AI Engines · 9 min read · July 15, 2026
How Medical Practices Get Recommended by ChatGPT and Gemini
More people are skipping Google and asking ChatGPT or Gemini directly: 'Which family practice near me is still taking new patients?' If your practice doesn't turn up in that answer, you don't exist for that patient. Generative engine optimization is how you make sure AI systems know your practice, describe it accurately, and recommend it — factually, verifiably, and within the limits of German medicinal advertising law.
Why patients now ask ChatGPT instead of Google
Your patients' search habits have shifted, more quietly than most practice owners realize. The person who used to type 'dermatologist appointment Munich Schwabing' into Google now asks a full question in ChatGPT: 'I've had an itchy rash on my arm for weeks — which specialist in Munich should I see, and what should I look for when picking an appointment?' The AI doesn't answer with ten blue links, it gives one concrete recommendation. Your practice is either in that answer or it isn't.
That's the real difference from a classic search engine. On Google you can compete for the third or fourth spot and still get the click. With ChatGPT and Gemini there are often only two or three practices named at all. Visibility has become binary: named or invisible. In crowded specialties like dermatology, orthopedics, or psychotherapy, that's what decides whether new self-pay and private patients ever learn you exist.
Keep this in perspective: AI systems don't replace a visit to the doctor, but they're increasingly the first point of contact. They pre-filter, they sort, they recommend. If you ignore that pre-filter, you lose patients before the phone even rings.
How ChatGPT and Gemini actually get their information about your practice
Language models don't invent their recommendations, they pull them from sources they judge trustworthy. For medical practices, that mainly means your Google Business Profile, review platforms like Jameda or the health insurers' doctor-review portal, directories such as the physician search run by the Associations of Statutory Health Insurance Physicians, your own website, and sometimes regional press coverage. Whatever is stated consistently across these sources is what ends up in the AI's answer.
The problem for most practices is inconsistency. The website says 'Dr. Meier & Colleagues Practice,' Jameda says 'Meier Family Practice,' and the Google profile still shows the old address from before the move. A person can work around that. A language model can't — contradictions like that read as uncertainty, and the AI would rather not name your practice at all than risk repeating something wrong.
Your first job isn't marketing, it's data hygiene. Name, address, phone number, specialty, hours, and services need to match, word for word, across every source. It sounds mundane, but it's the single highest-leverage fix available to you.
The questions your patients are really asking the AI
To show up in recommendations, you need to know what's actually being asked. With doctors, the queries rarely use the plain specialty label. Typical phrasings look like 'Which family doctor in Cologne-Ehrenfeld is still accepting new statutory-insurance patients?', 'I need a short-notice appointment with a gynecologist in Leipzig, who has availability?', or 'Which practice in Stuttgart offers acupuncture for migraines?'
What stands out is how tightly these questions hang on concrete needs: statutory-insurance acceptance, accessibility, languages spoken by the team, specific procedures, short-notice slots, experience with children or anxious patients. Name these attributes explicitly on your site. An AI can only recommend what exists somewhere as text — if nothing states that you offer diabetes education or Turkish-language consultations, you won't surface for those questions.
Collect the real questions from your own front desk. What do callers ask before booking? What shows up in prospective patients' emails? Those sentences are your best raw material, because they're phrased exactly the way patients talk to the AI.
Build a website the AI can actually understand
Language models like clarity. Long, meandering welcome copy full of phrases like 'holistic care from the heart' doesn't help them. What helps is unambiguous, fact-dense pages: one service page per specialty, clear headings, body text that answers a specific question. Instead of 'we offer state-of-the-art diagnostics,' say exactly which exam you run, for which complaints, and who it's suited to.
Real FAQ sections are especially effective. Answer 'Do you accept new statutory-insurance patients?' in full sentences on your page, and you've handed the AI a ready-made, citable answer. Add structured data in the background — schema markup for medical organizations. That lets the machine recognize, unambiguously, that this is a medical practice with specific specialties and hours.
On every page, think in terms of location plus occasion. 'Dermatologist mole screening Dresden Neustadt' is more specific, and therefore more findable, than a general dermatology page. The more precisely your content matches a real concern, the more likely you become the answer.
Reviews and reputation as a trust signal
AI systems weigh reputation heavily when deciding whom to recommend. A practice with many current, varied reviews reads as more trustworthy to the model than one with no footprint online. That doesn't mean buying a flood of five-star reviews — quite the opposite. Both review platforms and language models recognize suspicious patterns, and in healthcare that backfires harder than it helps.
Instead, run an honest, systematic process. At the end of an appointment, ask satisfied patients for a review — don't oversell it — and respond professionally to critical ones. Handling criticism matter-of-factly is itself a quality signal. Mind confidentiality throughout: in public replies, never confirm that someone was even a patient, and never disclose treatment details.
Consistency over time beats a one-time spike. A steady trickle of real feedback over months builds a stable picture that shows up in AI answers. Reputation here is a marathon, not a sprint — and that's exactly what makes it hard to fake.
The legal limits: take medicinal advertising law seriously
Unlike a restaurant, as a doctor you operate under strict advertising rules. The German Medicinal Advertising Act (Heilmittelwerbegesetz) and medical professional codes prohibit misleading, promotional, or comparative advertising. Claims like 'the best cardiologist in town' or promises of a cure are off-limits, even inside a page written to please an AI. GEO for doctors means informing accurately and verifiably, not exaggerating.
That's not really a restriction, it's an advantage. Language models already prefer sober, verifiable facts and distrust superlatives. Describe cleanly which services you offer, what qualifications your team holds, and which patient groups you're suited for, and you satisfy the AI and professional law at the same time. Facts beat marketing language in both worlds.
Be especially careful with before-and-after photos, testimonials about individual treatments, and specific success figures, particularly in aesthetic specialties. When in doubt, check your state medical association's guidelines or get a quick legal review before publishing.
How to check whether the AI already knows about you
Before you optimize anything, get an honest baseline. Ask ChatGPT, Gemini, and Google's AI Overviews the same questions your patients would ask: 'Recommend a family practice in my neighborhood,' 'Which orthopedist in my town treats sports injuries?' Note whether you're named, whether the details are correct, and who gets recommended instead.
You'll often find surprises: outdated hours, the wrong specialty listed, or a practice that no longer exists being recommended. Each mismatch is a concrete task. Also check which sources the AI is drawing from — many systems will name them if you ask — so you know which listing to fix next.
Repeat this check on a regular cadence, say quarterly. AI models get retrained and their answers shift. What's accurate today can be stale in six months. This routine costs little time and tells you plainly whether your work is paying off.
A realistic roadmap for your practice
Start with the basics, not the polish. First, clean up your Google Business Profile, reconcile every directory listing, and lock down consistent core data. This first step tends to deliver the biggest jump, because it removes the most common cause of invisibility: contradictory information the AI doesn't trust.
Next, rework your website around real patient questions: clear service pages, an honest FAQ section, structured data behind the scenes. In parallel, set up a steady, ongoing review process. Treat it as a standing habit built into daily practice, with one team member responsible for keeping it going, not a one-off campaign.
Give it time. GEO doesn't work overnight, because models need time to absorb new information. But practices that start now build a lead that latecomers will struggle to close later. Get ChatGPT and Gemini to know your practice correctly today, and you'll be the one getting recommended tomorrow, while others are still deciding whether any of this matters.
Name your specialties clearly
The AI recommends you when it can classify your practice unambiguously. So don't just write 'family doctor' — name concrete specialties: diabetology, travel medicine, outpatient minor surgery, disease management programs. The more precisely your website and directory listings word these services, the more reliably ChatGPT can match you to a specific patient question.
Think in the phrasing real patients use. Almost nobody types 'diabetologist' — they ask 'Where in Rosenheim can I get my blood sugar under control long-term?' If your copy picks up that everyday language and connects it to the clinical term, you build exactly the bridge the language model needs.
Don't scatter your specialties across ten subpages. A clearly structured service overview with a short description per offering is easier for the AI to parse than long-form copy where the actual core competency gets buried.
Make availability and new-patient admission transparent
One reason the AI won't name a practice is simple: it's unclear whether you're even taking new patients. State explicitly on your website whether your patient roster is open, whether there's a wait, and which insurers you're approved for. The model treats that clarity as a signal to actively recommend you.
Office hours, appointment booking, and emergency instructions belong on a fixed, clearly worded page too. If a patient asks 'Which practice near me is still open today?', the AI can only answer if your hours are current and consistent — ideally identical between your Google Business Profile and your website.
Keep this information consistently maintained. Nothing hurts your chances of being recommended more than conflicting details across listings. An outdated holiday notice or a wrong phone number reads to the model as an uncertainty signal and can knock you out of the answer entirely.
Answer common patient questions as a mini FAQ
Add a short section to your website with real patient questions, answered directly: How does the first exam go? Do I need a referral? How do I get a prescription without an appointment? Question-and-answer blocks like this are ideal for ChatGPT and Gemini, because they match exactly the format the AI uses to answer patients.
Keep every answer to two or three sentences, factual, with no promises of a cure. Stick to verifiable statements about process, equipment, and logistics. That keeps you inside the legal limits while still giving the model concrete, citable substance.
As a starting point: collect the ten most common questions from your front desk over four weeks, write a short answer to each, publish them together, and check back after a few weeks to see whether ChatGPT is naming your practice for exactly those questions.
Common questions
Is it even legal to optimize my medical practice for AI recommendations?
Yes, as long as you stay within the Medicinal Advertising Act and your professional code. Factual, verifiable information about services, qualifications, and hours is permitted and even expected. What stays off-limits is promotional advertising, superlatives like 'best doctor,' promises of a cure, and misleading claims. GEO for doctors is fact maintenance, not advertising — which happens to be exactly what AI systems prefer anyway.
How long does it take before ChatGPT or Gemini recommends my practice?
Think in months, not days. Directory and Google-profile corrections tend to show up relatively quickly, since many AI answers pull from these sources live. The models' trained base knowledge updates more slowly. A realistic window is three to six months of building consistent data and real reviews. Check your progress quarterly with your own test questions.
Do I need to pay for reviews or ads to get named?
No. Bought reviews are risky in healthcare and get flagged by both platforms and models. The most effective lever is free: clean, uniform core data across every source and a fact-rich website. An honest review process built on real patient voices costs you attention in daily practice, not an ad budget. Paid ads don't influence editorial AI recommendations.
Read on