Where AI Gets Its Healthcare Facts in 2026

When people use AI to search for healthcare information, who gets cited, and why? And what can you do to increase your chances of being visible in the first place?

// By Andy Ibbotson //

If you’ve been in a healthcare marketing conference session this year, you’ve probably seen SEMrush’s infographic, “Where AI Gets Its Facts.” It lists the top domains cited by large language models (LLMs) like ChatGPT and Perplexity.

Reddit tops the list at 40.1 percent, followed by Wikipedia at 26.3 percent, YouTube at 23.5 percent, Google at 23.3 percent, and so on.

This chart has become shorthand for how AI models source their answers, and many health system marketing and communications teams quote it as if it applies to them. It doesn’t.

It tells you where AI engines get facts about the internet in general, but it says almost nothing about where AI engines get facts when a patient asks, “What’s the best hospital for knee replacement near me?” or “Which hospital has the best cardiology program?”

Healthcare has its own citation hierarchy. Reddit and Wikipedia, the two most-cited domains on the general web, drop well down the list in our healthcare-specific industry benchmark, with Reddit sitting at 6.4 percent citation share and Wikipedia at just 2.3 percent — a fraction of their share on the general web of 40.1 percent and 26.3 percent respectively.

This data raises a bigger question than which sources AI cites most: Is your system named in the answer at all?

Read on to learn how citation share changes when looking only at healthcare domains, and why citation position matters as much as citation share.

Log in below to access this content: