We Asked Four AI Platforms to Recommend a Dentist in San Francisco. They Named 76 Different Ones.

Finding a good dentist is already stressful enough without AI making it worse.

Everyone who’s moved to a new city knows the feeling. You need a dentist, you don’t know anyone, and the decision feels weirdly high-stakes for something you’ll think about twice a year. This is someone who’s going to put sharp instruments inside your mouth. You want to get it right.

So you ask ChatGPT. “Best dentist in San Francisco.” And you get an answer. Three or four names, confidently presented, sometimes with a quick note about each one. Feels helpful. Feels trustworthy.

Except we tracked dozens of these queries across four different platforms, and the results were unlike anything we’ve seen in any other category.

Seventy-Six Dentists and Nobody Won

When we tracked the same types of queries about private schools in Los Angeles, the top two platforms agreed on the best schools. Harvard-Westlake and Brentwood came out on top of both ChatGPT and Gemini. There was a clear consensus at the top, with disagreement only in the middle.

With dentists in San Francisco, there was no consensus at all.

Across ChatGPT, Gemini, Perplexity, and Grok, a total of 76 different dental practices were recommended. The most-recommended practice in the entire study scored 48 out of 100 on our combined visibility metric. For context, Harvard-Westlake scored in the high 90s in the school study. No dentist came close to that kind of dominance.

In every other category we’ve tested, at least one or two businesses owned the top spot across platforms. With dentists, nobody owns anything.

The Ghost Practices

The strangest part wasn’t the fragmentation. It was the exclusivity.

All Smiles Dental was recommended by Grok in every single query we tracked. Every one. On Grok, it’s the best dentist in San Francisco, full stop.

On ChatGPT, Gemini, and Perplexity, All Smiles Dental doesn’t exist. Not once. Not in any query. Zero.

Lands End Dental had an identical pattern on a different platform. Perfect score on Perplexity. Recommended every time. On the other three platforms, never mentioned.

Downtown Dentist SF showed up in nearly every Gemini response. On everything else, nothing.

Studio 32 Dentistry showed up in nearly every Perplexity response. Elsewhere, invisible.

We kept finding these. Practices that are the top recommendation on one platform and a complete ghost on the others. Not just underperforming. Actually absent. As if they were dentists in different cities.

A patient who opens Perplexity will almost certainly see Lands End Dental. A patient who opens ChatGPT will never hear the name. Same neighborhood, same week, same question.

AI Can’t Decide What to Call Your Dentist

In other categories, we’ve seen AI confuse businesses with similarly-named competitors. Kitchen showrooms got mixed up with the cabinet brands they sell. Schools with common names got confused with institutions in other states.

With dentists, the confusion is more direct. AI can’t decide whether to recommend the practice or the person.

One San Francisco dentist appeared in the data three separate times, under three different names. Once as “Robert Soto, DDS.” Once as “Dr. Robert Soto.” Once as “Robert Soto DDS.” Same person, same practice, but AI treated each version as a different recommendation. The combined visibility of that practice was split across three entries instead of one.

The same thing happened with at least five other practices. “Itani Dental” in one response, “Dr. Samer Itani (Itani Dental)” in another. “Skyline Dental Studio” here, “Dr. Marjan Fathi (Skyline Dental…)” there.

For the patient reading the response, this probably isn’t confusing. They’ll figure it out. But it tells us something about how shaky the foundation is. If AI can’t even settle on what to call the practice, how confident can it really be in the recommendation?

Why Dentists Are Different

We’ve now tracked AI recommendations across restaurants, kitchen showrooms, private schools, and dental practices. The dentist data was the most chaotic by a wide margin.

Part of that is the nature of the business. A city might have a handful of truly famous restaurants and a few elite private schools. But there are hundreds of dental practices in San Francisco, and most of them serve their neighborhoods rather than competing citywide. There’s no “Harvard-Westlake of dentistry” that every source writes about.

Which means AI is guessing more than usual. When it can’t find a clear signal pointing to two or three obvious winners, it fills the response with whatever it can find. And since each platform searches different sources, each platform finds different practices.

But there’s another reason this matters more for dentists than for restaurants. Nobody gets anxious about choosing the wrong pizza place. Choosing the wrong dentist is a different kind of risk. A root canal costs $1,000. An implant can run $3,000 to $5,000. Veneers, $15,000 or more. And beyond the money, there’s the trust that comes with letting someone perform a medical procedure on your body.

The higher the trust required in a decision, the more weight a confident AI recommendation carries. And with dentists, that confidence is built on the thinnest data we’ve ever seen.

What This Looks Like From the Other Side of the Chair

For a dentist reading this, the numbers tell an uncomfortable story.

A new patient is worth somewhere around $3,000 to $5,000 in the first year, between cleanings, X-rays, and whatever treatment they need. A patient who stays for a decade represents $15,000 to $30,000 or more. These aren’t abstract numbers. This is what walks through the door (or doesn’t) based on a recommendation that took an AI half a second to generate.

Most practices track their new patient sources carefully. Google Ads, referrals, insurance directories. But almost none are tracking whether ChatGPT mentions them when someone in their zip code asks for a recommendation. And if the data here is any indication, the answer for most practices is no. Of 76 practices mentioned, the vast majority appeared only once across all queries. A single mention on a single platform is a lucky accident, not visibility.

The few practices that showed up on more than one platform shared something in common. They tended to have presence across multiple review sites, mentions in local editorial content, and enough information online that AI could confidently describe what makes them different. A practice with strong reviews on both Google and Yelp, mentions in a “best dentists in SF” article, and a website that lists specialties, insurance accepted, and languages spoken gives AI something to work with. A practice with only a Google Maps pin often doesn’t make the cut.

Finding Out

The five-minute manual check we’ve written about works for dentists the same way it works for any business. Open ChatGPT, ask for a dentist recommendation in the relevant area, and see what comes back.

But as this data makes clear, checking one platform isn’t enough. A practice could be the top recommendation on Perplexity and completely invisible on ChatGPT. The only way to know the full picture is to check all of them. And then check again next month, because the answers change.

Reachd.ai tracks this across platforms automatically, so a practice can see exactly where it stands, where competitors show up, and what’s driving or blocking recommendations. For a business where each new patient is worth thousands of dollars, whether ChatGPT recommends the practice belongs in the marketing budget as a real line item, not a curiosity.

Because somewhere in San Francisco right now, someone with a toothache is asking ChatGPT for help. And the answer they get depends entirely on which app they happened to open.