AI Matches Aesthetic Patients on Suitability, and Most Clinic Websites Give It Nothing to Match On

Patients do not ask who is nearest. They ask who will not ruin their face

The questions people put to an assistant before an injectable or a laser are not "aesthetic clinic near me." They are "best clinic for natural-looking lip filler in [city]?", "acne-scar treatment with minimal downtime near me?", "best clinic for melasma on darker skin?" Each is a matching problem: the patient has a skin type, a downtime tolerance, a budget and an outcome they are afraid of missing, and is asking the assistant to find the provider that fits.

Proximity and a star average tell an engine nothing about whether you treat melasma on deeper skin tones, or whether your injector's work reads as subtle. To match you, an assistant needs readable statements: what you treat, in whom, at what price, with how much downtime, and where you would say no. Most clinic websites state none of it in text: it lives in a before-and-after carousel, a PDF price list, an Instagram highlight, a booking widget. A clinic that cannot be read cannot be matched, which is a different failure from ranking badly, and a more fixable one.

One thing this article deliberately leaves alone: licensure, medical directors and who may legally inject. That argument belongs to AI visibility for med spas.

The assistant is fluent in Botox and Juvéderm. Your clinic is a detail attached to the product

5W, a PR and GEO agency that sells exactly this kind of visibility work, published a Medical Aesthetics AI Visibility Index in June 2026: 65 patient-intent prompts, 5 AI engines, 92 candidate brands. Read the scores with the seller in mind. Among injectables it put Botox at 95, Juvéderm 83, Dysport 78, Restylane 77, Sculptra 69, Daxxify 65, Xeomin 56 and Belotero 15, and it puts AbbVie/Allergan and Galderma together at 80% of drug and device AI citation share (5W).

So when a patient asks about their tear troughs, the fluent half of the answer is a product name; the clinic, if it appears at all, is where that product happens to be administered.

That inverts the usual advice. You are not competing with a manufacturer to own the answer; you are competing to be the provider it attaches to. So write at the level of the modality the engine is already fluent in: a page for tear trough filler, one for Profhilo and skin boosters, one for laser resurfacing, each stating indications, alternatives, downtime and who it is not for.

Six things a patient is matching on, and how many of them a clinic site states in words

An assistant fills in the same variables an injector fills in during a consult, and it can only fill them from text. Open a treatment page with images and scripts disabled: what survives is what the engine sees.

• Skin type and tone. "Melasma on darker skin" carries a safety edge — device settings, pigment risk, sometimes the decision not to laser at all. Answered only if your site names the skin types you treat and the devices you use.
• The concern, named as the patient names it: acne scarring, melasma, tear troughs, jowls, "tired eyes." Not "skin rejuvenation," which matches nothing.
• Downtime, in days, per modality. "Minimal downtime" is a qualifier built into the prompts this article asks you to test, and it is worth checking whether any page on your site answers it in a sentence an engine could lift.
• Price, as a number. A price behind "book a consultation" is one the assistant cannot quote, so it quotes a directory's range or a competitor's page.
• The aesthetic goal. "Natural," "subtle," "not frozen" are the patient's real acceptance criteria, and almost nothing on a clinic site addresses them in words rather than photographs.
• The refusal. Contraindications, and the treatments you decline to sell, are the most matchable text a clinic can publish: ruling yourself out for the wrong patient is how an engine learns to rule you in for the right one.

Laser, microneedling, peel, filler or skin boosters: the assistant arbitrates, and it will do it without you

"Profhilo vs skin boosters — who offers this safely?" and "is laser or microneedling better for acne scars?" are modality-selection questions, and the assistant does not decline them: it produces a comparison and a shortlist. How it arrives at either is not disclosed, and no published study measures how assistants shortlist aesthetic clinics or what they weigh when they do — treat the mechanism described here as something to test on your own market, not as a measured finding. What is observable is that the comparison an assistant repeats has to come from text somebody wrote — a manufacturer, a magazine, a directory, and only sometimes a clinic.

Patients take that answer to the reviews. The American Society for Dermatologic Surgery's 2025 consumer survey found 94% of consumers use rating and review websites to choose a cosmetic dermatologic provider, and 70% are considering a cosmetic procedure (ASDS). ASDS is the field's industry society and its page states no sample size, so treat both figures as directional. The assistant proposes; the review page disposes.

So publish the comparison you already give verbally, including the cases where neither treatment is right. An engine can quote a clinic that writes "microneedling will not lift an ice-pick scar — here is what will." It cannot quote a services grid. For the general fixes in order, see how to improve your AI visibility.

Your proof of results is a JPEG, and in California the law has rules about it

Evidence is the signal aesthetic patients want most and the hardest to publish. A before-and-after is the proof, and to an engine it is an unreadable binary — plus a regulated one.

California Business and Professions Code §651 prohibits false or misleading claims and images in the advertising of professional services, unsubstantiated scientific claims, misleading testimonials, and altered or otherwise misleading before-and-after images (California Legislative Information). That is California; other states' medical boards write their own advertising rules and they vary. Nothing here is legal advice — have counsel review your before-and-afters, testimonials and any superiority claim before they go up.

The resolution also makes you readable: describe the evidence in text beside the image — how many cases, which protocol and settings, the interval between the photos, and a plain statement that the same result may not occur for every patient. That caption is legible to an engine and close to the substantiation an advertising rule expects anyway.

Ranked by how much each source tells an engine about who a treatment is for

No one has published a directory-level citation-share table for aesthetics. 5W scores RealSelf 86 and calls it "the gatekeeper" (5W) — a vendor's characterization on a marketing page, not a measured share. The list below runs on healthcare-wide data: an untested hypothesis for aesthetic clinics, a starting bet rather than a finding.

The order is not by citation volume. It is by matchability: how much of the suitability question — which patient, which modality, what evidence that it worked — each source can actually hand an engine. A source with a big citation share that says nothing about who a treatment suits will not get you onto a shortlist built out of skin type and downtime. A directory can corroborate a suitability fact or, when it goes stale, contradict it; it cannot state one your own pages never made.

• 1. Your own site, for Gemini — 52.15% of its citations come from brands' own sites, against ChatGPT's near-mirror 48.73% from third-party listings (Yext, October 2025, 6.8 million citations across 1.6 million AI answers; Yext sells listings management, so a finding that listings decide everything sells listings management). First because it is the only source in this list where the suitability variables can exist at all. Action: on each treatment page, write the indication, the skin types it suits, the downtime in days, the price and the case you would turn away — the five inputs to a match, in crawlable HTML.
• 2. The manufacturers, whose vocabulary the answer is already speaking — AbbVie/Allergan and Galderma hold 80% of drug and device citation share (5W; a PR and GEO agency selling into this category). Second because modality is half of the match, and the engine's modality fluency is theirs, not yours. Action: name the product and device you actually use — Profhilo, a named filler line, a named laser platform — so the engine has something to attach a clinic to when it reaches the recommendation half of the answer.
• 3. RealSelf and niche aesthetic directories — outcome evidence, and the only place a subjective query has to land. Perplexity takes 24% of citations on subjective queries from niche industry directories (Yext; Yext sells listings management), and "which clinic does the most natural filler" is exactly that kind of query; 5W scores RealSelf 86 with no citation share published (5W). Action: make the profile carry outcomes and concerns, not a service menu — cases treated, per concern, with the intervals; log the whole line as a hypothesis.
• 4. Reddit, for ChatGPT — it cites 15.4 sources per answer against Gemini's 3.3 (Semrush AI Visibility Index: 126 million prompts, 1,200+ brands, 22 verticals, January to April 2026; Semrush sells AI-visibility tracking) (PPC Land). Fourth because "who did your lips, and does it look natural" is a suitability judgment made by patients, at length, in text — the closest thing to outcome evidence an engine can read, and the one source here you cannot write. Action: read your city's threads to learn the words patients use for the result they want, then answer those words on your own pages.
• 5. General healthcare directories (Healthgrades, Vitals, WebMD) — 52.6% of healthcare citations in the same Yext analysis (Yext; listings-management vendor). The biggest measured share on this list and, for matching, the thinnest: these profiles carry a specialty and a location, and a specialty is not a suitability statement. Action: file the concerns you treat — acne scarring, melasma, pigmentation on deeper skin tones — into the profile fields that accept them, and expect it to move less than the share suggests.
• 6. Instagram, Facebook and Yelp, for Google AI Mode — 11.4 sources per answer; AI Overviews use 9.2, YouTube the largest single source (Semrush; Semrush sells AI-visibility tracking). Last because it is where a clinic's evidence is most nearly invisible: the result lives in the photograph, and the engine reads only the caption. Action: caption the case — concern, modality, number of sessions, interval, and the plain line that the same result may not occur for every patient.
• One caution: the overlap between the brands an answer mentions and the sources it cites runs only 30% to 64% (Semrush). Chasing citations alone can leave a clinic a footnote in an answer that recommends someone else.

Six questions a patient types, and how to run them yourself in fifteen minutes

Put the questions below to ChatGPT, Gemini and Perplexity, in a fresh chat each time, and write down every clinic named, in order, plus every product and directory the answer leans on. That is the whole method, and it is free.

Then check what the answer says about you, not only whether it says it. SOCi's 2026 Local Visibility Index (roughly 350,000 locations; SOCi sells local-marketing software) put business-data accuracy at about 68% for ChatGPT and Perplexity against 100% for Gemini, which is grounded in Google Maps, and found ChatGPT recommends a given location in 1.2% of answers and Gemini in 11%, against 35.9% for Google's local 3-pack (SOCi).

Patients rarely catch an error: in rater8's June 2026 survey of about 1,000 US adults — rater8 sells review-collection software to medical practices — 60% accept an AI summary without verifying it, while 66% report hitting inaccurate provider information in AI answers (rater8).

The manual pass is a snapshot — answers move across runs, cities and model updates — and it touches no patient data: public questions, public answers, no chart.

• "Best clinic for natural-looking lip filler in [city]?"
• "Where do I get subtle anti-wrinkle injections in [city] — I don't want to look frozen."
• "Acne-scar treatment with minimal downtime near me — laser or microneedling?"
• "Best clinic for melasma on darker skin in [city]?"
• "Profhilo vs skin boosters — who offers this safely in [city]?"
• "How much should tear trough filler cost in [city], and who is safe to do it?"
• "Is [your clinic name] good for natural-looking filler?" That one shows what the engines already believe.

What aesthetic clinic owners ask before changing any of this

Five objections come up in nearly every conversation, including the ones that cut against us.

• "If I publish prices, won't I start a price war?" Possibly, and that trade-off is yours. What is not in doubt: a price an assistant cannot read is one it fills in from elsewhere — a directory range, a competitor's page, or thin air.
• "Isn't this just SEO with a new name?" Not quite: local-search strength does not transfer cleanly. SOCi's 2026 index — SOCi sells local-marketing software and therefore has an interest in the gap it reports — found that in retail only 45% of the top-20 brands in traditional local search reached the top of AI recommendations (Search Engine Land) — the gap covered in AI visibility vs local SEO.
• "Has anyone measured this for aesthetic clinics?" No. No published study measures how often assistants recommend aesthetic clinics, or how many aesthetic patients ask one at all. The aesthetics-specific numbers that do exist describe how patients choose a provider — ASDS on review sites, 5W on which product brands the engines cite — not how the engines behave toward clinics; every AI figure here is healthcare-wide or broader. Ask anyone quoting you an aesthetics-specific AI number for their methodology.
• "My patients come from Instagram, not ChatGPT." Both can be true. BrightLocal's Local Consumer Review Survey 2026 (n=1,002 US consumers; BrightLocal sells local SEO and reputation tools) found 45% of consumers now ask AI to recommend a local business, up from 6% a year earlier (BrightLocal) — a channel-level number, not a statement about your patients, which is the argument for measuring your own.
• "Will this guarantee I get recommended?" No, and be wary of anyone who says otherwise. There is no position to buy inside a ChatGPT answer, and the answer moves with the phrasing, the clinics around you and the next model update. What you can move is the probability that the next answer names you.

Start from a measurement, not a guess

RecoSignal's free AI Visibility Snapshot puts patient-style questions to ChatGPT, Gemini and Perplexity, then reports how often your clinic is named, your share of recommendations against nearby clinics, and the biggest gap behind it. For the mechanics of one engine, see how to get recommended by ChatGPT.