GEO for Healthcare: How Patients Find Doctors When They Ask AI

A patient asking an AI assistant “who’s a good dermatologist near me” does not get a results page to scroll. The model names a short list and stops, often pulling details (hours, insurance, reviews) from a directory the practice never built and may not actively manage. That is a different game than ranking in Google’s local map pack, and the practices winning it are not always the ones with the best medicine, just the clearest footprint on the directories AI engines actually trust.

Here is what a body of 2026 research on patient AI adoption and provider citation actually shows. (If you work with restaurant groups or home services businesses too, see our companion guides to GEO for restaurant groups and GEO for home services.)

How Many Patients Are Already Asking AI to Help Find a Doctor?

Two separate 2026 surveys, run by different organizations with different methods, point the same direction.

The exact percentage moves depending on how the question is asked and which patients are counted, but the direction does not: adoption roughly doubled in a year by more than one measure, and it is not slowing down.

Where AI Engines Actually Get Provider Data

The old assumption that a practice’s own website is the main thing AI reads does not hold up for healthcare specifically. A citation-tracking study spanning 6.8 million AI citations across ChatGPT, Gemini, and Perplexity found healthcare has the highest listings-dependence of any vertical measured: 52.6% of healthcare citations come from directory listings, not a practice’s own site. A practice’s website only enters the running at all if a directory already links to it.

Within that directory layer, rater8’s breakdown of what each AI tool actually pulls from shows the engines agree on a “big three”, Healthgrades, Vitals, and WebMD, but otherwise split:

A separate Doctor Rank audit of Perplexity specifically found Zocdoc acting as the primary citation driver for local healthcare queries on that engine, which lines up with rater8’s finding. For broader authority content, BrightEdge reported that NIH.gov alone captured roughly 60% of citation share inside Google’s AI Overviews for healthcare queries, a reminder that government and institutional sources still carry real weight for informational questions even when they carry none for “find me a provider” questions.

What Actually Gets a Practice Named

Reviews carry more weight, and more risk, than a year ago. rater8’s 2026 report found 75% of patients will not book a provider rated below 4.0 stars, and 55% have walked away from at least one provider because of online reviews, up 15 points from 2025. 66% say how a provider responds to reviews affects their trust in that provider, up 24 points. A profile with a low rating, old reviews, or no responses is a real liability, not a cosmetic one.

Completeness on the right directories matters more than completeness everywhere. Since the big three (Healthgrades, Vitals, WebMD) feed every major AI engine to some degree, and Zocdoc and RateMDs feed Claude and Perplexity specifically, a practice with gaps on those five platforms is invisible to a meaningfully different degree than a practice with gaps on a directory none of the engines read from.

AI gets provider information wrong often enough to matter. rater8 also found 66% of patients had encountered incorrect provider information from an AI tool, and 60% still trusted an AI summary without verifying it. That cuts both ways for a practice: a wrong address or an outdated insurance list posted to a directory does not just sit there unseen, it gets repeated by an AI assistant as fact.

Patients trust Google’s AI Overviews more than they trust the organic results below them. Patients named AI Overviews their most-trusted result type (37%), ahead of organic results (20%), the local map pack (13%), and paid results (7%). An accurate, complete Google Business Profile is still foundational even as patients shift toward chatbots for the same question.

The GEO Playbook for Healthcare Practices

• Claim and complete the big three first: Healthgrades, Vitals, and WebMD. Every major AI engine draws on at least one of these. A gap here is a gap every engine can see.
• If you see patients directly, get on Zocdoc and RateMDs. Claude and Perplexity lean on these specifically, and a separate audit found Zocdoc the primary citation driver for Perplexity’s local healthcare answers.
• Treat review responses as part of the record, not customer service. With patient trust in review-response behavior up 24 points in a year, an unanswered negative review reads as neglect to the next patient and to the AI summarizing your profile.
• Audit every directory for accuracy, not just presence. Insurance accepted, hours, and specialties listed inconsistently across Healthgrades, Vitals, and your own site create the kind of conflicting signal that produces the wrong answer in an AI summary, the exact failure 66% of patients say they’ve already run into.
• Keep Google Business Profile current even as AI adoption grows. Patients still trust AI Overviews, which draw heavily on Google’s own data, more than any other result type.
• Watch the specialty and news mentions, not just the directories. For informational queries, authority sources like NIH.gov and major health publishers still carry outsized citation weight; a practice with a quoted specialist or a published case note gains ground AI engines read as credibility signal.

Common Mistakes to Avoid

• Treating all directories as equally important. A gap on a directory none of the major AI engines read from costs little. A gap on Healthgrades, Vitals, WebMD, Zocdoc, or RateMDs costs real visibility.
• Ignoring negative reviews instead of responding to them. Review-response behavior now visibly affects patient trust, and an AI assistant summarizing a profile has no way to see context that only exists in an unwritten response.
• Assuming a strong website replaces strong directory profiles. For healthcare specifically, directory listings dominate AI citations more than in almost any other industry; a beautiful website that no directory links to may never enter an AI engine’s candidate set at all, the same way stopping SEO work too early erases months of accumulated gains: a strong foundation still has to stay maintained.
• Letting outdated information sit on a directory. Two in three patients say they’ve hit incorrect AI-sourced provider information, and most still trust it without double-checking. An old address or dropped insurance plan does not just confuse a human reader, it becomes a wrong answer repeated at scale.

How to Measure AI Visibility for a Healthcare Practice

Run a fixed set of patient-style questions (“best pediatric dentist in [city],” “who takes [insurance] near me”) across ChatGPT, Gemini, and Perplexity on a recurring schedule, and log which directories and sources each answer cites. That list tells you exactly where the next cleanup should happen, whether that is a Healthgrades photo gap or a stale Zocdoc insurance list. This is the same discipline behind measuring what an SEO campaign is actually doing: track the leading indicators on a schedule, not a single lucky screenshot.

Getting Started

The pattern holds across every study here: AI engines recommend providers based on what directories and review platforms say about them, filtered through whichever sources each engine happens to trust most. A practice’s own website plays a smaller role in healthcare than in almost any other industry. Building and maintaining accurate, complete, actively managed profiles on the handful of platforms that actually feed AI answers, rather than spreading effort evenly across every directory that exists, is the real work of GEO for a healthcare practice.

GEO for Healthcare: Frequently Asked Questions

What is GEO for healthcare?

GEO, or generative engine optimization, is the practice of structuring a healthcare provider’s directory listings, reviews, and online presence so AI tools like ChatGPT, Gemini, and Perplexity name it when a patient asks for a recommendation. It runs alongside traditional local SEO rather than replacing it.

Which directories matter most for AI visibility: Healthgrades, Vitals, WebMD, Zocdoc, or RateMDs?

All five, for different engines. Healthgrades, Vitals, and WebMD feed most major AI tools at some level, while Zocdoc and RateMDs specifically drive Claude’s and Perplexity’s local healthcare answers. A practice with gaps across all five is invisible to a meaningfully wider set of AI tools than one with gaps on a single platform.

Does a strong practice website matter if AI mostly cites directories?

It still matters, but less than in most other industries. A 2026 citation-tracking study found healthcare has the highest listings-dependence of any vertical studied, at 52.6% of citations from directories rather than first-party sites. A website still needs to exist and rank, but directory accuracy is the bigger lever for AI-specific visibility.

How much has patient use of AI for finding a doctor grown?

By more than one measure, it roughly doubled in a year. A 2026 industry report found AI influence on doctor choice rose from 17% to 36% year over year, while a separate measure from the same period found AI use for provider search rose from 31% to 47%. A broader Pew Research survey found 34% of US adults use AI chatbots for some health reason.

Can AI get provider information wrong?

Yes, and often enough to matter. One 2026 survey found 66% of patients had encountered incorrect provider information from an AI tool, and most still trusted the AI summary without verifying it further, which makes keeping directory listings accurate a patient-safety issue as much as a marketing one.