SEO

Do Patients Find Dentists Through ChatGPT and Google's AI? What We Measured

We put five San Diego patient questions to ChatGPT and Perplexity on two days in October 2026. What the answers named, which sources they cited, what changed between days, and what we did not measure.

10 min read
A desktop computer on a white desk showing a grid of image search results, lit by pink and purple light, with a keyboard, a candle and a small framed photo beside it.

The claim behind the AI SEO pitch

If someone has pitched you AI SEO this year, the pitch probably opened with a claim that patients now choose their dentist by asking ChatGPT. This post, part of our Grow Your Practice Online series, checks that claim two ways: what the published surveys actually say, and what we saw when we put five San Diego patient questions to ChatGPT and Perplexity on two days in October 2026. We report how often the answers named a practice, which sources they cited, what changed between days, what we did not measure, and whether either tool names our own firm.

What the surveys say, and what they do not

The figure ChatGPT itself cited when we asked comes from BrightLocal, a local marketing software company. Its 2026 Local Consumer Review Survey used a panel of 1,002 US adults and reports: "Use of ChatGPT and other generative AI tools for local recommendations has grown rapidly, rising from 6% last year to 45% and becoming the third most popular source of business recommendations." Put plainly, 45% of the adults BrightLocal surveyed now use AI tools for local recommendations, up from 6%. That covers local businesses of every kind, not dentists, and it measures use, not who people end up booking.

We found no published figure specific to dentistry that we could verify. The tools agree. When we asked ChatGPT "do patients use ChatGPT to find a dentist" on October 23, 2026, its answer cited the BrightLocal figure and then said "these figures do not establish the exact percentage of patients who use ChatGPT specifically to find a dentist." Perplexity's answer to the same question said "the sources here don’t establish how common this is specifically for dental searches." So we measured what the answers contain instead.

How we ran our test

We wrote five questions the way a patient might type them, each naming a place in San Diego County, and asked both tools each question once a day. AI answers can change with the person, the place and the day, so the setup matters:

  • The questions, word for word: best dentist in North Park San Diego; emergency dentist open Saturday in Chula Vista; family dentist in Escondido with good reviews; dentist in Vista CA that takes Delta Dental; Invisalign dentist in La Jolla.
  • Tools: ChatGPT with web search, and Perplexity, both logged out, in a new desktop browser profile with no history.
  • Dates: the evenings of October 23 and October 24, 2026, Pacific time. We planned three days of runs; this post reports the two we ran.
  • Location: the browser connected through a Los Angeles datacenter internet address, and neither tool showed a location. San Diego came only from each question, so the answers reflect the place named, not what a patient on a phone in San Diego would see.
  • Valid runs: a run counts only when the page held a real answer. Perplexity often showed a sign-in request or a Cloudflare security check instead. We did not try to get around either; we counted those runs and excluded them. On day 1, four of the five Perplexity questions stayed blocked after three retries each. On day 2, two of the five hit a security check and were not retried.
  • One rerun: on day 2, ChatGPT's first pass returned four empty pages; we asked those four again with a longer wait and counted the full answers.
  • Sources: we recorded each answer's links plus source names printed as text, such as Yelp. Both tools hide some sources behind a count, so source figures are a minimum.

What the answers looked like

Every valid answer named specific practices, two to five per answer, with phone numbers and usually addresses. Across both days the answers named a practice in 14 of 14 valid runs (100%) and cited a practice's own website in 11 of 14. Directories and review sites appeared mostly on the insurance and reviews questions. ChatGPT answered the Delta Dental question from Delta Dental's own provider directory on both days, Perplexity cited Zocdoc for it on day 2, and the Escondido reviews question drew Yelp and review platforms such as Birdeye on both days; Yelp also appeared once on the emergency question. No answer cited a news site, a publisher or a dental association. We do not name the practices the tools mentioned.

Our reading: the tools behaved like a search engine that reads practice websites and listings, then writes a short list. The details they repeated (phone numbers, Saturday hours, emergency arrangements, accepted insurance) are details a practice publishes itself or keeps in a listing.

Valid runs only. ChatGPT and Perplexity, logged out, Los Angeles connection, October 23 and 24, 2026
MeasureChatGPT day 1ChatGPT day 2Perplexity day 1Perplexity day 2All valid runs
Valid runs out of five questions5 of 55 of 5 (4 after a rerun)1 of 53 of 514 of 20
Named at least one practice5 of 55 of 51 of 13 of 314 of 14
Cited a practice's own website4 of 53 of 51 of 13 of 311 of 14
Cited a directory or review platform (link or name)3 of 52 of 50 of 11 of 36 of 14
Cited a news site, publisher or dental association0 of 50 of 50 of 10 of 30 of 14
Showed star ratings and review counts2 of 54 of 50 of 10 of 36 of 14
Gave phone numbers5 of 55 of 51 of 13 of 314 of 14
Named Medi-Edge Marketing0 of 50 of 50 of 10 of 30 of 14

Day 1 against day 2

The core of each list held and the rest rotated. Of the 19 names ChatGPT gave on day 1 across the five questions, 12 came back on day 2 for the same question. By question, the overlap ran from one name of three (the Chula Vista emergency question) to four of five (the La Jolla Invisalign question). Perplexity had only one question answered on both days, and three of its four day 1 names came back. The pattern of sources held: practice websites first, directories mostly on the insurance and reviews questions, no publishers.

What changed most was access. Perplexity answered one of five questions on day 1 and three of five on day 2. Two days is too few to call a trend, but enough to show that one screenshot of an AI answer, yours or a vendor's, is a single draw that may look different tomorrow.

How this compares with vendor reports

Three companies that sell AI visibility work have published findings on the same subject, each with a different method.

5W, a public relations firm that sells this work to dental practices and groups, says in its 2026 Dental AI Visibility Report that answers to patient questions come from "a tight cartel of nine directory and review properties," with practices appearing inside those directories "but rarely as independent voices." The methodology section on its web page names query types and procedure categories but not how many queries were run, when, or from where. Our runs did not match it: a practice's own website was cited in 11 of 14 valid runs and a directory or review platform in 6. Different questions may explain the gap.

Insites, which sells website and visibility audits to agencies, asked ChatGPT and Perplexity about a 10,000-business sample of US local companies: 55.58% of the ChatGPT answers Insites checked fully matched the Business Profile, and the most common error was an incorrect phone number. We did not check accuracy, but every valid answer in our runs gave phone numbers, and six of the 14, all from ChatGPT, showed star ratings and review counts. One ChatGPT answer added: "Ratings and review counts are from available listings and may change."

Citevio, an agency that sells AI visibility work to cosmetic and Invisalign dentists, posted a scan of dental review profiles on Dentaltown in August 2026. On whether reviews change ChatGPT's recommendations, it wrote: "Honestly: we don't know, and this data can't tell you." That is the right tone for this subject, and why we publish our method with our numbers.

What Google says about its own AI answers

ChatGPT and Perplexity are not Google, and we did not test Google's AI Overviews or AI Mode on these patient questions. Google has, however, published guidance for site owners on those two features, last updated December 10, 2025. It says: "There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary." And: "You don't need to create new machine readable files, AI text files, or markup to appear in these features. There's also no special schema.org structured data that you need to add." Among the fundamentals it does list is "Checking that your Merchant Center and Business Profile information is up-to-date." Visits from these features are counted with the rest of your Google search traffic in Search Console: "they're reported on in the Performance report, within the 'Web' search type."

Our advice: be wary of any proposal whose main deliverable is an AI text file or extra markup. Google says neither is needed for its AI features, and we have seen no controlled evidence that they buy a place in ChatGPT's answers.

What a practice controls

Nothing here guarantees that an AI answer will name your office; nobody can promise that. These are the inputs we saw the tools read, and each also helps patients who never use AI. Most of them overlap with the basics in our guide to ranking for dentist near me searches. Our advice:

  • Your website states in plain text the facts patients ask about: each service you offer, the insurance plans you are in network with, your hours including Saturdays, how you handle emergencies, and the neighborhood you are in. The answers in our runs repeated exactly these kinds of facts.
  • Your Google Business Profile matches your website on name, phone number, hours and web address. Google lists keeping it up to date among the fundamentals for its AI features.
  • Your entry in each insurer's provider directory is correct. ChatGPT answered the Delta Dental question from Delta Dental's directory on both days.
  • Your main directory and review profiles are claimed and consistent. Our post on which dental directory listings matter covers which ones to keep.
  • New reviews keep arriving at a steady pace. Our post on how many Google reviews a dental practice needs sets a target from your own competitors.
  • Run your own check a few times a month: ask both tools two or three questions a patient would ask about your area, logged out, and note any wrong phone number, hours or insurance detail. Fix it where the answer got it (your site, your profile or the listing it cited), not in the AI tool.

Candidly: does our own firm show up?

We ran the same kind of test on ourselves. On October 23, 2026, with the same logged-out setup, we put 20 questions to both tools: seven about choosing or paying for an SEO or marketing provider (three of them naming San Diego), ten common practice-owner questions, and three that name Medi-Edge Marketing. ChatGPT answered all 20 and Perplexity 19; one Perplexity question stayed behind a sign-in request on all three attempts. Both tools named us, correctly identified, on all three questions that included our name. On the 17 that did not, ChatGPT named us in none, and Perplexity in none of its 16 valid answers. Both tools also said independent reviews of our firm are thin; ChatGPT put it as "the independent review history appears limited." A separate check in September, ten marketing-buyer questions in Google AI Mode in a signed-in browser on one day, named us in none.

We offer local SEO, so this is not a flattering result to publish. We publish it because it is the test we would ask you to apply to anyone selling AI visibility, us included: ask to see the method, the dates, the number of valid runs and the misses.

What this test cannot tell you

Read our numbers with these limits in mind:

  • It is small: five questions, two tools, two days, one pass a day, 14 valid runs. One answer moves a percentage a lot.
  • It is not local: the answers came through a Los Angeles datacenter connection from a logged-out desktop browser with no history, not from a phone in San Diego. Signed-in users, saved locations and past chats can change what a tool says.
  • We did not test Google's AI Overviews or AI Mode, Gemini, Microsoft Copilot, Claude or Apple's assistants on these patient questions.
  • We did not check whether the phone numbers, hours, ratings or insurance details in the answers were correct.
  • We measured what the answers said, not what patients did: no clicks, calls or bookings, and no estimate of how many patients use these tools.
  • Perplexity's results rest on four valid runs, too few for a rate.

See what patients find when they look you up

Request a free audit and we will send a personalized video review of your website and local search presence. If you are at the San Diego Dental Convention on November 20-21, 2026, come by Booth 9 and we can look at it with you.

Free Consultation

Ready to Grow Your Practice?

Medi-Edge Marketing specializes exclusively in medical practices. Get a free audit of your digital presence: no commitment, no sales pressure.

Get My Free Audit