What AI Engines Say When Patients Ask for a Dentist: Our Data From 62 Practices

When we asked ChatGPT, Claude, Gemini and Perplexity to recommend a dentist for a specific service in each practice's own city, all four engines named the practice only 13% of the time, and none of them named it 40% of the time. In almost half the searches, the engines split: some named the practice and others didn't. And practices with more Google reviews weren't named more often. Practices with a larger web footprint were.

Why we started measuring this

Dentists ask me some version of the same question, often after one of my talks: "When a patient asks ChatGPT for a dentist, does it name me?" The honest answer used to be "let's look." We'd open a chat window, type a question, and see what came back.

One chat window is an anecdote, though, not a measurement. So every month we run the same searches, the same way, for every practice we track, and keep every answer. This article shares what the October 2026 run showed, with the method laid out so you can check our work.

How we ran the test

On October 1, 2026, we ran our monthly AI visibility sweep for 62 dental practices in 35 states. We ran up to six searches per practice, phrased the way a patient types them, "best [service] in [city, state]," using the services each practice told us it wants to grow: dental implants, Invisalign, cosmetic dentistry, sedation, sleep apnea care and so on. We also ran one generic search for every practice, "best general dentist in [city, state]." (Five practices had only the generic search.) The sweep ran October 1; ChatGPT's answers were re-run October 4 after an error.

We sent every search to four engines through their own APIs, with live web search turned on where the engine offers it:

• ChatGPT (OpenAI)
• Claude (Anthropic)
• Gemini (Google)
• Perplexity

A practice counted as named only when the answer included the practice's name or its doctor's name, and we kept the text of every answer so a person could check the match. When an engine returned an error, we left that answer out instead of counting it as a miss, so a glitch can't make a practice look invisible.

That gave us 899 service-search answers, and 224 searches that all four engines answered.

Finding 1: All four engines named the practice in only 13% of searches

For the 224 searches all four engines answered, here's how many engines named the practice:

The middle of that table is the part most dentists miss. In almost half the searches, the practice was named by some engines and not others. A practice that checks ChatGPT, sees its name and stops there has seen one answer out of four, and the patient down the street using Gemini may be getting a different list.

Finding 2: Each engine has its own habits

Across all service searches, this is how often each engine named the practice:

These engines don't read the web the same way. Some lean harder on maps and business listings, some on directories and "best of" lists, some on a practice's own pages. That's why a single fix rarely moves every engine at once, and why we track all of them.

Finding 3: Specific beats generic

The generic "best general dentist in [city]" search named the practice in 25% of answers (62 of 248). Searches for a specific service named it in 35% (312 of 899), and 34% (282 of 820) if you leave out the service searches that were themselves for a general or family dentist.

That matches what we see in practices every month. AI engines answer a specific question with the practice that has clearly and specifically answered it. A practice with a full page, a video and a few posts about implants gives the engine something to point to when a patient asks about implants. A practice with one line about implants on a services list doesn't.

It also means the generic search is the wrong scoreboard. A practice can look invisible on "best dentist in town" and still be the answer on the treatments it cares about most.

Finding 4: Web footprint tracked with being named. Review count didn't.

We split the practices into thirds two ways and compared how often each group was named in service-search answers.

By Google review count(54 practices with a confirmed Google Business Profile):

By web footprint, measured as the number of search terms a practice's website ranks for in Semrush (52 practices with data):

Reviews still matter. Patients read them, and they're part of what makes a practice worth recommending. But in these practices, more reviews didn't go with being named more often. The practices whose websites rank for more search terms were named more often.

This is a correlation in 52 to 54 practices, not proof of cause. It is consistent with how these engines work: they repeat what they can find and read, and a site that answers more questions gives them more to find.

Finding 5: The answers move every month

We compared each engine's answer to the same search on September 1 and October 1. In 17% of those pairs (140 of 833), the answer flipped: a practice that was named wasn't, or the other way around.

So one check, on one day, in one engine, isn't a measurement. The direction over several months is the signal. That's why our reports compare each month to the month before instead of showing a single score.

What this means for your practice

1. Check all four engines, service by service. Ask each one the question your patients ask about the treatments you want to grow, in your city. Write down who gets named.
2. Build one clear answer for every service. A full page, written answer-first, plus the same answer on video and in your posts. Specific questions get specific answers.
3. Keep your facts identical everywhere. Name, address, phone, doctors and services should match on your website, Google Business Profile and directories.
4. Treat reviews as the floor, not the strategy. Keep them steady and recent, then put your energy into content the engines can quote.
5. Measure monthly and watch the trend. Expect some answers to move month to month. Judge the work by the direction across a quarter.

If you want to see these searches run for your own practice, that's what our AI visibility work does every month. My guide to AI search for dentists also includes a 15-minute self-audit you can run today.

The limits of this data

• It's 62 practices, and they all work with us. They're not a random sample of every dental practice in the country, so treat the percentages as a picture of these practices, not a national average.
• It's one run on one day. Finding 5 shows how much single answers move.
• Two engines aren't in these numbers. Google's AI Overviews and Microsoft Copilot returned no usable data in this run.
• Correlation isn't cause. Finding 4 shows what the named practices had in common, not proof of what made them named.

Frequently asked questions

How often does ChatGPT recommend a specific dental practice?

In our October 2026 test of 62 practices, ChatGPT named the practice in 31% of searches for a specific service in that practice's city. Gemini and Claude named it in 40%, and Perplexity in 28%. Results vary by service, city and month, so check your own practice in all four engines.

Do more Google reviews get a dentist recommended by AI?

Not on their own, in our data. Practices with the most Google reviews were named in 35% of AI answers, and practices with the fewest in 38%. Practices with the largest website footprint were named in 44% of answers, compared with 30% for the smallest. Reviews matter to patients, but in our data the size of a practice's web footprint lined up with being named more than review count did. That's a pattern, not proof of cause.

Why does ChatGPT name my practice but Gemini doesn't?

Each engine searches and weighs sources differently. In our test, all four engines named the practice in only 13% of searches. Being named by one engine doesn't mean you're named by the others, so a real check covers ChatGPT, Claude, Gemini and Perplexity, plus Google's AI Overviews.

How should a dental practice measure its AI search visibility?

Run the same patient-style searches, for the services you want to grow, in every major engine, every month, and record who gets named. Count a result only when your practice or doctor is actually named. Compare month over month, because about one answer in six can change between checks.

Sources

• Digital Floss AI visibility sweep, October 1, 2026: 62 dental practices in 35 states; ChatGPT, Claude, Gemini and Perplexity asked through their own APIs with web search on; a practice counted as named only when the answer named the practice or its doctor; engines that returned an error excluded, never counted as a miss.
• Semrush, organic search terms per domain (the web footprint measure in Finding 4): https://www.semrush.com/