliberately not stored.
Which engines are tested
Perplexity Sonar and Grok, both grounded-answer engines — and only those. Every screen names the engines a particular run actually reached, because the list can differ from run to run.
This is not a measure of ChatGPT, Claude, Gemini, or any other assistant. Those are not probed, nothing here says anything about them, and a module claiming "visibility across AI assistants" would be claiming coverage nobody measured.
How appearance is measured
Each question is asked several times on each engine, because a generative answer differs between identical requests. What is reported is the share of runs your clinic was named in — an appearance rate, not a position.
There is no rank, and there will not be one. A rank across reruns of a generative answer is noise with a number on it, and the provider itself declines to report one.
Rates are shown as bands — not named, named occasionally, named in most answers, named consistently — rather than percentages. Across five runs, one extra appearance moves the figure twenty points, and showing that as a change would be reporting sampling as progress.
When a rate is withheld
If too few runs completed, no rate is shown at all. Three runs can only produce 0, 33, 67 or 100 per cent, and any of those read as a visibility rate implies a precision the sample cannot carry.
What a zero means
A clinic named in none of the runs is a real and useful finding. But matching is done by looking for your clinic's name in the answer text, so a clinic written a different way than we hold it can read as absent. The matching method is shown beside every reading for exactly this reason.
Cited domains
Alongside the appearance rate, you get the domains those engines quoted, and which of them are yours. This is the actionable half: you cannot make an answer engine mention you, but you can see that the sources it quotes are directories you have no presence on.
Your monthly allowance
Each plan includes a number of prompts per month. Answer-engine probes are the most expensive thing ClinicRadar does, so a run is trimmed to what your allowance leaves rather than refused outright — the questions you marked conversion-critical are measured first, and the screen tells you how many were deferred.
What is not stored
The generated answers themselves are not kept. They are long passages of third-party prose naming businesses that are mostly not yours, and storing them would put a pile of unverified claims about other clinics inside your workspace. What is kept is the appearance rate, the runs behind it, the cited domains, and the passages that named you.
What this does not claim
It does not attempt to influence answer-engine output. An engine's description of your clinic is a record of what it said on a date, not a statement of fact, and ClinicRadar makes no clinical claim of any kind.
Was this page helpful?