How ChatGPT and AI Overviews Actually Decide Which Doctor to Recommend

A prospective patient no longer just Googles "best private knee surgeon London." Increasingly, they open ChatGPT, or scroll straight past the blue links to Google's AI Overview, and ask: "Who is the best private knee surgeon in London?

The machine gives them one answer. Maybe two or three names. If your name isn't on it, you don't lose a ranking position. You disappear from the conversation entirely.

Most consultants assume this works like Google SEO always has - get your website right, and you're visible. Unfortunately, it doesn't. AI recommendation engines decide who to name using a different mechanism altogether, and understanding it is now essential to your marketing strategy.

1. ChatGPT Doesn't Always "Look You Up" — But Increasingly, It Does‍ ‍

By default, ChatGPT answers from its training data, not a live web search. But OpenAI's own documentation confirms that ChatGPT will automatically trigger a web search when it determines current, real-world information would improve its answer - and a search like "who is the best private knee surgeon in London" is exactly the kind of local, time-sensitive question that tends to trip this. Perplexity is built to search live by default for effectively every query.

When that search does fire, the model is pulling from indexed web pages, review platforms, and directories at that moment - not treating your own homepage as gospel. You can write "London's leading hip specialist" on your site all day; the model is checking whether anyone else is saying it too.

2. Google AI Overviews Are Leaning Less on Your Ranking, Not More‍ ‍

Google explicitly confirms its AI features use "query fan-out" - breaking your patient's question into multiple related sub-queries and pulling citations from across that wider cluster, not just the page that ranks for the exact phrase they typed.

The practical effect: Ahrefs' most recent research, based on over 4 million AI Overview citations, found that only 38% of cited pages also rank in Google's top 10 organic results for that query - down sharply from 76% just a year earlier. Ranking well still absolutely helps, but it's no longer close to a guarantee, and the trend is moving further away from "just rank your own site," not towards it.

Different platforms, same underlying principle: the model is looking for evidence about you that exists independently of your own website.

3. Third-Party Validation Is the New Currency‍ ‍

This is the part that should sound familiar if you've read why Google needs PR to rank your clinic — because it's the same principle, applied to a newer technology. Google's own guidance for AI features confirms it still relies on the same E-E-A-T signals (Experience, Expertise, Authoritativeness, and Trustworthiness) that determine standard search rankings.‍ ‍

AI platforms are explicitly trained to favour content with clear expertise signals: verifiable author credentials, citations from authoritative outlets, and consistent third-party corroboration of who the expert actually is. A journalist at a national title naming you as "leading consultant [X]" and linking to your clinic is precisely the kind of signal these models are built to weight heavily. Your own marketing claims are not.

In other words: the AI is doing, algorithmically, exactly what a sceptical patient would do by instinct - asking "does anyone independent of this person actually vouch for them?"

4. Reviews and Directories Now Matter More, Not Less

Because ChatGPT and Perplexity pull directly from review platforms and directories at the moment of the query, your presence (and your standing) on these sites is no longer a "nice to have" reputational asset - it's live input data for the recommendation engine itself. A thin or outdated directory listing isn't just a missed opportunity; it's a gap in the evidence the model uses to decide whether to name you at all.

5. What This Means for Your Marketing Strategy‍ ‍

Chasing keywords on your own website is no longer the whole game. To be the name an AI model surfaces, you need a consistent footprint of independent validation across the web: national and trade press coverage, up-to-date directory and review profiles, and clear, consistent association between your name and your specialty wherever you're mentioned.

This is precisely what strategic PR builds — deliberately, and well before your competitors even realise the goalposts have moved.

A note on the detail here: neither Google nor OpenAI publishes the exact mechanics of how sources get selected, and Google itself warns that AI Overviews are "rapidly evolving" and can make mistakes. What's above reflects the best available official guidance and independent research at the time of writing — and the direction of travel, if anything, is moving further towards external validation, not away from it.

Don't wait to find out you've been left off the list.

At RL Comms, I secure the kind of Tier-1 media coverage and digital footprint that gets ambitious consultants named - by journalists, by Google, and now by AI. Get in touch to discuss a bespoke retainer.

Not ready for an ongoing commitment? Book a PR Foundations Power Hour to audit your current visibility across search and AI, and get a strategy in place.