Let me tell you about a shift I am seeing that most therapists have not noticed yet. When someone asks ChatGPT or Perplexity “find me a therapist for anxiety in Chicago” — and yes, people are doing this — the AI does not scan search engine results the way Google does. It pulls information from your website in a completely different way. If your site is not structured for AI extraction, it is invisible to that entire search.
This is not some future hypothetical. Google AI Overviews now appear in over 40% of searches. ChatGPT handles billions of queries per month. A meaningful percentage are health-related. If your practice is not optimized for AI retrieval, you do not exist in the search the patient actually runs. This guide breaks down how each AI platform works and what you need to do differently from traditional SEO.
What Changed and Why You Should Care
Traditional SEO ranks pages based on relevance, authority, and backlinks. GEO (Generative Engine Optimization) determines whether AI systems extract and cite your content when answering questions. The mechanics differ: SEO rewards keywords and backlinks. GEO rewards structured data, entity clarity, and definitive answers. SEO produces a list of options. GEO produces a recommendation.
Here is how the two approaches diverge in practice:
| Factor | Traditional SEO | GEO (AI Search) |
|---|---|---|
| Primary signal | Backlinks, domain authority | Structured data, entity trust |
| Content style | Keyword-optimized, 1500+ words | Direct answers with supporting context |
| Target format | Ranked list of URLs | Extracted quote with citation |
| Citation format | Title + meta description in SERP | Inline answer + source URL |
| Success metric | Position 1-10 in Google SERPs | Appears in AI-generated answer |
| Geographic relevance | Google Maps + local pack ranking | Address + schema cross-referencing |
| Key vulnerability | Algorithm updates (Helpful Content, Core) | Citation accuracy, source freshness |
Understanding this table changes how you prioritize. A page that ranks #3 in Google but contains clear, structured, authoritatively sourced answers may rank higher in AI outputs than the site sitting at #1 — because AI extraction favors clarity over domain authority alone.
How Each AI Platform Retrieves and Cites Therapy Providers
Not all AI search engines work the same way. Optimizing for one does not guarantee visibility in another. Here is how each major platform handles therapy-related queries.
ChatGPT Search (GPT-4 and GPT-4 Turbo with browsing)
ChatGPT with browsing capabilities retrieves live web results and synthesizes them into conversational answers. It heavily weights content from sites with clear schema markup, fresh publication dates, and strong topical authority. For therapy queries, ChatGPT cross-references Google Knowledge Graph data — so your consistency across Knowledge Graph, Wikipedia, and your own site matters. ChatGPT cites sources inline, so the goal is to be the source it extracts from, not the source it ignores.
Key optimization: Ensure your practice name, NAP, and specialty appear identically on your website, Google Business Profile, and any directory listings. ChatGPT resolves entity identity across sources — mismatches cause it to discount your content.
Claude (Anthropic)
Claude’s approach to web content is more cautious. Anthropic trains Claude to prioritize authoritative, well-cited sources, especially in healthcare. Claude is more likely to extract from .edu, .gov, and established medical organizations for generic queries. For local therapy queries, Claude looks for verified credentials — license numbers, professional association memberships, and verifiable education history. Claude also values recency: content published or updated within the last 12 months is significantly more likely to be cited.
Key optimization: Include your license number and state on your about page and profile schema. List your professional memberships (APA, ACA, NASW, etc.) prominently. Keep every page updated with a recent “last updated” date.
Perplexity
Perplexity is the most citation-forward AI engine. It cites sources with numbered footnotes and pulls from multiple pages to build its answer. Perplexity heavily favors pages with FAQ schema, clear Q&A formatting, and direct answers to specific questions. For therapy queries, Perplexity often extracts from FAQ pages, service descriptions, and blog posts that answer specific patient questions (“What does EMDR therapy treat?”, “How long does CBT take to work?”). Perplexity also selects sources based on authority signals at the domain level.
Key optimization: Build FAQ sections with FAQ schema on every service page. Answer common patient questions in the first paragraph of each section. Ensure your domain has sufficient backlink authority to pass Perplexity’s domain score filter.
Google AI Overviews
AI Overviews appear directly in Google search results above traditional listings. They extract from pages that Google already ranks well — but not exclusively. Pages ranking on page 2 can appear in AI Overviews if they contain the most definitive answer to a specific sub-question. AI Overviews favor structured lists, tables, and step-by-step instructions. For therapy queries, Google’s AI Overviews typically pull from pages with MedicalBusiness or HealthAndBeautyBusiness schema, clear author attribution, and YMYL-compliant content.
Key optimization: Add author bios with credentials to every clinically relevant page. Use tables and bulleted lists for treatment comparisons. Ensure MedicalBusiness schema is validated and error-free. For detailed schema implementation guidance, see our Technical SEO Checklist for Therapist Websites.
Three Things AI Search Needs From Your Site
Schema markup. Structured data tells AI engines explicitly what your content means. MedicalBusiness schema on your homepage. Person schema on your about page. Service schema on each service page. FAQ schema for question-answer pairs. Without schema, AI engines must guess what your content is about — and many of them will guess wrong.
Entity clarity. Your practice needs consistent representation across Google Knowledge Graph, Wikidata, Apple Business Connect, and Bing Places. Name, credentials, specializations, address — identical everywhere. AI systems cross-reference these sources to build their understanding of who you are. Every inconsistency creates ambiguity that reduces citation confidence.
Definitive answers. AI extracts direct answers, not context-heavy paragraphs. Your content should answer questions in the opening sentences of each section. “Anxiety therapy helps by…” is better than “Anxiety is a common condition that affects millions of people and has been studied extensively…” The AI is looking for the answer, not the warm-up.
Schema Markup: Your AI Visibility Blueprint
For schema implementation details, see Technical SEO Checklist. Schema markup is genuinely the single most impactful thing for AI visibility. Here is the specific markup each platform relies on:
MedicalBusiness Schema (Required by all platforms)
This schema must be on your homepage. It tells AI systems that your site represents a legitimate healthcare provider. Required fields: @type, name, description, url, telephone, address (streetAddress, addressLocality, addressRegion, postalCode), medicalSpecialty, and knowsAbout (list modalities and conditions you treat). Optional but recommended: sameAs (links to your GBP, Psychology Today, LinkedIn), aggregateRating, and priceRange.
Person Schema with Medical Credentials (Claude and Perplexity favor this)
Every clinician needs Person schema on their profile. Include: name, honorificSuffix (PhD, LCSW, LMFT), knowsAbout (specialties), knowsLanguage, and most critically — identifier with propertyID set to “license” and value set to your actual license number. Claude specifically uses license numbers as authority signals.
FAQ Schema (Perplexity extracts heavily from this)
Perplexity builds its answers from FAQ pages and FAQ-annotated sections more than any other content type. Add 5-10 FAQ items per service page covering the most common patient questions. Each FAQ item needs question and acceptedAnswer with text — not just one-line answers, but 2-3 sentence responses that an AI can extract as a complete answer.
How to validate your schema: Use Google’s Rich Results Test (search “Rich Results Test” — it will be the first result) to validate every schema you implement. Schema markup with errors is treated as though it does not exist. Test individually per page.
The llms.txt Standard: A New Visibility Opportunity
llms.txt is an emerging standard that creates a dedicated file AI engines can read for structured information about your site. It sits at yourdomain.com/llms.txt and provides a machine-readable summary of your practice — specialties, location, credentials, and recommended content for AI extraction.
Unlike robots.txt which controls access, llms.txt directly feeds AI engines with curated information. Here is what your llms.txt should contain:
Your practice name, a one-paragraph description, a list of specialties with links to dedicated service pages, your key credentials and affiliations, your service areas, and links to your most authoritative content (about page, FAQ page, clinical approach page).
To create yours, save a plain text file as llms.txt in your site root. Format: one section heading per line prefixed with #, one link per line with description. Example structure:
# Austin Trauma Therapy
Evidence-based trauma therapy for veterans and first responders in Austin, TX.
– About: https://yoursite.com/about
– Trauma Therapy: https://yoursite.com/trauma-therapy
– EMDR: https://yoursite.com/emdr-therapy
– FAQ: https://yoursite.com/faq
Upload to your server root. Verify it loads at yourdomain.com/llms.txt. Update it when you add new services or content.
Entity Signals: Building Your Digital Identity
AI search engines build entity profiles by cross-referencing your practice across multiple authoritative sources. Think of it as your practice’s digital passport. Every source where your NAP (name, address, phone) appears consistently is a stamp on that passport. The more stamps, the more AI trusts your entity.
Essential entity-building sources for therapists:
- Google Knowledge Panel — Wikipedia or equivalent structured data that triggers a Knowledge Graph entry. If you have notable credentials, publications, or professional history, a Wikipedia page or Wikidata entry creates the strongest entity signal.
- Google Business Profile — Verify and complete every field. Full hours, services, photos, Q&A, posts. GBP is Google’s primary entity source for local businesses.
- Apple Business Connect — Separate from GBP. Apple Maps and Siri draw from this. Many therapists skip it, which means their practice doesn’t exist in Apple’s entity graph.
- Bing Places — Bing powers parts of ChatGPT Search’s local results. Bing Places listing is mandatory for ChatGPT visibility.
- Professional directories — Psychology Today, GoodTherapy, TherapyDen, Alma, Headway. Each creates a citation that reinforces your entity profile. AI engines trust practices with consistent data across all these sources.
- Healthcare-specific sources — NPI registry, state licensing board, professional association directories (APA, ACA, NASW, AAMFT). These carry outsized weight for healthcare entities.
Robots.txt: The Gatekeeper for AI Crawlers
Your robots.txt must explicitly allow: GPTBot, ChatGPT-User, ClaudeBot, and PerplexityBot. Many WordPress configurations block these crawlers by default or do not address them. If AI engines cannot crawl your site, they cannot cite you — regardless of how well you have optimized your content.
Add these lines to your robots.txt:
User-agent: GPTBot Allow: / User-agent: ChatGPT-User Allow: / User-agent: ClaudeBot Allow: / User-agent: PerplexityBot Allow: / User-agent: * Disallow: /wp-admin/ Disallow: /wp-includes/ Disallow: /feed/ Disallow: /trackback/
Important note: The most permissive crawler rules should appear before the catch-all User-agent: * block. Crawlers match the most specific rule first. If you place a restrictive User-agent: * block before the AI crawler rules, some AI crawlers may still be blocked under certain configurations. Rules at the top of the file take precedence for matching crawlers.
How to Track AI Visibility (It Is Different From Rankings)
You cannot manage what you do not measure. Tracking AI visibility requires different tools and metrics than traditional rank tracking.
Traditional rank tracking: Tools like SEMrush, Ahrefs, and BrightLocal track keyword positions in Google’s 10 blue links. These remain useful but do not tell you whether your content appears in AI-generated answers.
AI visibility tracking: Several platforms now offer GEO-specific tracking. Platform like Perplexity’s built-in sourcing reports, and third-party tools like GEO Tracker and AIOSEO’s AI visibility module, track whether your content is cited in AI responses. These tools automate the testing — they query AI platforms with your target keywords and report whether your site appears as a source.
What to track monthly:
- Number of AI citations across ChatGPT, Perplexity, Claude, and Google AI Overviews
- Which pages are being cited — identify which content drives AI visibility
- Which competitors are appearing in AI results that you are not
- Changes in citation frequency after content updates or new schema implementation
- Correlation between AI citations and website traffic from organic search
Manual spot-checking also works: test your target keywords in Perplexity and ChatGPT-4 with browsing weekly. Note which sources appear and whether your site is included. A spreadsheet tracking 10-20 core queries across platforms tells you more than a vanity dashboard.
Five GEO Mistakes Therapists Make
1. Blocking AI crawlers unintentionally. Most WordPress sites using security plugins (Wordfence, Sucuri, iThemes) block unknown user agents by default. AI crawlers are new — they may be classified as “unknown.” Whitelist them explicitly in your security plugin settings in addition to robots.txt.
2. Writing warm-up paragraphs instead of direct answers. “Anxiety affects millions of Americans and can manifest in various ways including…” — this is a warm-up. The AI needs “Cognitive Behavioral Therapy treats anxiety by restructuring negative thought patterns…” — this is the answer. Lead with the answer. Add context after.
3. Inconsistent NAP across entity sources. Your Google Business Profile says “Suite 200” but your website says “Ste 200.” Your Psychology Today profile omits your middle initial that appears on your license. Every inconsistency reduces AI citation confidence — AI engines detect these mismatches and may discount your entity entirely.
4. Using only traditional SEO schema. Most therapists implement LocalBusiness schema but skip MedicalBusiness, Service, and FAQ schema. AI platforms extract from specialized schema types. LocalBusiness tells AI you are a business. MedicalBusiness tells AI you are a healthcare provider. The difference matters for AI citation confidence.
5. Ignoring non-Google entity sources. Google Business Profile optimization is standard. Apple Business Connect, Bing Places, and professional directory citations are often neglected. ChatGPT Search draws from Bing’s index. Claude validates against professional directories. Focusing only on Google leaves you invisible to half the AI search ecosystem.
The Relationship Between SEO and GEO
Strong SEO is the foundation for GEO. Schema markup helps both Google rankings and AI extraction. Backlinks that build domain authority also signal trustworthiness to AI systems. Content structured for search engines is also content AI can parse. Building for one builds for both. The relationship is complementary, not competitive.
That said, the optimizations differ in emphasis. SEO favors comprehensive content that covers a topic from all angles. GEO favors concise, self-contained answers that can be extracted and cited independently. The ideal approach: write comprehensive SEO content, then ensure the first paragraph of every section can stand alone as an AI-ready answer. You get both goals with one piece of content.
AI Search Queries Therapists Should Target
AI search queries differ from traditional Google queries. Patients phrase questions conversationally. Here are specific query patterns therapists should optimize for:
- Direct comparison queries: “What is the difference between CBT and DBT for anxiety?” “EMDR vs talk therapy for trauma — which is more effective?” — Answer these in FAQ sections and comparison pages.
- “How to find” queries: “How to find a therapist who specializes in OCD in Denver?” “How to find a LGBTQ-affirming therapist in Portland?” — Your about page and specialty pages should include language that answers the how-to dimension.
- Cost and logistics queries: “Does insurance cover couples therapy in New York?” “How much does a therapist cost without insurance in Austin?” — Publish a transparent fees and insurance page with definitive numbers.
- Condition-specific queries: “What is the best therapy for childhood trauma?” “What kind of therapist treats binge eating disorder?” — These map to your specialty pages. Answer the therapy question directly in the first paragraph.
- Credentials queries: “What does a licensed clinical social worker do vs a psychologist?” “What credentials should a trauma therapist have?” — Build a credentials/qualifications page that answers these comparison questions.
To find these queries for your specific location and specialty, search in Perplexity with “therapist [your city] [your specialty]” and note the follow-up questions Perplexity suggests. Those are the exact queries patients are asking AI.
Where Private Practices Have an Advantage
Online therapy platforms like BetterHelp and Talkspace compete aggressively for generic terms. They cannot credibly target narrow specialty and location combinations. “Trauma therapist for veterans in Austin, TX” is a search only a local practice can authentically answer. AI search rewards specificity. A solo practice with deep expertise in a narrow niche can outrank a platform with millions in marketing spend. The strategy is the same as traditional SEO: narrow the target, own the niche.
AI search actually amplifies this advantage. When Perplexity or ChatGPT answers “What is the best therapy for veterans with PTSD in Austin?”, they need a specific, local, credentialed provider answer. A national platform’s generic content does not satisfy that query. Your content — with MedicalBusiness schema, local entity signals, and directly relevant expertise — perfectly matches the AI’s retrieval criteria. This is the GEO opportunity most therapists have not yet seized.
Your 90-Day GEO Action Plan
Week 1-2: Audit your current AI visibility. Test 10 target queries in Perplexity, ChatGPT with browsing, and Google (with AI Overviews). Record whether your site appears. Identify which competitors appear that you do not.
Week 3-4: Implement schema markup — MedicalBusiness on homepage, Person with license on about page, Service on each specialty page, FAQ on service pages. Validate every schema with Google’s Rich Results Test. Add llms.txt to your site root.
Week 5-6: Update robots.txt to explicitly allow all AI crawlers. Verify AI crawler access by checking server logs for GPTBot, ClaudeBot, and PerplexityBot user agents. If none appear after two weeks, investigate firewall or security plugin blocks.
Week 7-8: Rewrite the first paragraph of every key page to function as a standalone answer. Cut warm-up text. Lead with the definitive response to the page’s primary question. Add 5-10 FAQ items with FAQ schema per service page.
Week 9-10: Build or update your entity presence. Claim Apple Business Connect. Verify Bing Places. Ensure NAP consistency across Psychology Today, GoodTherapy, and professional directories.
Week 11-12: Retest the same 10 queries. Compare appearance rates before and after optimization. If your site still does not appear, investigate whether the pages are properly indexed, whether crawlers are accessing them, and whether your schema validates without errors. Iterate from there.
Start with schema and robots.txt — they cost nothing and deliver immediate AI visibility improvements. Build out from there.
