Local SEO · 10 min read · Published November 2026

How Dentists Get Recommended by ChatGPT: What an 18-Answer Test Actually Showed

We asked ChatGPT, Google's AI Overview and Gemini to recommend dentists 18 times and recorded which of the 10 practices in our Logan, Utah pilot sample (November 2026) they named. Half were never named once. Here are the findings, what the answers reveal about how AI tools pick practices, and the playbook that follows.

1. The Test: 10 Practices, 18 AI Answers

Patients have started asking AI assistants the question they used to type into Google: "Who is the best dentist near me?" ChatGPT, Google's AI Overview and Gemini now answer with a shortlist of named practices. If your practice is not on that list, you do not exist for that patient — no matter how good your website is.

Rather than guess how those shortlists get built, we ran a small, honest test. This article shares what we found in our 10-practice Logan, Utah pilot sample (November 2026), what the answers themselves reveal about how AI tools pick practices, and the playbook that follows from the evidence.

We took a sample of 10 dental practices in one city — Logan, Utah — spanning general dentistry, orthodontics, endodontics, periodontics and pediatric dentistry. We then asked three AI tools (ChatGPT, Google's AI Overview and Gemini) six questions a patient might ask.

The six questions were: Who is the best dentist in Logan, UT? Who is a good family dentist in Logan, UT? Who is a good orthodontist in Logan, UT? Who offers dental implants in Logan, UT? I have a toothache — who is an emergency dentist in Logan, UT? And who is a good dentist near Utah State University in Logan, UT?

That is 18 AI answers. We recorded every practice named, in order. Separately, we ran the five core searches on Google ("dentist in Logan, UT", "best dentist in Logan, UT", and so on) and recorded which of the 10 practices appeared in the Map Pack or the top 30 results. Everything below comes from those two passes. It is a pilot sample of 10 practices in one city — treat the findings as directional, not as universal laws.

Key takeaway: A 10-practice Logan, Utah pilot (November 2026): six patient questions put to ChatGPT, Google's AI Overview and Gemini — 18 answers, every named practice recorded — plus a Google Map Pack and top-30 pass over the same 10 practices.

2. What We Found: Invisibility, Reviews, and Three Different Shortlists

Half the practices were invisible to AI. Only 5 of the 10 sampled practices were named by any AI tool, in any of the 18 answers. The other five — real practices with Google listings and reviews — were never recommended once.

Reviews alone did not decide it. One practice in the sample has 473 Google reviews and a perfect 5.0 rating. It received zero AI mentions across all 18 answers. The most-reviewed practice in the sample (732 reviews, also 5.0 stars) was named exactly once. Review volume is clearly part of the picture — Google's AI answers displayed star ratings and review counts next to the practices they named — but a big review count by itself did not buy a recommendation. One caveat: two of the high-review practices in our sample are specialists, and our six questions included no root-canal question, so specialists were partly tested outside their specialty. The 473-review practice, though, is a general practice — and it was still shut out.

The three AI tools do not agree with each other. Each tool produced a different shortlist for the same question. The most-named practice in the sample earned most of its mentions from a single tool while being nearly absent from the others. Being "visible to AI" is not one thing — it is at least three separate tests, and a practice can pass one and fail the rest.

Key takeaway: Only 5 of 10 practices were named at all; a 473-review, 5.0-star general practice got zero mentions; and each AI tool produced a different shortlist for the same question.

3. What We Found: Different Games, Missing Websites, and Invented Facts

Google search and AI answers are different games. Across the five core Google searches, only 1 of the 10 practices — the orthodontist, on the orthodontist search — appeared in a Map Pack at all. Meanwhile the AI tools named practices that Google search could not find: an implants specialist in the sample does not appear in the Map Pack or the top 30 results for "dental implants in Logan, UT", yet ChatGPT named it second for the implants question. Winning Google does not automatically win AI, and losing Google does not automatically lose AI.

A missing website did not disqualify a practice — but it is still a hole. That same implants specialist has no website link on its Google listing at all (the listing literally shows "Add website"), and ChatGPT recommended it anyway. So AI tools will surface a practice from listing and review data alone. But note the other half of that finding: the practice is invisible in Google search for its own core service, and a patient who wants to check it out has nowhere to land. AI found it despite the gap, not because the gap is fine.

AI answers can be wrong about you — check what they say. In the emergency-dentist answer, ChatGPT displayed a phone number containing "555" — the pattern reserved for fictional numbers — for one practice. In other words, the AI confidently showed patients a number that does not work. If you never ask the AI tools about your own practice, you will never know what they are telling your patients.

Same brand, wrong practice. The AI tools repeatedly recommended a different location of a multi-location brand — not the location in our sample. If you run or belong to a multi-location practice, each location's online identity has to stand on its own, or the AI will happily send patients to the wrong office.

Key takeaway: Map Pack visibility and AI visibility diverge; a practice with no website link was still recommended by ChatGPT; one AI answer showed a non-working 555 phone number; and multi-location brands risk AI sending patients to the wrong office.

4. What the AI Answers Actually Pull From

We cannot see inside these tools, and anyone who claims an exact recipe is guessing. But the 18 answers themselves show their workings.

Listing data. Google's AI answers printed each named practice's rating, review count and address — the contents of a complete, accurate business listing.

Review substance, not just counts. ChatGPT's answers described practices in patient-experience language — gentle, good with kids, easy to reach, upfront about prices. That language mirrors what patients write in reviews, which is one reason the content of your reviews matters, not only the total.

Service clarity. Answers matched practices to the specific service in the question — implants, orthodontics, emergencies. Practices were described by what they clearly offer, in plain words.

The open web. The answers blended listing facts with descriptions that only exist on practice websites and directories — names, services, positioning. A practice that says clearly what it does, and where, gives the tools something accurate to repeat.

Key takeaway: Observed in the answers themselves: complete listing data, the substance of reviews, plainly stated services, and consistent facts across the open web. No exact recipe exists — anyone claiming one is guessing.

5. The Playbook: What to Do, in Order

Step 1 — Complete your Google Business Profile like it is your homepage. Correct primary category (the implants specialist above is listed under a precise "dental implants periodontist" category, and it surfaced for the implants question), every service listed, hours, photos — and a working website link. Our Google Business Profile guide walks through the full setup.

Step 2 — Give each high-value service its own clear page. Implants, orthodontics, emergency care, Invisalign: one page each, in plain language, saying who it is for and what it costs in shape if not in numbers. In our Google pass, the sampled practices that did show up in results for service searches were showing up with dedicated service pages. AI answers describe services in plain words — give them plain words to work with.

Step 3 — Make your name, address and phone identical everywhere. Your listing, your website, directories, your socials. Inconsistent details are how AI tools merge you with another location — or invent details, like a 555 phone number, to fill a gap.

Step 4 — Build reviews steadily, and respond to them. Keep asking, keep the flow recent, and reply. But take the review finding seriously: reviews are necessary groundwork, not a guarantee. Pair them with the steps above instead of treating the count as the whole strategy. Our guide on how many Google reviews a dental practice needs explains how to find your real local benchmark.

Step 5 — Let the crawlers in and structure your site. Make sure AI crawlers are not blocked and your site carries basic structured data (schema) identifying your practice, services and location. Our dental website crawler and schema checklist covers this in order. Honest note from our sample: no single technical factor explained the winners — the practices named most often had the basics stacked together, not one magic item.

Step 6 — If you have multiple locations, build each one separately. Separate listing, separate reviews, separate service pages per location. Otherwise, as the wrong-location finding showed, the AI recommends whichever location's data is strongest and sends your patients there.

Step 7 — Test yourself the way a patient would, every month. Open fresh sessions in ChatGPT, Google and Gemini. Ask the six question types above for your city. Write down who gets named — and check every fact the tools state about your practice, especially your phone number. Visibility you never measure is visibility you are guessing about.

Step 8 — Don't neglect the Map Pack while you chase AI. Only 1 of our 10 practices appeared in any Map Pack — that is paid-attention territory you are also probably missing. The factors are different but the groundwork overlaps; our breakdown of what decides the Maps 3-Pack for dentists covers it.

Key takeaway: In order: a complete Google Business Profile, a clear page per high-value service, identical name, address and phone everywhere, steady answered reviews, crawler access and schema, a separate identity per location, monthly self-testing, and the Map Pack basics underneath.

6. Find Out Where Your Practice Stands

Everything in this article started with one question: does AI recommend us at all? LocoraAI's free checkup answers that for your practice — it audits your Google presence and your AI visibility together and shows you the gaps in the order this playbook fixes them.

For the deeper mechanics, read next: AI Visibility for Dentists, on how ChatGPT decides which practices to recommend, and our step-by-step guide to getting your dental practice recommended by ChatGPT. More guides for dental practices live on our page for dentists.

Key takeaway: Start by finding out whether AI recommends your practice at all — the free checkup audits your Google presence and AI visibility together and orders the gaps the way this playbook fixes them.

Frequently asked questions

How do I check whether ChatGPT recommends my dental practice?

Open ChatGPT in a fresh, logged-out session and ask the questions a patient would ask — "best dentist in [your city]", "emergency dentist in [your city]", and the same for your highest-value services. Repeat in Google's AI Overview and Gemini, because the three tools give different shortlists. Do it monthly, and note every detail the tools state about you; in our pilot test, one AI answer displayed a fabricated 555 phone number for a practice.

Do more Google reviews guarantee that AI will recommend my practice?

No. In our 10-practice Logan, Utah pilot sample (November 2026), a general practice with 473 reviews and a 5.0 rating received zero mentions across 18 AI answers, and the sample's most-reviewed practice (732 reviews) was named once. Reviews matter — AI answers display ratings and counts — but they work together with listing completeness, service clarity and your wider web presence, not instead of them.

Can AI recommend a practice that has no website?

It happened in our test: a periodontics practice with no website link on its Google listing was named second by ChatGPT for the dental-implants question. But the same practice was invisible in Google search for that service, and patients had nowhere to land. Treat an AI mention without a website as luck, not a strategy — a complete listing and a real website remain the groundwork.

Why does AI keep recommending a different location of our brand?

Because the tools build each answer from the strongest, clearest data they can find, and locations with thin or merged data lose to the location with the complete listing, the most reviews and the clearest pages. Give every location its own complete listing, its own review flow and its own service pages so each one can be recommended on its own merits.