How to get patients
from ChatGPT.

A growing share of patients no longer start at a list of ten blue links. They ask ChatGPT, or read the AI overview Google now places above its own results, and they take the answer the machine hands back. For a medical or dental practice, that changes the goal: the job is no longer only to rank, it is to be the source the model cites when it answers, the best dermatologist near me, who treats this, where should I go.

The discipline for that is called generative engine optimization, or GEO. It is new enough that almost no healthcare practice has done it deliberately, which is exactly why the field is open. What follows is what GEO actually is, how AI engines decide which practices to name, and the specific, mechanical playbook we run to earn those citations.

The short version, for skimmers.

• AI engines answer the question directly and cite a few sources. Being one of those sources is the new first page.
• They pull from the open web, not just Google’s top ten. Research on generative engines finds a large share of cited pages do not rank in the traditional top results.¹
• They reward content that is answer-first, self-contained, statistic-backed, and quotable, because the model lifts sentences out of context to compose its reply.¹
• They lean on your entity, the consistent, credentialed identity of your practice across your site, your Google profile, and the authoritative directories.
• There is no paid slot for the organic citation. You earn it.

What is generative engine optimization?

Generative engine optimization (GEO) is the practice of structuring your content and your online presence so that AI answer engines, ChatGPT, Google’s AI Overviews, Perplexity, and the rest, cite your practice when they answer a patient’s question. Where SEO optimizes to rank a clickable link, GEO optimizes to be the trusted source a model quotes in an answer the patient may never click past.

The two are siblings, not rivals. Everything that earns authority in search, real expertise, structure, trust, citations, helps in GEO too. The difference is emphasis: generative engines favor passages that stand on their own, carry a statistic or a clear definition, and read as quotable fact, because that is what the model can safely lift into its composed answer.¹

How AI engines decide which practices to name.

An answer engine retrieves candidate sources from across the web, reads them, and composes a reply that cites the few it trusted most. Three things decide whether your practice is in that set.

Retrievability.The engine has to be able to find and read you. That means crawlable, well-structured pages on your own site and a consistent presence on the authoritative surfaces these models lean on for local and medical questions, your Google Business Profile, your specialty’s directories, and real reviews. A practice that exists only as a thin template site with no profile footprint is invisible to the retrieval step.

Quotability. Research on generative engines found that adding clear citations, statistics, and quotable, authoritative phrasing measurably increased how often a source was cited, by a meaningful margin over generic copy.¹ Answer-first paragraphs, definition sentences, and numbered steps are not just good for featured snippets anymore; they are the format the model can lift.

Trust and entity clarity. Medical answers are your-money-or-your-life territory, and engines weight credibility heavily. Visible physician authorship, credentials, citations to primary sources, and a clean, consistent identity across the web, reinforced by structured data that names your providers and procedures, all tell the model you are a safe source to quote.

The GEO playbook for medical practices.

None of this is exotic. It is disciplined execution of things a serious practice should already be doing, tuned for retrieval and citation instead of clicks.

1. Write answer-first. Open every page and section with a direct, 40-to-60-word answer to the question it targets, phrased so it makes sense lifted out of context. The model reads the first lines hardest.
2. Back claims with numbers and sources. A statistic with a credible citation is more quotable than an adjective. Cite primary sources for any clinical or prevalence claim, the same discipline that earns medical trust.
3. Structure with schema. Physician, MedicalProcedure, FAQ, and Review schema tell the engine exactly who your providers are and what you treat, which is how it attributes a claim to a credentialed source. See the healthcare schema graph.
4. Win the entity, not just the page. Keep your name, address, specialty, and providers identical across your site, Google Business Profile, and the directories. Inconsistent identity makes a model hesitate to cite you.
5. Earn real reviews and local authority.The same signals that decide the map-pack feed the model’s view of who is a legitimate local provider. Review velocity and a strong Google profile do double duty.
6. Stay fresh. Generative engines favor current content, which is an advantage a practice can act on immediately: ship, update, and date your pages.
7. Tell the engine what you don’t do.The sharpest, most-overlooked move in medical AEO is defining the practice’s boundaries in plain content and schema: no Medicaid, no walk-in urgent care, no pediatric cases, whatever applies. Explicit exclusions stop a model from hallucinating services a patient will be let down to find missing, and they sharpen the entity at the same time.

A note on patient data.

Chasing AI visibility does not change the rule that matters most: patient information never goes to a third-party platform it shouldn’t. Keep PHI out of analytics and ad tools as you build out intake and tracking around these new surfaces. The mechanics are in our guide to HIPAA-compliant acquisition.

The window is open now.

The reason to act is timing. AI answer engines are still young, their citation behavior is still being shaped, and almost no healthcare practice is optimizing for it deliberately. Freshness counts, the field is uncrowded, and the practices being cited today are compounding an authority advantage that is harder to unseat the longer it stands. This is the rare moment where moving early is most of the win.

Questions practices ask about AI search.

How do I get my practice recommended by ChatGPT?

You become a source the model can find and trust. That means clear, answer-first content on your own site, consistent presence across the authoritative profiles AI engines pull from (Google Business Profile, your specialty's directories, real reviews), credentialed author attribution, and structured data that tells the engine who you are and what you treat. There is no paid placement; you earn the citation the way you earn an organic ranking, but the signals are tuned for retrieval rather than ten blue links.

What is the difference between SEO and GEO?

SEO optimizes to rank a page in a list of links a person then clicks. GEO, generative engine optimization, optimizes to be the source a language model cites when it answers the question directly, with no list to click. They overlap heavily (both reward authority, structure, and trust), but GEO rewards self-contained, quotable, statistic-backed passages because the model lifts sentences out of context to compose its answer.

Does an llms.txt file help me get cited?

Not in any way that has been demonstrated. llms.txt is a proposed convention for listing your key pages in a flat file, and it is cheap to publish, but as of 2026 there is no evidence the major AI engines use it to decide citations. Treat it as harmless housekeeping, not a ranking lever. The things that actually move citation are content quality, structure, authority, and entity consistency.

Can I pay to appear in AI answers?

Not for the organic citation itself, today. AI engines compose answers from retrieved web content and cite the sources they used; that placement is earned, not bought. Some surfaces are beginning to test ads around AI answers, but the recommendation a patient reads as the model's own answer is not a paid slot. That is exactly why it is worth optimizing for now, before it is.

How do I measure whether AI is sending me patients?

Imperfectly, for now. Watch for referral traffic from chat and AI-overview surfaces in your analytics, ask new patients how they found you and add an AI option, and periodically prompt the major engines with the searches a patient would use to see whether you are named. It is less precise than rank tracking, but the direction is readable, and the practices being cited today are compounding an advantage.

1. 1. Aggarwal et al., “GEO: Generative Engine Optimization,” research introducing and measuring generative-engine citation factors (statistics, quotations, and cited sources among the highest-impact). Published research, 2024.

AI answer engines change quickly, and citation behavior is not fully documented by the platforms. The patterns here reflect current research and Macbach’s operating approach; specifics will keep shifting.