How AI decides which aesthetic clinic to recommend
A patient asks one question. Before answering, an AI assistant rewrites it, runs several searches, retrieves dozens of sources and compares the clinics it finds.
The recommendation is only the final step. Here is one real example.
Follow the process ↓How Gemini turned one question into five clinics.
A single patient question became three searches and 56 scanned results. Follow the path to see which clinics Gemini recommended, where they ranked, and how repeated visibility across several searches shaped the final answer.
From one prompt, Gemini ran three searches and scanned 56 results to name five clinics. Every recommended clinic appeared in the results through its own website. One appeared across all three searches, two appeared across two searches, and two appeared once. If your clinic is absent from the searches AI runs, it is far less likely to enter the final answer.
Hover any clinic or query to trace the path AI followed.
Check whether AI recommends your clinic →Gemini did not search the patient’s exact wording.
It created three narrower queries based on the treatment, desired result and location, then scanned 56 results across them. The clinic does not control which searches the engine creates — only whether its website and wider presence answer them.
Most clinics disappear before the answer is written.
Being in the search results is not being cited. Being cited is not being recommended. Each step removes most of the clinics that came before it.
Two things have to happen.
The clinic must appear in the searches and sources the engine chooses to use.
The retrieved evidence must be relevant, authoritative and clear enough for the clinic to be named in the answer.
Found. The clinic appears in one or more of the searches the engine runs.
Cited. The engine retrieves or references a page mentioning the clinic. It has entered the research process.
Verified. There is enough consistent evidence — website, profiles, registers, reviews — to understand the clinic.
Recommended. The clinic survives the comparison and is named to the patient.
Citation visibility matters. Recommendation visibility is the outcome.
The solution depends on where the clinic disappears.
A clinic absent from the searches may need stronger treatment, concern or location visibility.
A clinic that is retrieved but not selected may need deeper content, clearer practitioner evidence or stronger independent support.
A clinic with inconsistent information may need its website, Google Business Profile, professional profiles and relevant directories brought into alignment.
This is why AI search optimisation cannot be reduced to adding a file, installing schema markup or repeating a keyword. The correct work depends on the point where the clinic is being lost.
See how Penydarren improves AI visibility →A clinic appearing in a search result or citation was not automatically counted as recommended. It had to be named in the answer itself.
- Patient question
- How a patient asks an AI assistant to find or compare a clinic.
- AI response
- The answer generated by ChatGPT, Gemini, Claude or Perplexity.
- Background search
- A distinct search the engine creates while researching its response.
- Citation instance
- An occasion where a page appeared in the source set for a response.
- Clinic mention
- An occasion where a clinic was named in the answer prose.
- Recommended clinic
- A clinic named in the answer itself as a suitable option.
Run the same patient question through another engine and the searches, sources and final clinics can all change.
See the findings across all four engines →How AI recommendations work
Does ranking first mean the clinic will be recommended?
No. A high position makes a page more likely to be retrieved, but the engine still compares the evidence before naming a clinic. The top-ranking clinic is not automatically the final recommendation.
Does being cited mean being recommended?
No. A clinic or source can be retrieved and cited without the clinic appearing in the final answer. Being cited means it entered the research process; being recommended means it was selected and named to the patient.
Why do the engines produce different answers?
Each engine can rewrite the question differently, run different searches, retrieve different sources and select different evidence, so the same question can return different clinics.
Can recommendations change over time?
Yes. Search results, sources, models and wording all change, so answers vary between testing rounds. Visibility should be measured across several questions and repeated over time.
See where your clinic enters the process.
We run relevant patient questions through ChatGPT, Gemini, Claude and Perplexity, then show whether your clinic was found, cited, verified or recommended.