From invisible to cited: winning AI answers for a London clinic
Patients were arriving having 'asked ChatGPT'. The clinic was not in any of those answers. Two competitors were, repeatedly.
The problem
What was going wrong
Traditional rankings were solid but referral patterns were shifting. Across a fixed set of 60 patient-intent prompts run against AI Overviews, ChatGPT search and Perplexity, the clinic was cited in none. Its own content was written as marketing copy — long, hedged, and impossible to quote.
The work
What we did about it
- Established a prompt-set baseline across four answer engines and recorded which sources were being cited instead.
- Rewrote 40 clinical information pages into self-contained, quotable passages with defined terms, dated evidence and named clinical authors.
- Consolidated the clinic's entity footprint — identical descriptions, services and credentials across the site, directories, professional bodies and profiles.
- Added author schema tied to real clinicians with verifiable credentials, plus FAQ and MedicalWebPage markup where genuinely applicable.
- Set AI crawler policy deliberately and published a llms.txt summary of the clinic's services.
We assumed this was a future problem. It was already affecting bookings when we started.
Services used
What this programme involved
Figures are taken from the client's own analytics and CRM over the stated period and are shared with permission. Client name withheld under NDA.
Portfolio
More work
Design, build and growth programmes we can talk about publicly — with the numbers the client signed off.
Next step
Your situation is probably not identical
But the method is. Send us the numbers and we will tell you which part of this applies to you.