Contents
Reputation management for medical aesthetics is no longer about collecting five star reviews. Patients now ask ChatGPT, Gemini, Claude and Perplexity which clinic to trust before they ever open Google. What matters is whether the engines can find your reviews, read them, verify they belong to you, and repeat them back when a patient asks who is safe.
What a patient actually does before they book
They ask a question in plain language. Not “lip filler London” but “which clinic in London should I go to for lip filler if I have had a bad result somewhere else and I am nervous about it happening again”.
The engine answers with two or three named clinics and a sentence of reasoning for each. Sometimes it names a practitioner rather than a clinic. Sometimes it refuses to recommend anyone and lists what to look for instead, which is its own kind of answer, and one that quietly removes every clinic from consideration at once.
That answer is assembled from whatever the engine can find and trust about you: your reviews, your website, your Google Business Profile, directory entries, register listings, forum threads, press mentions, and any place a real patient has written about their experience. If your proof is scattered, thin, or sitting somewhere an engine cannot read, you are simply not in the answer. There is no second page to be on. The patient never learns that your clinic existed.
This is the part that catches clinic owners out. You can be the best injector in your city, with a waiting list and a wall of genuine five star reviews, and still be absent from every AI recommendation in your postcode. Being good and being legible to a machine are two different problems, and only one of them is currently being solved by most clinics.
Your review count is not failing you. It has stopped being the measurement.
For fifteen years the reputation scoreboard in aesthetics was simple. Number of reviews, average star rating, and where you sat in the local map pack. Clinics got very good at moving those three numbers, and the numbers genuinely predicted bookings.
They predict less now, because a growing share of patients never see them. When an engine answers a recommendation question it does not show a ranked list of clinics with star ratings to compare. It gives a short answer naming a few options. Your 400 reviews do not appear. What appears is whether the engine decided to say your name.
So a clinic with 400 reviews that no engine cites can lose to a clinic with 90 that every engine names. That is not a claim about which clinic is better. It is a statement about what the engines can currently read, verify and repeat.
The scoreboard that matters now is a different one: for a given set of patient questions, on a given date, on named engines, does your clinic get mentioned, what does the engine say about you, and where did it get that from. That is measurable. It is just not measured by the tools most clinics are using.
Your reviews are not the problem. Where they live is.
Most clinics we look at have decent reviews. What they do not have is those reviews in a form an engine can actually use. Four things break it, and they break it quietly:
Reviews trapped in platforms
Sitting only inside a booking system or on a platform that blocks AI crawlers in its robots file, so the engines never see them at all. The reviews are real. They are just invisible to the thing making the recommendation.
No structured data
Reviews on the site as plain text with no Review or AggregateRating markup, and no Organization or Person markup tying the practitioner to the clinic. The engine has to guess, and guessing favours whoever did not make it guess.
Inconsistent clinic identity
A different trading name, address format or phone number across Google, directories, registers and your own site. The engine cannot confirm these are the same clinic, so it discounts the lot rather than risk conflating two businesses.
Nothing answering the hard questions
Safety, qualifications, prescriber status, what happens if something goes wrong, how corrections are handled. The questions patients genuinely ask, unanswered anywhere you control, so the engine answers them from somewhere you do not.
None of these are marketing problems in the traditional sense. They are legibility problems. The work is unglamorous and largely invisible to patients, which is exactly why so few clinics have done it.
The four places AI looks for proof about your clinic
| Where | What it decides | What we do |
|---|---|---|
| Your own site | Whether the engine can verify who you are, who treats, and what you treat | Review and AggregateRating schema, practitioner credentials and prescriber status marked up, treatment pages rewritten to answer questions rather than list services, FAQ blocks that match real patient phrasing |
| Google Business Profile | Still the most cited source when an engine is asked for a local clinic recommendation | Identity consistency, category accuracy, review velocity and response quality, the Q&A section actively managed rather than left to strangers, services and attributes completed properly |
| Third party platforms and registers | Whether independent corroboration of your clinic exists at all | Presence and consistency across the review sites, directories and practitioner registers the engines actually pull from, including the ones specific to aesthetics rather than generic business directories |
| Open discussion | What the engines repeat when a patient asks about risk, complications or bad results | Monitor the forums and communities where your clinic and your treatments get discussed, and track what the engines say back when asked directly |
Every one of these is measured, not assumed. We run a fixed set of patient questions across the named engines on a set schedule, save the answers verbatim with the date attached, and show you what changed between runs. Where something cannot be confirmed, it is recorded as not verified rather than counted as absent. Those are different states, and treating them as the same is how reporting starts lying to you.
What gets tracked, and how often
Reputation work without measurement is decoration. The part that makes this defensible is that every claim we make about your visibility has a date, an engine name and a saved answer behind it.
- Named engines, not “AI”. ChatGPT, Claude, Gemini, Perplexity, Google AI Mode and Google AI Overviews are checked separately, because they disagree with each other constantly and an average across them hides the thing you need to see.
- A fixed question set. The same patient questions each run, so movement means something. Ad hoc spot checks feel productive and prove nothing, because answers vary between sessions and between users.
- Verbatim answers, saved and dated. Not a score with no working shown. You can read exactly what the engine said about your clinic on the day, including the runs where it said nothing.
- Sources captured where given. When an engine cites what it drew on, that gets recorded, because it tells you which of the four places above is actually carrying your reputation.
- The gaps reported as loudly as the wins. The questions where you did not appear are the working list. A report that only shows the good runs is a brochure.
One honest caveat, stated up front rather than buried: AI answers fluctuate between runs and between users. A single check is one data point from one session. That is precisely why the measurement is repeated on a schedule instead of screenshotted once and put in a deck.
Reputation management for medical aesthetics is regulated, and AI writing tools do not know that
Medical aesthetics advertising in the UK sits under the ASA and CAP codes, and parts of it sit under medicines law. Botulinum toxin is a prescription only medicine, and prescription only medicines cannot be advertised to the public. That single rule catches an enormous amount of routine clinic marketing, including a great deal of what sits in Instagram captions and on treatment pages right now.
Beyond that, before and after imagery has its own conditions, testimonials that make medical claims carry risk, and language implying a guaranteed or risk free outcome is a problem regardless of how carefully it is hedged elsewhere on the page.
Generic AI copy tools know none of this. Asked to write a page about a toxin brand for a clinic, they will produce something confident, fluent and non compliant, and it will read well enough that nobody catches it. We have seen clinics publish exactly that in good faith.
So anything written for a clinic here is written against those rules first, not checked against them afterwards. Where something already live looks exposed, it gets flagged to you plainly rather than quietly left in place because raising it is awkward.
Not legal advice. We work to the published ASA and CAP guidance and to the medicines advertising rules, and we flag risk where we see it. Final sign off on regulated claims stays with your clinic and, where relevant, your prescriber. If a claim needs a regulatory opinion, we will say so rather than guess.
What good actually looks like for a clinic
It is worth being concrete about the finish line, because “improve your AI visibility” is not a target anyone can hold you to.
You get named
For the recommendation questions that matter in your area and your treatments, your clinic appears by name in the answer, on more than one engine, on more than one run.
The reasoning is yours
When the engine explains why it named you, the reason reflects what you actually do well: the practitioner qualification, the complication handling, the specific treatment focus. Not a generic line it invented.
The hard questions land safely
Asked about risk, corrections or bad results near your clinic name, the engine draws on material you wrote rather than a forum thread from four years ago.
It holds
The position survives repeat runs over months rather than appearing once and vanishing. Durability is the whole point, and it is only visible if you are measuring on a schedule.
What the first ninety days look like
- Weeks one to two, the baseline. The full question set run across the named engines and saved. Identity audited across your site, Google Business Profile, directories and registers. You get the honest starting position including every question where you were absent.
- Weeks three to eight, the legibility work. Schema on the site, identity inconsistencies resolved, Google Business Profile completed and its Q&A taken in hand, review flow set up so new reviews land where engines can read them, and the unanswered patient questions answered on pages you control.
- Weeks nine to twelve, the second read. The same question set run again on the same engines. Compared against the baseline verbatim by verbatim. What moved, what did not, and what the next quarter targets.
Identity and schema corrections tend to register first because they remove a blocker rather than build an asset. Shifting what an engine says in an open recommendation question is slower, and depends heavily on how much independent proof about your clinic already exists. Any timeline promised more precisely than that is guesswork.
Reputation is one pillar of five
Reputation and trust does not work in isolation. An engine that trusts your reviews but cannot crawl your site, or can crawl your site but finds no independent corroboration anywhere, still will not recommend you with any confidence.
So reputation is measured alongside AI visibility, SEO and technical health, automation, and PR and earned media. One score, five pillars beneath it. You can buy the reputation work on its own, and plenty of clinics do, but the baseline reads all five because a reputation problem is very often a technical problem wearing a disguise.
A worked example of that. A clinic with excellent reviews that no engine cites usually turns out to have a crawler or schema problem, not a reputation problem. If we only sold you reputation work you would pay for the wrong fix and see nothing move.
Self serve, or we run it for you
- Start with The Baseline™.It stands on its own, no plan needed, and our plans carry no setup fee if you carry on.
- You see where you actually stand. Named engines, dated answers, saved verbatim, including the questions where you did not appear at all.
- Then you choose. Run the fixes yourself in the dashboard with the priority list in front of you, or have us run them. Either way the measurement continues, so you can see whether the work moved anything rather than taking it on trust.
What clinic owners ask first
Find out what AI is telling patients about your clinic
Start with The Baseline™
Can you remove a bad review?
No, and anyone promising that is either breaking platform rules or lying to you. What we do is make the honest picture legible: get genuine reviews into a form the engines can read, make sure responses are handled properly, and make sure the questions behind a bad review are answered somewhere you control. A well handled complaint often reads better to both patients and engines than an unbroken run of five stars.
How is this different from normal reputation management?
Traditional reputation management optimises what a person sees when they search your clinic name. This optimises what an engine says when a patient asks a question that does not contain your name at all. Those are different problems, and the second is where new patients increasingly come from.
How long before anything changes?
Identity and schema fixes can register within weeks because they remove an obstacle. Shifting what an engine says in an open recommendation question takes longer and depends on how much independent proof exists to begin with. We report at a fixed cadence either way, including the runs where nothing moved.
Will you write our review responses?
On the managed plan, yes, drafted for your approval and written against ASA and CAP rules. On self serve you get the templates and the schedule and your team sends them.
We are a single practitioner. Is this overkill?
Often the opposite. A single practitioner clinic usually has less independent proof circulating than a group, so the engines have less to work with, and a modest amount of structured work moves things further than it would for a large chain.
What if we appear nowhere at all?
That is a common starting point and it gets recorded honestly rather than smoothed over. The Baseline™ shows you the exact questions where you did not appear, and that list becomes the plan.
Do you guarantee we will be recommended by ChatGPT?
No. Nobody can, and a guarantee would tell you something unflattering about whoever offered it. Answers vary between runs and between users. What we commit to is the measurement discipline: same questions, named engines, dated verbatim answers, gaps included, so you can see what is actually happening rather than being told.
Is any of this against platform rules?
No. There is no review manipulation, no incentivised reviews, no fake accounts and no attempt to game a platform. The work is making genuine proof findable and verifiable, which is what the platforms and the engines both say they want.
We already have an SEO agency. Does this replace them?
Not necessarily. There is overlap on schema and technical work, and it is worth deciding who owns what so the same job is not billed twice. Plenty of clinics keep their SEO agency and add this alongside. We would rather tell you where the overlap sits than pretend there is none.
Can you work with our booking system and existing review flow?
Usually. The aim is to get reviews landing where engines can read them without disrupting a flow your front of house already follows. If your current system traps reviews somewhere invisible, we will say so and suggest the smallest change that fixes it.
What do you need from us to start?
Access to your Google Business Profile, your website, and whatever review platforms you already use. Then a short conversation about which treatments and which area actually matter commercially, because that determines the question set the whole measurement is built on.
What happens if the results are bad?
You get told. The reporting shows the runs where nothing moved and the questions where you are still absent. If a plan is not working we would rather change it than keep reporting around it.
AI Visibility Expert and Founder of The GEO Agency, helping travel, hospitality and property companies win visibility in AI search. 28 years of digital marketing, brand strategy and enterprise digital transformation experience with brands including Virgin Media, O2, Samsung, LEGO, RBS/NatWest and ITN. Author of the ChatGPT Blueprint, Anthropic Claude Partner Network member and Claude Certified Architect.
