Barely at all. We asked ChatGPT and Perplexity four versions of a local-dentist search, general, cosmetic, Invisalign and dental implant, across the same 527 US dental practice websites in June 2026. ChatGPT named 0 of 372 practices for "best cosmetic dentist" and 1 of 371 for "best Invisalign dentist." Perplexity named at most 2.4% of practices for any of the four versions.
If you are weighing whether to build an Invisalign landing page or rewrite your cosmetic-services copy so that AI notices your practice, this is the data to look at first. We did not test whether adding a page changes anything, that is a different, harder question we address directly below, but we did test whether the wording of the search question itself changes who gets named. The short version: not much. What that means for your practice, and the honest limits of what this test can tell you, follows.
What we tested: four search intents, one set of practices
We ran the same underlying local-dentist search four ways: a general version, plus three that named a specific intent, cosmetic, Invisalign, and dental implant. Every version ran against the same practice base scanned between 6 and 26 June 2026, on both ChatGPT and Perplexity, so the wording of the question is the only thing changing between rows of the table below.
Read each row against its own valid-reading count, not against the full 527. The next section explains why that count moves from row to row.
| Engine | Query we asked | Valid reading | Named | Share |
|---|---|---|---|---|
| ChatGPT | "best dentist in {city}" | 373 | 1 | 0.3% |
| ChatGPT | "best cosmetic dentist in {city}" | 372 | 0 | 0.0% |
| ChatGPT | "best Invisalign dentist in {city}" | 371 | 1 | 0.3% |
| ChatGPT | "best dental implant dentist in {city}" | 372 | 1 | 0.3% |
| Perplexity | "best dentist in {city}" | 245 | 6 | 2.4% |
| Perplexity | "best cosmetic dentist in {city}" | 244 | 4 | 1.6% |
| Perplexity | "best Invisalign dentist in {city}" | 245 | 4 | 1.6% |
| Perplexity | "best dental implant dentist in {city}" | 245 | 4 | 1.6% |
Of the 527 practices in our published base, 506 have at least one query-level reading; the other 21 do not. We do not know whether that group would score better or worse, and we are not willing to guess. A further 20 of the 506 are excluded from the table above for a different reason (a malformed city label, explained below), leaving 486 usable for the comparison. Across all eight rows, ChatGPT's named share never rises above 0.3% and Perplexity's stays inside a 0.8-point band, 1.6% to 2.4%. Whichever way you ask the question, most of the practices in our sample are not the ones getting named.
Why does the valid-reading count change from row to row?
Only slightly, and that is what makes the comparison meaningful. The four ChatGPT denominators sit within two practices of each other, 371 to 373, and the four Perplexity denominators within one, 244 to 245. None of the eight rows lost a count to a measurement error; the small gaps come from individual practices missing a reading for one query but not another. In practice, all four rows per engine are reading almost exactly the same set of practices.
That is a meaningful correction from an earlier version of this dataset, where one batch of scans had a corrupted city label ("Columbus Oh Cosmetic Dental Clinics" instead of "Columbus") that leaked the word "cosmetic" into queries that were not asking about cosmetic dentistry. We caught it, excluded the 20 affected practices from the table rather than guess at the right bucket for them, and the remaining denominators are now close enough that comparing rows side by side, general versus cosmetic versus Invisalign versus implant, is a fair comparison rather than one distorted by different-sized groups. The practical rule still holds: read each percentage against the reading count printed in its own row, and don't divide the "named" column by 527.
Were these clinics labeled "cosmetic," or was the question?
The question. None of the 527 practices in our sample, and no subset of them, were tagged as cosmetic, Invisalign, or general dentistry. What changed between rows of the table is the wording we sent to ChatGPT and Perplexity, not the population of practices being asked about.
That distinction changes how the cosmetic row should be read. The correct sentence is: when we asked ChatGPT for the best cosmetic dentist in these seven metros, it named none of the 372 practices we had a reading for. The incorrect sentence, which we are not making, is that some percentage of cosmetic dental practices are invisible to AI. We have no denominator for that claim, because we never sorted practices by specialty in the first place. The cosmetic row (372) and the Invisalign row (371) are, within a practice or two, the same set of dental practices as the general row (373), just asked about differently.
What this data does not show
We did not test whether adding Invisalign content, renaming a service page, or writing more about cosmetic procedures changes any of these numbers. We changed the wording of the question, not anything about the practices themselves, so this is a read of how two AI engines answer different search intents today, not a before-and-after test of any fix a practice could make.
The practices behind the Invisalign and cosmetic rows overlap heavily with the wider set we cover in why isn't my dental practice showing up in ChatGPT, where crawler access, structured data and Bing indexing turned out to matter more than reviews. If your practice wasn't named in any of the four rows above, on-page content naming a specific service is unlikely to be the one lever that changes that on its own, though we have not measured that directly either. For the technical and structural signals we have measured, see AI SEO for cosmetic dentists.
What can your practice check today?
The checker below reads your own site's technical AI-readiness in seconds: crawler access, structured data, and Bing footprint among them. It does not run our four-query test against your practice specifically, that would take the kind of scan we ran city by city, but it will tell you whether the fixable, structural gaps behind this dataset apply to your site.
Prefer to run the queries yourself? See how to check what ChatGPT says about your practice.
How we measured this
The base is 527 US dental practice websites scanned between 6 and 26 June 2026 across Charlotte, Austin, Raleigh, Nashville, Columbus, Tampa and Salt Lake City, plus 21 practices in 9 smaller nearby towns; our wider testing framework is in how we measure AI visibility. For each practice with a query-level reading, we recorded whether ChatGPT and Perplexity named it in response to four separately worded local-dentist questions. One batch of 20 Columbus-area practices was scanned with a corrupted city field ("Columbus Oh Cosmetic Dental Clinics" instead of "Columbus"), which made the query text itself malformed for that batch, so we excluded those 20 from the table above rather than count a broken question as a reading. One possible reason ChatGPT's share barely moves with intent: for local questions, ChatGPT runs a Bing search and works from roughly the same 20 to 30 results regardless of how the question is phrased, according to Search Engine Land's 2025 breakdown of ChatGPT's local search mechanism. If the underlying candidate pool changes little between "best dentist" and "best cosmetic dentist," the names ChatGPT can choose from wouldn't change much either. We have not confirmed that this is the actual mechanism behind our numbers, only that it's consistent with them.
Source: Damian Rollison, Search Engine Land, May 2025.
For scale, SOCi's 2026 Local Visibility Index measured how often local businesses across roughly 350,000 locations and 2,751 brands, mostly outside dentistry, get recommended by each assistant: about 1.2% for ChatGPT and 7.4% for Perplexity. Our dentist-specific figures sit inside that same order of magnitude, ChatGPT from 0.0% to 0.3%, Perplexity from 1.6% to 2.4%, though the two studies measure different industries and shouldn't be read as a like-for-like gap.
Source: SOCi 2026 Local Visibility Index, via Search Engine Land, January 2026.
- A snapshot, not a trend. This is one June 2026 scan window. Profound tested about 80,000 prompts per platform and found the domains cited in AI answers change 40 to 60% month to month for identical questions, so a practice invisible here in June may surface later, and the reverse.
- Two engines, not three. Gemini was not included in this scan.
- Sample, not census. Seven metros plus nearby towns, not a national sample.
- Correlation is not a test. Nothing here compares a practice before and after a content change; it compares four differently worded questions against the same set of practices at one point in time.
Source: Profound, AI search volatility, 2025. Full dataset: citevio.com/data. A city-level look at the same base is in the AI visibility report for Charlotte dentists, where the one ChatGPT-visible Invisalign result in our table above was scanned.
Common questions
Will adding an Invisalign page get my practice recommended by AI?
We don't know, and this page doesn't test that. What we measured is how ChatGPT and Perplexity answer today when a patient asks a cosmetic or Invisalign-specific question versus a general one, across the same 527 practices, not what happens if any one of those practices changes its content. In our June 2026 read, ChatGPT named 1 of 371 practices for the Invisalign version of the query and 0 of 372 for the cosmetic version, so most practices in our sample were not appearing either way. Adding a page is a reasonable step, but we have no data showing it changes whether AI names you.
Does asking AI a cosmetic-specific question change whether it recommends a dentist?
Barely, in our data. ChatGPT stayed in a narrow band, 0.0% to 0.3%, across all four intents we tested. Perplexity moved a little more, from 1.6% up to 2.4%, but the general query was the one that performed best, not any of the three specific-intent queries: cosmetic, Invisalign and implant all landed at exactly 1.6%. Neither engine showed a large jump when the question got more specific.
Which engine is more likely to name a dentist for these queries?
Perplexity, consistently. Across the four intents we tested, Perplexity named between 1.6% and 2.4% of the practices with a valid reading; ChatGPT stayed between 0.0% and 0.3%. That gap is consistent with the wider pattern in SOCi's 2026 Local Visibility Index, which put ChatGPT's general local-recommendation rate at about 1.2% against Perplexity's 7.4% across roughly 350,000 locations in many industries, not dentistry specifically.
Were the practices in this study cosmetic dentistry specialists?
No. The 527 practices are dental practices generally, family, general, orthodontic and cosmetic clinics scanned across seven US metros. We did not tag any practice as a cosmetic specialist; the word "cosmetic" only appears in the question we asked the AI. See how we measure AI visibility for the full method.