A patient who asks an AI engine about symptoms gets one synthesized answer and a short list of cited links, and for most clinics none of those links are theirs. When LLM Pulse analysed nearly 825,000 citations from AI answers to US health questions, hospitals and providers made up just 1.0% of them, excluding pages owned by the brands the study tracked. A healthcare AEO strategy is the work of earning a place in that list for the questions where you have a real claim to authority.
The good news is that most of it builds on work you may already be doing. The less comfortable news is that the engines don't all cite the same kinds of sources, so one plan won't cover them all.
What a healthcare AEO strategy changes, and what it doesn't
Answer engine optimization is search optimization aimed at the passages AI tools quote, rather than the blue links beneath them. For Google's own AI features, the company says there are no additional requirements and no special optimizations beyond being indexed and eligible to show with a snippet in regular Search.
That statement deserves to be taken at face value. Google also says you don't need new machine-readable files, AI text files or special markup to appear in AI Overviews or AI Mode, and that inclusion isn't guaranteed even when every condition is met.
So the fundamentals carry the weight: crawlable pages, helpful content, a good page experience. Google's statement covers its own products, and we haven't found equivalent published guidance from the other engines, which makes the writing and trust work below the safest investment.
Which sources do AI engines cite for health questions?
It depends heavily on the engine and on the kind of question. BrightEdge tracked citations over 14 weeks, from October 2025 to January 2026, and found ChatGPT leaning on major hospital systems far more than Google's AI features do.
| Share of citations from elite hospital systems | ChatGPT | Google AI Overviews | Google AI Mode |
|---|---|---|---|
| Symptom queries | 57% | 20% | 18% |
| Treatment queries | 52% | 20% | 17% |
Those numbers come from BrightEdge's healthcare AI citations analysis, which found ChatGPT cited government sites and medical specialty organizations at much higher rates too. Consumer health media gained little from either platform.
Our reading, and it is only that, since neither study measured independent clinics: a practice shouldn't expect to displace a hospital system on a generic definition of a condition. The openings are the questions a hospital's national pages can't answer, such as what a visit to your clinic involves, who you treat, what you accept and how to book.
Write the answer first, then the evidence
Open each page, and each section of it, with a direct answer in plain text, then add detail. Google notes that important content needs to be available in textual form, so an answer that lives only inside an image, a carousel or a PDF isn't a candidate.
Phrase headings the way a patient would ask the question, and answer it in the first two or three sentences so the passage still makes sense if it's quoted alone. A heading like "How long does a first physical therapy visit take?" earns more than "Our Approach".
The LLM Pulse data points the same direction. In that study, authoritative sources such as government bodies, academic journals and medical publishers appeared earlier in answers than social content, which suggests the first credible, clear answer an engine finds has an edge.
Show who wrote it and who checked it
Trust is the lever you control most directly. Google's helpful-content guidance says that of the elements it lists, trust is the most important, and that health topics get even more weight: content must be highly accurate and consistent with established expert consensus.
In practice that means a visible byline on every clinical page, linked to a real bio with credentials. Google specifically warns that invented author profiles, false credentials or AI-generated headshots are deceptive and mark a page as low quality.
A reviewer line helps too: the clinician who checked the page and the date they did. We cover the broader case for this in our piece on healthcare digital marketing strategy and why trust is the new SEO.
Does schema markup help healthcare pages?
Photo by Bibek ghosh on Pexels
Not as a lever for Google's AI features. Google says no special schema.org structured data is needed, though any markup you add should match the text a visitor can see on the page.
That doesn't make markup pointless. The schema.org MedicalWebPage type has fields for the intended medical audience and the specialty, and it inherits reviewedBy and lastReviewed from WebPage, which describe exactly the review trail your visible page should already show.
Treat markup as an accurate label for what's on the page. If the byline and review date aren't visible, adding them only in code gets you nothing and breaks Google's matching rule.
Keep your clinic and provider details consistent
Engines answering "urgent care near me" style questions need clean facts about who you are and where you are. Google's guidance asks you to keep Business Profile information current, make content findable through internal links, and verify your site in Search Console to diagnose technical problems.
Check also that robots.txt and your CDN or host aren't blocking crawlers. It's a boring failure, and a very common one on sites that have been through a redesign.
Name, address, phone, hours and the list of providers should read the same on your site and your profiles. For the wider picture of how the site, SEO and ads fit together, see our healthcare digital growth stack, and if you want a team to run the technical and content side, that's what our digital marketing and growth service covers.
How to measure whether AI engines cite you
Build a fixed list of 20 to 30 questions your patients really ask, run them in each engine on a schedule, and record which pages get cited. Compare the same list month to month rather than judging from a single check.
The reason to do it repeatedly is volatility. In BrightEdge's data, Google AI Overviews had a citation volatility of 20.2%, against 2.1% for ChatGPT, so a result that looks like a win or a loss in one week may not survive the next.
BrightEdge also advises monitoring both platforms, since a source can be cited in one and absent from the other. Its advice to treat video as experimental fits what it saw: ChatGPT cited no video, while Google's video share swung by more than 50 percentage points over three months.
We've written up a healthcare AI search case study if you'd rather see this kind of tracking in a real setting. Whatever you find, fix the gaps in the order patients would feel them: a missing answer on a service page before a missing schema field.
Frequently asked questions
Do I need an llms.txt file or other AI-specific files?
For Google's AI Overviews and AI Mode, no. Google says you don't need to create new machine-readable files, AI text files or markup to appear in them. We haven't found equivalent published guidance from the other engines.
Will FAQ schema get my clinic cited?
Google says no special structured data is required for its AI features. If you add markup anyway, it has to match the visible text on the page.
Why does my clinic show up in one AI engine but not another?
The engines draw on different source mixes. BrightEdge found ChatGPT favoring hospital systems, government sites and specialty organizations, while Google's AI features spread citations more widely, so a source can be cited in one and absent from the other.
Should I invest in video or social content for AI visibility?
Treat it as an experiment. LLM Pulse found social and user-generated content made up 8.5% of citations in AI Overviews and 1.9% in ChatGPT, and BrightEdge saw ChatGPT cite no video at all.
Cover photo by Karolina Grabowska www.kaboompics.com on Pexels
Sources
- AI features and your website — Google Search Central
- Creating helpful, reliable, people-first content — Google Search Central
