How Medspas Show Up in ChatGPT: The 6-Step GEO Checklist I Run for Aesthetic Clinics

OpenAI said in October 2025 that ChatGPT had reached 800 million weekly users. Some of those people are your future patients, and they have stopped typing “botox near me” into Google. They open ChatGPT, ask “which medspa in Scottsdale is best for first-time Botox”, and book with whichever clinic the answer names. I have spent 9 years doing search work for service businesses, and this is the biggest shift in patient discovery I have watched happen in real time. Almost no medspa is prepared for it.

This post is the generative engine optimization (GEO) checklist I run for aesthetic clinics. Six steps, the reasoning behind each one, and a monthly test so you know whether any of it is working. Nothing here requires a developer on staff, and most of it costs nothing but time. If you want to see where AI visibility fits inside a full patient-acquisition system, my medspa marketing hub lays out the whole picture.

Patients changed where they ask. Most medspas have not noticed.

The classic funnel still works: a patient googles “lip filler cost”, clicks a few results, fills out a form. But a second funnel has grown beside it, and it behaves differently. The patient describes her whole situation to an AI assistant in plain language. “I’m 34, I’ve never had injectables, I’m terrified of looking frozen, who should I see in Denver and what should it cost?” The assistant answers with two or three named clinics and a reason for each.

Three properties make this second funnel impossible to ignore:

• It compresses choice. Google shows ten organic results plus ads plus a map pack. ChatGPT typically names two to four businesses. Either you are in the answer or you do not exist in that conversation. There is no page two.
• It carries borrowed trust. A recommendation from an AI assistant reads like advice from a well-read friend, not like an ad. Patients who arrive this way come in warmer and ask fewer price-shopping questions.
• It hides in your analytics. Visits from AI answers usually show up as direct traffic or as referrals from chatgpt.com and perplexity.ai. Most owners never connect those sessions to AI recommendations, so the channel grows unmeasured.

I am not telling you to abandon Google. Search still drives most discovery, and the fundamentals overlap heavily. What I am saying is that the work below makes you visible in both places, and the clinics that do it early will compound an advantage while competitors keep debating whether AI search “counts”.

How ChatGPT actually decides which medspas to name

Understanding the mechanics tells you exactly where to spend effort. AI assistants answer local questions in two ways.

1. From training data

The model has read a snapshot of the public web. If your clinic appears in directories, press coverage, listicles and review platforms that existed before the training cutoff, the model may “remember” you. You influence this slowly, by existing in lots of credible places over a long period.

2. From live retrieval

This is the mode that matters for local intent. When someone asks ChatGPT with search enabled (or Perplexity, or Gemini) about medspas in a city, the assistant fires off web searches behind the scenes, reads the top results, and composes an answer from what it just read. Those sources are usually a mix of map listings, review sites like Yelp and RealSelf, “best medspa in [city]” listicles, and individual clinic websites.

The implication is blunt: to be named in the answer, you need to be present and legible in the sources the assistant reads. That is the entire game. The academic paper that coined the term GEO tested this directly and reported visibility improvements of up to 40% in AI answers for pages that added statistics, quotations and source citations. Machines reward pages that hand them clean, attributable facts.

Everything below is a way of doing exactly that.

Step 1: Make your site machine-readable with schema markup

Schema markup is structured data added to your pages that labels your business facts in a vocabulary machines understand. Your homepage might look beautiful to a human, but to a crawler it is a soup of divs. Schema turns “we’re a med spa in Austin open till 7” into fields a model can extract without guessing.

For a medspa I implement, at minimum:

• MedicalBusiness or LocalBusiness on the homepage, with your name, full street address, phone, geo coordinates, opening hours, price range, and sameAs links pointing to your Google Business Profile, Instagram, Facebook and review profiles.
• FAQPage on treatment pages, marking up the real questions patients ask, with answers in the 40 to 80 word range. These blocks get quoted nearly verbatim in AI answers more often than anything else I publish.
• Service or Offer markup for individual treatments, ideally with honest price ranges. Models strongly prefer recommending businesses whose pricing they can actually state.

You do not need to hand-write code. Rank Math or Yoast on WordPress will generate most of this from settings, and a developer can fill the gaps in an afternoon. One caution while you are in the site code: medspas carry a specific legal exposure around tracking scripts on booking pages, and I wrote a separate HIPAA website checklist for medspas that you should run during the same audit.

Not sure what your site currently exposes to crawlers? I will run the audit with you on a screen share, no charge and no pitch. Book a free 30-minute call and bring your homepage URL.

Step 2: Publish an llms.txt file

llms.txt is a plain-text file that sits at yourdomain.com/llms.txt and gives AI systems a curated, human-written summary of your site: who you are and which pages matter. It was proposed in September 2024 as a convention, the way robots.txt once was. Adoption among AI crawlers is still uneven, and I will not pretend it is a magic switch. But it takes about 20 minutes, costs nothing, and its entire function is to put an accurate description of your clinic directly in front of any system that looks for one.

A medspa version looks like this:

# Glow Aesthetics
> Single-location medical spa in Austin, TX. Botox, dermal fillers,
> laser hair removal, HydraFacial. Founded 2019. Medical director:
> Jane Smith, MD.

## Services
- [Botox and Dysport](https://example.com/botox/): pricing, FAQs, before/after gallery
- [Dermal fillers](https://example.com/fillers/): lip, cheek and jaw filler details
- [Laser hair removal](https://example.com/laser-hair-removal/): packages and candidacy

## About
- [Meet the team](https://example.com/team/): injector credentials and licensure
- [Pricing](https://example.com/pricing/): transparent treatment pricing
- [Reviews](https://example.com/reviews/): patient testimonials

Write it honestly, link your strongest pages, and update it when services change. While you are in there, open robots.txt and make sure you are not blocking GPTBot, ClaudeBot or PerplexityBot. I still find medspa sites that block AI crawlers because a security plugin shipped that way by default, which quietly removes the clinic from every AI answer it could have appeared in.

Step 3: Build pages AI wants to quote

Retrieval-based assistants are quote machines. When one reads your page, it is hunting for sentences it can lift into an answer with confidence. Most medspa websites give it nothing: vague adjectives, stock photos, “contact us for pricing”, and paragraphs that never commit to a number.

Pages that get cited share a recognizable structure:

• Question-shaped headings. “How much does Botox cost in Austin?” as an H2, with the direct answer in the first sentence underneath it. Not three paragraphs of warm-up first.
• Real numbers. Price ranges, unit counts, session counts, downtime in days. If you estimate, label the estimate. A model prefers a page that says “est. $10 to $15 per unit, and most first-time forehead treatments need 20 to 30 units” over one that says “affordable luxury”.
• Tables. Treatment comparison tables and pricing tables get extracted constantly. They are dense and unambiguous.
• Named sources. When you attribute a claim to a journal or a manufacturer, you make the whole page safer to quote.

The same logic explains why benchmark content earns citations. My post on medspa Google Ads cost benchmarks is built as tables of clearly labeled estimated ranges precisely because that format is what AI answers reach for when someone asks what medspa advertising costs.

For your clinic, the priority list is short: one transparent pricing page, one page per flagship treatment with a question-shaped FAQ section, and one local page that states plainly which neighborhoods and suburbs you serve. That is the corpus an assistant needs before it can recommend you for “best medspa for [treatment] in [city]”.

Step 4: Make your NAP boring and identical everywhere

NAP means name, address, phone. AI models do entity resolution: they try to work out whether “Glow Aesthetics”, “Glow Aesthetics Med Spa” and “Glow Aesthetic Center LLC” are the same business. Every inconsistency lowers the model’s confidence, and a model with low confidence quietly names a different clinic instead of yours.

Audit these in one sitting:

• Google Business Profile (the single most-read source for local AI answers)
• Yelp, RealSelf and Healthgrades listings
• Instagram and Facebook bios
• Your own site footer and contact page
• Any chamber-of-commerce or local directory pages

Same exact business name, same address format, same phone number, same hours, everywhere. If you rebranded or moved in the last few years, hunt down the stale listings and fix or remove them. This is unglamorous work and it moves the needle more than most owners believe, because consistency is one of the few signals a machine can verify entirely on its own.

If you would rather hand this whole checklist to someone who has run it before, that is literally what I do for clinics. Grab a free 30-minute call and I will tell you which steps your site already passes and which ones are costing you mentions.

Step 5: Grow a review corpus that models can read

Reviews are your third-party evidence layer. When an assistant decides whether to recommend you, the text of your reviews tells it what you are actually good at, and review platforms rank well in exactly the searches assistants run behind the scenes.

What matters for AI visibility specifically:

• Specificity beats volume. A review that says “best laser hair removal experience I have had in Austin, Maria explained every setting” connects your clinic to a treatment and a city in retrievable text. Coach happy patients, gently and compliantly, to mention the treatment they had.
• Spread across platforms. Google reviews matter most, but RealSelf and Yelp show up in AI source lists for aesthetics queries constantly. Do not let everything pool in one profile.
• Respond to every review. Owner responses add more crawlable text, prove the entity is active, and let you naturally restate your services and location.
• Never fake or buy reviews. Beyond the legal exposure, models read patterns across thousands of reviews, and synthetic clusters look like synthetic clusters.

Third-party corroboration is also why I keep my own track record on a platform I do not control: 222 completed jobs on Upwork, a 97% job success score, Top Rated Plus status and 37 five-star reviews. I mention it partly as credibility and partly because it is the lesson itself. Evidence that lives on an independent platform is exactly the kind machines weight, for my agency and for your clinic alike.

Step 6: Publish free tools. They are citation magnets nobody in your market uses.

Here is the gap that surprised me most when I built my competitor research. As of June 2026, per their own sites, neither Thrive (160+ staff) nor Studio 3 Marketing (roughly 190 to 200+ staff), two of the most visible agencies selling medspa marketing, offers a single free, no-signup tool. Everything sits behind a contact form. The same pattern holds across most medspa websites themselves: galleries, brochure pages, booking forms, nothing interactive.

That is an opening, because AI assistants love recommending tools. Ask ChatGPT how many units of Botox a first-timer needs and it will happily point to a calculator if a good one exists to point to. Ungated interactive pages earn links and bookmarks, and they surface in answers as “you can estimate this with…”. A medspa could build:

• A Botox unit and cost estimator by treatment area
• A treatment downtime planner (“my event is June 20, what can I safely book before it?”)
• A laser hair removal session and package calculator
• A “which facial fits my skin” quiz with honest logic behind it

I practice this myself. The free, no-signup tools I publish at my tools page exist because ungated utility is the most durable citation bait I know of. If you are evaluating agencies and want to see who actually operates this way versus who only talks about it, I broke the field down in my Thrive alternatives comparison for medspas.

How to test your AI visibility every month

GEO without measurement turns into superstition. In the last week of every month, run the same prompt set and log the answers:

1. “Best medspa in [your city]”
2. “Where should I get Botox in [city]? I am a first-timer and nervous”
3. “Best place for laser hair removal near [neighborhood]”
4. “Is [your clinic name] reputable? What do reviews say?”
5. “[Your clinic] vs [main competitor], which should I choose?”

Run them in ChatGPT with web search on, in Perplexity, in Gemini and in Copilot. For each one, record which clinics get named and, crucially, which sources the assistant cites. Those citations are your to-do list. If a “12 best medspas in [city]” listicle keeps appearing, get into it. If RealSelf profiles keep getting cited, strengthen yours. Reverse-engineering the sources beats guessing every single time.

Also check GA4 for referrals containing chatgpt.com and perplexity.ai, and add an “AI assistant (ChatGPT etc.)” option to the “how did you hear about us?” question on your intake form. Imperfect, but it makes the channel visible. And when you interview agencies, ask them to run this exact test on their own brand, live, in front of you. In the research for my ranked list of medspa marketing agencies, very few could.

Five mistakes that keep medspas out of AI answers

I see the same handful of own-goals on almost every medspa site I audit. Fix these before you build anything new:

1. Pricing hidden behind consultation forms. A model cannot recommend you for value if it cannot find a single number on your site. Publish ranges, label estimates as estimates, and let the page do the pre-qualifying for you.
2. Treatment menus as PDF downloads. AI retrieval reads HTML pages. A beautiful PDF price menu is close to invisible. Rebuild it as a normal page with a table.
3. Content that only renders in JavaScript. Some site builders inject your actual text after page load. Many crawlers never see it. View your page source and check that your headings and pricing exist in the raw HTML.
4. Crawler-blocking security defaults. Firewall plugins and CDN bot rules frequently block GPTBot and PerplexityBot out of the box. Whitelist them deliberately.
5. Inconsistent business identity. A renamed clinic with five stale directory listings reads, to a machine, like five low-confidence businesses instead of one strong one.

None of these takes more than a day to fix, and every one of them sits upstream of all six steps above. Cheap wins first.

What this costs, and the honest version of each route

If you are still working out what total marketing spend makes sense before picking a route, I wrote a stage-by-stage framework in how much a medspa should spend on marketing, and my complete rate card sits on the pricing page. The honest summary: the basics in this post are owner-doable in a weekend or two. Where clinics stall is the content layer, the citable pages and the tools, because those take sustained production over months. That is the part worth paying for, if you pay for anything at all.

The window is open right now

Every channel I have worked in over 9 years had an early period where effort was cheap and attention was underpriced. Google Ads in the late 2000s. Local SEO when the map pack launched. AI answers are in that period today. Most medspas have no schema, no llms.txt, no citable pricing, and no tools, which means a single-location clinic that does this work properly can beat brands with 50x the budget inside an AI answer.

That window will close the way they all eventually do. If you want help getting through it while it is open, book a free 30-minute call and I will map this checklist onto your site live, whether or not we ever work together afterward.

FAQ

Still unsure where to start? Book a free 30-minute call and we will find the highest-impact step for your clinic inside the first ten minutes.