Can ChatGPT See Your Practice? How to Check.
You do not treat a symptom you have not confirmed. Before you order a test, you have a specific question in mind and a specific reason a particular answer would change what you do next. Most practices approach AI visibility the opposite way: they assume ChatGPT either knows about them or it does not, without ever actually asking it, and they never look closely enough at the answer to know which problem they actually have.
That is worth fixing, because you cannot tell from the outside whether an AI assistant has never encountered your practice, has encountered outdated or wrong information about it, or knows about you but has nothing worth quoting. Those are three different problems, and each one calls for a different fix. Guessing which one you have wastes the fix.
Run the test the way a patient would
Open ChatGPT, Perplexity, or Gemini in a normal browser tab, not a specialized SEO tool, and ask it the way a patient actually would. Five prompts cover most of what you need:
- “Who is a good [your specialty] in [your city]?”
- “Tell me about [your practice name].”
- “Does [your practice name] take [a specific insurance you accept]?”
- “What does [a common procedure or visit type] cost at [your practice] or nearby?”
- “I need a [specialty] near [your neighborhood or nearby landmark], any recommendations?”
Ask each one in a fresh conversation, not a continuation of the last, since a long conversation lets earlier answers bias later ones. Do this on more than one assistant. ChatGPT, Perplexity, and Gemini each pull from a different mix of sources, so a clean answer on one and silence on another is itself useful information, not a contradiction to explain away.
Reading the answer: three outcomes, three different problems
Absence. The assistant names two or three other practices and never mentions you, or it gives a generic answer with no names at all. This is the most common result and the easiest to misread as a ranking problem. It usually is not. An AI assistant answering a local question typically draws from a short list of businesses with strong, current review and listing signals, not a long list the way a Google Maps search shows one. If you are not on that short list, you are not competing for eighth place, you are simply not being considered. The fix here is the same review and directory work that already matters for your Google Business Profile: current, accurate listings on the sites patients and search systems both trust, and reviews that keep arriving steadily rather than in one burst two years ago.
Misinformation. The assistant names you, but gets something wrong: an address you moved from three years ago, a doctor who left the practice, insurance you stopped accepting, hours that are off by a day. This is worse than absence, because a wrong answer sends a real patient to call a closed line or ask about coverage you do not offer, and it happened because it found information about you that you did not put there and cannot directly edit. Search your practice name plus the wrong detail and see which third-party site is the likely source: an old directory listing, a stale profile on an insurance site, an outdated news mention. You correct what you can control directly (your website, your Google Business Profile) and you contact the third-party site to correct the rest. This takes longer than it should. Start anyway.
Presence without substance. The assistant names you correctly and gets the facts right, but the answer is thin: your name in a list with no detail, while a competitor gets a sentence about what they specialize in or what patients say about them. Nothing is wrong here, but nothing is making the case for you either. This points back to your own site: does it state, in plain readable text, what you actually do and treat, and does it have current, specific reviews for an assistant to draw on when it wants to say something concrete rather than just list a name.
What this audit does not tell you
A clean answer today does not mean you are safe next month. These assistants answer from a mix of what they learned in training and what they can find live on the web right now, and a local business’s standing in that mix can shift as reviews accumulate elsewhere or a competitor’s listings improve. Treat this as something you check periodically, the same way you would not assume a chart is current just because it was accurate at the last visit.
It also will not tell you why an assistant chose someone else over you when both of you look similarly qualified on paper. The honest answer is often unglamorous: the other practice has more recent reviews, a more complete profile, or a website that states its services in plain text instead of behind a design element the assistant cannot read. That’s a checklist to work through, not a mystery to solve.
What to do Monday morning
Run the five prompts above on two different AI assistants and write down, for each one, which of the three outcomes you got: absent, wrong, or thin. Absence sends you to your reviews and directory listings. Misinformation sends you to find and correct the source, starting with your own site and Google Business Profile. Thin presence sends you back to your own service pages, asking whether they actually say, in plain text, what a patient would need to hear to choose you.