Something shifted in how patients find doctors - and most practices have not noticed it yet. The traditional path was straightforward: patient has a symptom, Googles a question, clicks through to a website, finds a phone number, books an appointment. That path still exists, but a growing number of patients are now stopping at step two and asking ChatGPT or Perplexity instead of clicking any results at all.
The AI search engines answer the question directly. "Who are the best cardiologists in Phoenix?" gets an answer - with names, brief descriptions, and sometimes a recommended next step - without the patient visiting a single website. For practices not represented in those answers, it is invisible patient loss. For practices that are cited by name in AI-generated recommendations, it is a referral channel that did not exist two years ago.
This is not a distant future concern. ChatGPT is currently the second-largest referral source for many healthcare marketing websites, sending more traffic than Instagram, LinkedIn, or any individual directory. Google AI Overviews now appear in roughly 25 percent of health-related searches. Perplexity has built a loyal user base of exactly the type of analytically-minded patient who researches thoroughly before choosing a physician. The window to establish a presence in these channels before they become crowded is open right now. For context on how this fits into a broader digital strategy, understanding how to fully optimize your Google Business Profile is the foundational first step - it feeds directly into AI search visibility.
How ChatGPT, Perplexity and Google AI Overviews Actually Work
The three major AI search platforms operate on different mechanics, and understanding those differences changes how you prioritize your optimization efforts.
ChatGPT with Browsing
When a user asks ChatGPT a question that benefits from current information - "find me a reproductive endocrinologist in Austin who specializes in PCOS" - the model activates its browsing capability and actively searches the web. It reads multiple sources, synthesizes the information, and generates a response that cites its sources. This means ChatGPT is not working from a static database of your practice information. It is finding what is currently indexed and publicly accessible, which makes your real-time web presence the determining factor.
Perplexity
Perplexity is built around real-time search with transparent sourcing. Every answer includes clickable citations, which reveals exactly which sources the AI weighted most heavily. This transparency is valuable for practitioners because you can search for queries relevant to your specialty and geography, see which sources are being cited, and reverse-engineer exactly what you need to build or update. Perplexity weights recency heavily - a blog post published last month outperforms an older page on the same topic, even if the older page has more backlinks.
Google AI Overviews
Google AI Overviews synthesize answers from indexed content in Google's own search index. The signals that influence AI Overview inclusion overlap significantly with traditional E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) factors, but weighted differently. Structured content - especially FAQ-formatted pages, schema markup, and clearly attributed first-hand expertise - appears in AI Overviews more frequently than keyword-dense traditional SEO content. Google is using AI Overviews to surface the most direct, authoritative answer to a question, which means your content needs to be built around answering questions, not around ranking for them.
Why Most Medical Practices Don't Show Up in AI Answers
The absence of a practice from AI-generated recommendations is almost always traceable to one of three root causes:
- Thin citation footprint - AI engines trust consistency across multiple independent sources. A practice with a good website but minimal directory presence, few reviews, and no third-party mentions looks invisible to an AI trying to verify its existence and reputation. The more independently a practice is mentioned and described across the web, the more confident an AI is in recommending it.
- Content written for search engines, not for answers - Most healthcare websites are optimized for keyword rankings, not for being cited as an authoritative source. AI engines look for content that directly and clearly answers a specific question, ideally in the voice of an expert. A "Services" page that lists "Cardiology Services" is not what gets cited. A page that answers "What should I expect at my first cardiology appointment?" in a physician's own voice is exactly what gets cited.
- Weak or stale review signals - AI search tools synthesize reputation, not just existence. A practice with 12 three-year-old reviews reads as stagnant and potentially closed. Fresh, specific reviews that mention the physician by name, describe conditions treated, and reflect a consistent positive experience are weighted heavily. Practices with active review generation programs are dramatically overrepresented in AI recommendations relative to their size.
The Authority Stack AI Search Engines Actually Trust
AI engines do not discover your practice the way a human researcher would - reading your website and forming a judgment. They aggregate signals from dozens of sources simultaneously and generate a composite picture of who you are, what you do, and how trusted you are. Building that composite picture is the core task of AI search optimization.
Tier 1: Your Core Listings
These are non-negotiable foundations that every practice should have fully built out before worrying about anything else:
- Google Business Profile - must be verified, have accurate NAP (name, address, phone), complete service descriptions, the correct primary and secondary categories, and a minimum of 25 reviews with an average above 4.3 stars
- Healthgrades profile - the single most-cited healthcare directory by AI engines; a fully completed Healthgrades profile with specialty details, office information, and insurance accepted is essential
- Zocdoc profile - actively referenced by ChatGPT for "find a doctor who accepts X insurance" queries; online booking capability makes it especially prominent
- Vitals.com - another tier-one directory that AI engines regularly cite
- WebMD Doctor Finder - legacy but still indexed and trusted by AI search tools
- Your own website with physician bio - must include the physician's name, credentials (MD, DO, fellowship training), specialties, and conditions treated, written in the first or second person
Tier 2: Specialty and Regional Signals
Beyond the core listings, your specialty determines which additional sources AI engines consult. A dermatologist benefits from being listed in the American Academy of Dermatology's physician finder. An orthopedic surgeon should appear in the AAOS member directory. Regional signals - local newspaper features, hospital system staff pages, medical school alumni directories - add geographic credibility that makes "best orthopedic surgeon in [city]" recommendations more confident.
Tier 3: Independent Mentions and Citations
The most powerful signal for AI search is being mentioned by name in content that is not your own. A quote in a local newspaper article about flu season, a contribution to a Healthline piece about managing chronic conditions, an interview segment on a local TV station's website - these independent mentions are interpreted by AI engines as third-party endorsements of your authority. Building a media presence, even a modest one, creates a citation trail that generic competitors cannot replicate.
Optimize Your Website Content for AI Citation
AI engines are not reading your website the way a patient would - they are scanning for structured, citable answers. The content format that gets cited most consistently across ChatGPT, Perplexity, and Google AI Overviews shares a few defining characteristics.
Free Weekly Newsletter
One Marketing Insight, Every Tuesday
Join 2,400+ practice managers getting actionable healthcare marketing tips every week.
Build a Dedicated FAQ Section for Every Core Service
AI engines love FAQ content because it is pre-formatted for direct citation. When ChatGPT encounters a page that asks and directly answers "How long does recovery take after a knee replacement?" it has found exactly what it needs to answer that question for a user. Build a FAQ section on every major service page and in standalone blog content, and use the actual questions patients ask - not the questions you wish they were asking. Tools like Google's "People Also Ask" feature, Reddit searches in your specialty, and your own intake form data are the best sources for high-value FAQ topics.
Write FAQ answers in the first or second person: "At our practice, we typically see patients return to light activity within three weeks" outperforms "Recovery from knee replacement takes three to six weeks" because it carries a source identity. The AI can cite "Dr. Smith at XYZ Orthopedics says patients typically return to light activity within three weeks" - a named, attributable claim rather than an anonymous fact.
Write Condition Pages With Genuine Clinical Depth
Every condition your practice treats should have its own dedicated page that goes beyond a paragraph description. These pages do not need to be academic literature - they need to be genuinely helpful to a patient who just received a diagnosis and is trying to understand their options. Address: what the condition is, how it is diagnosed, what treatment options exist, what your practice's specific approach looks like, and what recovery or management looks like long-term. Pages like this get cited by AI engines because they are substantive and authoritative, not because they are well-optimized for traditional SEO.
Use Your Physician's Name and Credentials Consistently
AI engines are building a knowledge graph about your practice as they index it. Using a consistent name format - "Dr. Sarah Chen, MD, FACC" - across your website, directories, and third-party mentions helps the AI connect all those separate pieces into one authoritative profile. Inconsistencies like "Dr. Chen," "Sarah Chen MD," and "Dr. S. Chen" across different sources create ambiguity that weakens your profile. This consistency is also important from a local SEO perspective, which is why understanding the full picture of digital visibility for physicians matters as a complement to AI search strategy.
Schema Markup: The Technical Signal Most Practices Miss
Schema markup is structured data you embed in your website code that explicitly tells search engines - and AI engines - what type of content they are looking at. It is one of the clearest signals you can send about your practice's identity, and most healthcare websites do not have it properly implemented.
The most valuable schema types for a medical practice include:
- MedicalOrganization schema - describes your practice as a medical entity with name, address, phone, specialties, and medical services offered
- Physician schema - creates a structured profile for each physician with credentials, specialty, affiliated hospital, and language spoken
- FAQPage schema - marks your FAQ content explicitly so AI engines can parse it as a question-and-answer structure for direct citation
- LocalBusiness schema - reinforces your geographic identity with address, hours, service area, and geographic coordinates
- Review schema - allows your aggregate review rating to appear in search results and be parsed by AI tools evaluating your reputation
Implementing schema correctly requires either technical knowledge or a web developer, but it is a one-time investment that pays ongoing dividends. A practice with clean, complete schema markup is unambiguous to any AI engine crawling its site - it is a named, credentialed physician at a specific address in a specific specialty. That clarity translates directly to citation confidence.
Review Strategy as AI Search Strategy
Generating a consistent flow of fresh, specific reviews is not optional for AI search visibility - it is table stakes. AI engines weight both quantity and recency, and they read the content of reviews, not just the star count. Here is what this means in practice:
Volume and Recency
A practice with 200 reviews over three years looks very different to an AI engine than a practice with 50 reviews from the last six months. The second practice appears active, trusted, and currently operating. The first may have a great average rating but its signal strength is decaying. Set a target of at least 3 to 5 new reviews per month per physician and maintain it continuously - not in bursts when you remember to ask.
Specificity
Generic five-star reviews ("Great doctor! Highly recommend!") contribute less to AI visibility than specific reviews that mention the physician by name, the condition treated, and the experience in detail. When ChatGPT summarizes a practice's reputation, it draws from review content. "Dr. Martinez took the time to explain my Hashimoto's diagnosis in a way no other doctor had" is exactly the type of content that leads an AI to recommend Dr. Martinez by name when someone asks about endocrinologists who are good at explaining thyroid conditions.
Platform Diversity
Google reviews are the most important single source, but AI engines also read Healthgrades, Vitals, Zocdoc, and in some cases Yelp for healthcare practices. Building a review presence across multiple platforms creates the independent corroboration that makes AI recommendations more confident. A patient who reads positive reviews about your practice on three separate platforms before asking ChatGPT will likely find that ChatGPT's answer confirms what they already found - which is exactly the reinforcement loop you want to create.
Tracking Your AI-Driven Traffic in Google Analytics
One of the most practically useful things you can do right now is set up proper tracking to see how much traffic AI search engines are already sending to your website. The data may surprise you - most practices are receiving more AI-referred traffic than they realize, but it shows up in ambiguous ways without proper configuration.
In Google Analytics 4, AI-driven referrals currently appear as:
- chatgpt.com - direct referrals from ChatGPT browser use or when users click citations
- perplexity.ai - referrals from Perplexity answers when users click through to your site
- claude.ai / anthropic.com - referrals from Claude search queries
- Direct / (none) - a portion of this bucket is AI-influenced, as users who saw your practice recommended in an AI answer type your URL directly or search your name without a referral chain
Create a custom channel grouping in GA4 that buckets chatgpt.com, perplexity.ai, claude.ai, and similar AI engine domains into an "AI Search" segment. Track this alongside your organic and paid channels as a growth metric. For newer practices still building their marketing foundation, this tracking provides early evidence of which AI channels are most responsive to your content and directory investments.
What to Do This Week
AI search optimization does not require a complete website overhaul. The highest-impact changes can be made in days, not months. Here is a concrete starting point:
- Audit your Google Business Profile - confirm all information is accurate, your primary category is specific (not just "Doctor"), and you have at least one photo uploaded in the last 30 days
- Claim and complete your Healthgrades and Vitals profiles - if you have not done this, it takes less than an hour and pays consistent dividends
- Add a FAQ section to your homepage or primary service pages - start with 5 to 7 questions that patients actually ask at intake, and write detailed, attributed answers
- Implement FAQPage schema markup - ask your web developer to add this to your FAQ content; it takes one to two hours and materially improves AI citation probability
- Start a review generation system - even a simple text message to patients after their appointment asking for a Google review, if done consistently, will build a review footprint that compounds over time
- Search for yourself in Perplexity and ChatGPT - ask "who are the best [your specialty] doctors in [your city]?" and see what comes up; this tells you exactly where you stand and what your competitors are doing that you are not
The practices that will dominate AI search in 2027 and beyond are building their authority stack right now, before the channel becomes as competitive as Google Maps or paid search. The entry cost is information and consistency, not advertising budget - which means early movers have a durable advantage that late entrants cannot simply buy their way past.