Generative engine optimization · Medical aesthetics
Ask AI for the best med spa and a directory answers 84% of the time. Ask about acne scars and it almost never does.
Profound, one of the bigger AI visibility trackers, ran the numbers on citation sources across brands and came back with a stat that gets repeated a lot: about 4% of AI citations point to a brand's own website, with the other 96% coming from somewhere else, mostly YouTube, Reddit, and review sites. If you've read anything about AI search this year, you've probably seen that number used to argue that owning your site doesn't matter anymore, only your presence on other platforms does.
We went back into our own med spa testing to see where that number lands in our category, and it depends entirely on what the person typed. A comparison-style question gets answered by a directory almost every time. A problem-style question sends the model straight to a dermatology practice's own page, again and again.
We ran the same five cities through the same setup for two different kinds of questions: "best med spa in [city]" and "I have acne scars, where in [city] can I get them treated." Same engine, same week, same location list. The only thing that changed was the shape of the question.
84% vs 10%
For "best med spa," 84% of the sources ChatGPT cited were directories: Yelp-style aggregators, review compilers, "top 10" list sites built to rank many businesses against each other. For the acne scars question, that number dropped to 10%. Out of 48 citations, 5 were directories, and the other 43 were mostly individual dermatology and med spa practice pages, plus one hospital system and a single citation from the American Academy of Dermatology's own patient guide.
Look at the actual pages behind that 10%. Biscayne Dermatology's acne scarring page in Miami. Colorado Dermatology Center's acne scar treatment page in Denver. A Chicago clinic's scar revision page. None of these sites carry directory-scale authority or a national ad budget. They're small local practices whose page happened to answer the exact question someone asked, in the city where they asked it.
It holds outside med spa too
We checked whether this pattern was specific to med spa, and it isn't. We ran the same query structure across nine other service categories, and every single one showed the same direction: symptom-shaped questions cite far fewer directories than rating-shaped ones. Personal injury lawyers went from 0% directory citations on a symptom-style question to 83% on "best lawyer near me." Dentists went from 2% to 73%. Plumbers, chiropractors, and vets all showed the same pattern at different strengths. Optometrists were the outlier, sitting at 46% even on the symptom-style question, likely because "why is my vision blurry" pulls in more general health sources than a plumbing or dental symptom would.
Why the split happens
The mechanism isn't complicated once you see both ends of it. A rating-shaped question asks the model to compare options it hasn't personally evaluated, so it reaches for a source that has already done the comparing: a directory, a ranked list, a review aggregator. A symptom-shaped question asks for a specific answer to a specific problem, and the page that answers it best is usually the practice's own treatment page, written for exactly that patient and that procedure. The model isn't choosing loyalty to small businesses over directories. It's matching the shape of the answer to the shape of the question.
What this means if you run a med spa
That split matters for where you put your effort. If you've been told the whole game is getting listed everywhere and collecting reviews, that advice is correct for one slice of your traffic and close to useless for the other. Directory presence helps you show up on "best med spa near me" and similar comparison questions, where 84% of our citations point to exactly that kind of source. It does close to nothing for the patient who already knows their problem and is asking how to fix it. That patient gets answered by whichever practice wrote the clearest page about their specific symptom, in their specific city.
Most med spa sites we've looked at don't have that page. They have a services list with "acne scarring" buried as a line item under a general aesthetics menu, rather than a dedicated page titled around the way a patient actually asks the question. The dermatology practices showing up in our 10% mostly do have that page, and it's usually short: what the treatment is, what it costs or roughly costs, what city it's in, and a few sentences of specificity that a directory listing can't match.
What we can't claim from this
We should be upfront about the limits here: we pulled from one day of testing, September 28, 2026, across five U.S. cities, using up to 48 queries for each intent in each category. The direction held across all ten categories, but the exact percentages will move with a bigger sample or a different week. We're also measuring which sources got cited rather than why the model picked one page over another, beyond the pattern visible in the pages themselves. And this is one AI engine on one kind of setup; we haven't run the same split against every model people search with.
If you run a med spa, the practical version of this is small: pick your three or four most common treatment questions, the ones patients actually type into a search bar, and check whether your site has a real page answering each one by name. Directory listings and reviews still earn their keep on the comparison side of your traffic. The symptom side is being decided by whoever bothered to write the page.
Source: askchat.studio's GEO probe database, 669 ChatGPT citations across "best [category] near me" and symptom-style queries in ten service categories (med spa, dentist, plumber, chiropractor, veterinarian, optometrist, gym, hair salon, auto repair, personal injury lawyer), five U.S. cities, collected September 28, 2026, through the OpenAI Responses API with web search grounding. External stat: Profound AI visibility study, cited via a third-party summary of 34 AI search studies. Correction, September 30, 2026: the med spa percentages were updated from 69%/8% to 84%/10% after we added five niche med spa directories to our classifier that a separate audit had already identified but that hadn't been added to the code yet. The cross-category numbers for the other nine service categories were unaffected.