Penydarren helps UK aesthetic and medical aesthetics clinics improve how they appear across ChatGPT, Gemini, Claude, Perplexity and Google AI search. We begin by identifying where the clinic is being lost.
It may be absent from the searches an engine runs. Its website may be retrieved without providing enough evidence. Its practitioners may be difficult to connect to the treatments they offer. Independent sources may be incomplete, inconsistent or pointing towards competitors.
We strengthen the part of the recommendation process connected to that gap, then repeat the testing to measure what changed.
AI assistants can use a mixture of search results, clinic websites, Google Business Profiles, professional registers, clinic directories, practitioner pages, reviews and editorial sources when answering a patient. The exact mixture changes according to the engine, treatment, location and wording of the question.
That means effective AI search optimisation cannot be reduced to adding an llms.txt file, installing schema markup or repeating a collection of keywords.
We examine the real searches and sources behind the recommendations, then improve the evidence available at the point where your clinic is being lost.
The clinic must appear in the searches and sources the AI engine chooses to use. This is where search visibility, local SEO, page relevance and technical accessibility matter.
A clinic that never enters the research process has very little chance of reaching the final answer.
Once found, the clinic must provide enough relevant and credible evidence to survive the comparison. The engine may need to understand:
Ranking can open the first lock. It does not automatically open the second.
AI engines do not always search the patient’s exact wording.
They may rewrite the question into several narrower searches involving the treatment, concern, location, practitioner type, reviews or desired result.
We identify the searches being created around your services and assess whether the right pages from your clinic are appearing. The work may include:
The objective is not simply to rank for a broad treatment name. It is to make the clinic available across the wider set of searches an AI engine may use before producing its answer.
A clinic can appear in search results and still be excluded from the recommendation.
This often happens when the available page is too generic, too brief or disconnected from the patient’s actual question.
We strengthen the pages that explain:
This may involve treatment pages, concern pages, practitioner profiles, comparison content, location pages or clearer connections between existing information.
The aim is not to publish more words. It is to provide distinctive and usable information that can be found, extracted and applied to an answer.
Medical aesthetics recommendations depend on more than the clinic name.
An engine may need to establish who provides a treatment, what they are qualified to perform and whether their experience is supported elsewhere.
We assess how clearly the clinic communicates:
We also compare those claims with relevant registers, practitioner profiles and other independent sources. The information should be clear, accurate and consistent wherever it appears.
Not every frequently cited source improves visibility.
AI assistants often retrieve professional registers, clinic directories, review platforms, practitioner profiles and editorial pages when researching a recommendation.
However, citation volume alone does not tell us whether a source is valuable. Our research found that some heavily cited sources produced little or no measurable lift because they repeatedly recommended only a small selection of their own registered clinics. Being listed on the source did not mean a clinic was more likely to reach the final answer.
We therefore assess more than how often a source appears. We look at whether it:
This is not about placing your clinic on every available directory. It is about identifying the independent sources that genuinely influence recommendations, then improving your presence where the evidence shows it matters.
Technical changes cannot guarantee a recommendation, but technical problems can prevent useful information from being discovered or understood.
Depending on the clinic, we may assess:
Any technical recommendation must support real clinic information. Schema markup does not replace substantive content, and an llms.txt file does not replace search visibility, practitioner evidence or independent support.
AI recommendations are variable.
The searches, retrieved sources and final clinic selections can change between engines and between testing rounds.
We measure visibility across a structured set of treatment, concern, practitioner and location questions. The testing separates four stages:
The clinic appeared in a search created by the engine.
The engine retrieved the clinic website or another source mentioning it.
The available sources provided enough consistent information to understand the clinic, practitioner, treatment and location.
The clinic was named in the final answer.
This allows us to measure more than a simple recommendation count. It shows whether the clinic is entering more searches, gaining stronger source visibility and moving closer to the final selection.
We run patient questions relevant to your treatments, practitioners and locations through ChatGPT, Gemini, Claude and Perplexity.
We record the searches created, the pages retrieved, the sources used and the clinics named.
We establish whether the clinic is absent from the research process, retrieved without being selected or undermined by incomplete evidence.
We identify the pages, profiles, listings, technical issues or evidence gaps most closely connected to the result.
Penydarren improves the agreed website content, local search presence, practitioner information, independent profiles and technical structure. The exact scope depends on the clinic and the findings.
We repeat the testing to measure whether the clinic is entering more searches, appearing in stronger sources and reaching more recommendations.
AI assistants often rely on the same search infrastructure that helps people find clinics through Google. If your treatment pages are not visible, your local signals are weak or your website does not clearly explain what you offer, the clinic is less likely to enter the searches behind an AI recommendation.
Penydarren takes traditional SEO further. We improve the foundations that help your clinic appear in search, then examine what happens after it is found:
This means the work can include technical SEO, local SEO, treatment and concern pages, internal linking, practitioner content and independent sources. The difference is that every improvement is also assessed against how AI assistants research, verify and recommend clinics.
SEO helps your clinic enter the process. AI visibility work helps it survive the comparison and reach the recommendation.
ChatGPT, Gemini, Claude and Perplexity do not publish a complete formula for selecting clinics, and their answers change over time. Penydarren cannot guarantee that a particular clinic will appear for a particular question.
What we can do is identify where the clinic is currently being lost, improve the information and evidence available to the engines and measure whether its visibility changes. That is more useful than promising a ranking nobody controls.
Penydarren works with:
The work is suitable for established clinics with genuine clinical expertise that is not yet being represented clearly enough across AI search.
AI search optimisation improves the information and evidence available when assistants such as ChatGPT, Gemini, Claude and Perplexity research and recommend businesses. For an aesthetic clinic, this can involve search visibility, local SEO, treatment content, practitioner evidence, structured data and relevant third-party sources.
No, although they overlap. SEO helps a clinic appear in search results. AI search optimisation also examines which results an engine retrieves, which sources it trusts and whether the clinic reaches the final recommendation. A clinic can rank highly without being recommended.
Generative Engine Optimisation, often shortened to GEO, is commonly used to describe work intended to improve visibility within AI-generated answers. Penydarren uses the clearer term AI visibility because the objective is not simply to optimise a page. It is to understand and improve the wider recommendation process.
We assess Google Business Profile where it is relevant to the clinic’s local visibility or appears within the searches and sources connected to its recommendations. Any work must be based on the actual gap rather than assuming the profile is always the deciding factor.
Yes. We implement appropriate schema markup as part of every engagement. This helps search engines interpret important information about the clinic, its practitioners, treatments, locations and website content more consistently. Schema markup is a supporting technical measure, but it does not guarantee inclusion in an AI-generated answer.
In our study, after accounting for other variables including domain rating and Google Business Profile, we found no evidence that clinics with an llms.txt file achieved greater visibility in AI recommendations. That does not prove the file will never have value. AI search is still developing, and the systems used to discover and interpret websites continue to change. We repeat our testing monthly to identify whether signals such as llms.txt begin to show a measurable relationship with visibility. Where appropriate, we may still create and maintain an accurate llms.txt file. However, it is not treated as a replacement for search visibility, substantive clinic content, practitioner evidence, technical accessibility or credible independent sources.
No. No consultancy controls the searches, source selection or final answer produced by an AI engine. Penydarren identifies the gaps that can be improved and measures whether visibility changes through repeated testing.
We will show you how your clinic appears across ChatGPT, Gemini, Claude and Perplexity, which competitors are being selected instead and what should be addressed first.