Short answer: we don't know, and we're not going to fake a confident answer to sound more authoritative than the data lets us be. In our June 2026 scan of 523 U.S. dental practices, the median practice had 250 Google reviews at a 4.9-star average — and 96.7% already sit at or above 4.3 stars, the average rating of the locations ChatGPT actually recommended in a separate industry study. What we have not done, and want to say plainly rather than imply, is test whether review count or rating causes a practice to get named. That's a different, harder question than "what's normal," and this post is deliberately narrow about which one it's answering.
How many reviews does the typical practice actually have?
In our June 2026 scan of 527 U.S. dental practices in 7 metros plus 9 smaller nearby towns, 523 had a rating on file. The median was 250 reviews at 4.9 stars. A quarter of practices had fewer than 102 reviews; a quarter had more than 474.
| Percentile | Google reviews | Star rating |
|---|
| Fewest (min) | 1 | 2.9 |
| 25th percentile | 102 | 4.8 |
| Median | 250 | 4.9 |
| 75th percentile | 474 | 5.0 |
| 90th percentile | 764 | 5.0 |
Two different shapes here. Review count spreads out a lot — a practice at the 90th percentile has more than 7 times the reviews of one at the 25th. Star rating barely spreads at all — three quarters of practices sit at 4.8 or higher. That ceiling matters for the next section: with almost every practice already bunched at the top of the scale, star rating alone doesn't do much to separate one practice from another in this sample.
Where do practices sit against the ratings AI-recommended locations averaged?
A separate industry study (SOCi's 2026 Local Visibility Index, reported by Search Engine Land) found that the locations ChatGPT actually recommended averaged 4.3 stars; Perplexity's averaged 4.1; Gemini's averaged 3.9. Those are averages of what got recommended in that study, not cutoffs any engine is known to enforce — worth repeating, because it's easy to misread as a pass mark.
Against our own 523-practice sample:
| Engine | Avg. rating of locations it recommended (SOCi study) | Our practices at/above that average | Share of 523 |
|---|
| ChatGPT | 4.3? | 506 | 96.7% |
| Perplexity | 4.1? | 516 | 98.7% |
| Gemini | 3.9? | 522 | 99.8% |
Nearly everyone in our sample already sits above all three reference points. If your practice is anywhere near the median above, star rating is very likely not what's holding your AI visibility back — but that's a statement about where the sample sits, not a statement about what causes a citation.
So — do reviews affect ChatGPT's recommendations, or not?
Honestly: we don't know, and this data can't tell you. Testing that would mean comparing the review profiles of practices an engine actually named against the ones it didn't, in large enough groups to mean something. In our own scan, ChatGPT named only 3 practices and Perplexity named 18 — and nearly all of them were already sitting in the same top-rated band as almost everyone else in the sample. Three names and eighteen names are too few to sort into groups and call the result a finding, so we're not going to dress up that noise as one.
If we get a future scan with enough named practices to run that comparison properly, we'll publish it. Until then, we'd treat any claim you see — ours or anyone else's — that reviews "affect" or "improve" AI citations as unproven. We haven't tested it, and we haven't gone looking for whether someone else has published a proper test of it either — so take "unproven" as our honest limit, not as a claim that no such study exists anywhere.
What we can say with more confidence
Reviews and ratings are close to a solved problem for almost every practice in our sample — so if you're wondering what to check next, review volume is probably not it. In a separate scan of the same practices, checked against whether ChatGPT actually named them, the signal that was almost universally missing wasn't reviews — it was structured data (schema markup) on the practice's own site.
Common questions
Do Google reviews affect ChatGPT's dental recommendations?
We don't know — we haven't run that test ourselves, and we're not going to guess at the answer. What we can say is that in our June 2026 scan, 96.7% of 523 practices already sit at or above the 4.3-star average of the locations a separate study found ChatGPT actually recommending.
What star rating do I need for ChatGPT to notice my practice?
There's no confirmed cutoff. 4.3?, 4.1? and 3.9? are the average ratings of the locations ChatGPT, Perplexity and Gemini recommended in a separate 2026 industry study — averages of an outcome, not thresholds the engines are known to enforce. In our scan, 96.7% of practices already clear the highest of those three averages.
Does buying reviews help?
No, and it isn't worth the legal exposure — Google's policies prohibit paying for reviews or filtering out unhappy patients, and it's unnecessary for most practices anyway: 96.7% of the ones we scanned already sit above the highest reference average without it.
Where can I see the full dataset and percentile breakdown?
The