Why doesn't ChatGPT recommend my med spa when people ask about treatments in my city?

Injectables, laser treatments, and body contouring sit in a gray zone between cosmetic and medical, and AI models are trained to be cautious anywhere health outcomes are involved. When a query touches injections, energy-based devices, or anything that penetrates skin, the model tends to hedge, cite established medical-adjacent sources first, and only name a specific provider when it has strong independent confirmation that the provider is legitimate.

That confirmation rarely comes from your own site, because your own site is inherently self-interested. It comes from platforms built specifically to verify aesthetic providers at scale, the same way a physician directory outranks a solo practitioner's homepage for a symptom-based medical query.

This filter also explains why a med spa can rank well in Google's traditional organic results while still being invisible in ChatGPT or Perplexity answers. Organic ranking rewards on-page relevance and backlinks; AI citation rewards independently verifiable trust, and those are not the same scorecard.

RealSelf has spent over a decade accumulating patient-written reviews, doctor Q&A threads, and a "Worth It" rating system tied to specific treatments in specific cities. When someone asks an AI model whether a treatment is worth it in their area, RealSelf is frequently the single most-cited source behind the answer, ahead of any individual practice website.

A med spa without a claimed, active RealSelf profile is effectively opting out of the citation pool for a huge share of treatment-comparison queries. This is not optional infrastructure for aesthetics the way it might be for other verticals; it is closer to a prerequisite.

Providers who answer patient questions on RealSelf regularly, not just once after signing up, build a visible track record of expertise that the model can associate with a named doctor or injector, which carries more weight than a bio page nobody outside your own site links to.

Allergan (Botox, Juvederm), Galderma (Restylane, Sculptra), Merz (Radiesse, Xeomin), and Ulthera each maintain a "find a provider" locator tied to their brand. These locators are structured data, tied to a recognizable brand name, and crawled regularly, which makes them exactly the kind of source an AI model treats as verification.

Most med spas either never registered for these locators or let the listing go stale after opening. Getting current, complete listings on every manufacturer locator relevant to your service menu is one of the fastest, lowest-cost fixes available in this vertical.

A page titled "Injectables" that lists Botox, filler, and Kybella in three sentences each cannot compete with a dedicated page on, say, jawline filler for men that covers dosage ranges, typical cost per syringe in your market, healing timeline, and who is not a good candidate. Depth at the individual-treatment level is what gives the model something specific to retrieve when a query names a treatment rather than a category.

The pattern we see across audits is consistent: practices with 15 to 20 treatment-specific pages outperform practices with five broad category pages, even when the broad pages have more total words on them.

This also matches how people actually phrase questions to ChatGPT. Nobody types "tell me about your injectables menu." They ask "how much does under-eye filler cost in Naples" or "how long does CoolSculpting downtime last," and a page structured around that exact phrasing has a direct match a category page simply does not.

Before-and-after galleries are strong trust signals for human visitors but often carry no structured data the model can parse, and AI systems are cautious about drawing conclusions from image content in a medical-adjacent category regardless. Pair every gallery entry with a written case description (treatment used, number of sessions, timeline) so the value exists in text, not only in the image.

Overclaiming language ("permanent," "guaranteed results," "cures") tends to push a model toward hedging your entire page rather than citing it, because the language pattern matches content the model has been trained to treat skeptically. Precise, bounded claims read as more credible to the model, not less.

A caption that reads "Results after 2 sessions of Ultherapy, effects typically last 12 to 18 months" gives the model a concrete, quotable fact. A caption that just says "amazing transformation" gives it nothing to work with beyond the image itself.

The fastest path to being named is treating third-party platforms as primary infrastructure rather than an afterthought.