How AI Tools Decide Which Therapists to Recommend

An illustrated walkthrough of how ChatGPT, Claude, Perplexity, and Google's AI Overview respond when a potential client asks them to recommend a therapist — and why the four platforms rarely agree.

Looking for the Google Map Pack walkthrough?

This page covers how ChatGPT, Claude, Perplexity, and Google's AI Overview recommend therapists. If you're looking for how Google's Map Pack works — the traditional local-search path that still drives most therapy inquiries — that's covered on a separate page.

View the Google walkthrough →

See whether AI tools recommend your practice

Enter your name and specialty. We'll run a real query against one of the major AI platforms and show you whether it recommends your practice — along with the specific signals the AI used to decide.

No email required, no information stored.

Run the AI Visibility Test →

What fragmentation looks like in practice

We asked ChatGPT, Claude, and Perplexity to recommend an anxiety therapist in Tacoma, Washington. Here are the actual responses from our April 2026 study — three platforms, three largely different recommendations:

← The same query asked to three different AI platforms produced three largely different therapist recommendations. Only a small overlap across platforms — in our full 80-query study, fewer than 15% of named therapists appeared on more than one platform.

Your city will look different. These aren't universal rankings — they're a snapshot of one market, one specialty, one point in time. What stays consistent is the pattern: four AI platforms rarely recommend the same therapists.

To see what AI tools return for your practice specifically, run the AI Visibility Test.

From our 80-query AI visibility study, Tacoma, Washington — April 2026. Real responses from ChatGPT, Claude, and Perplexity. Results in your city will differ. See the full research in our AI search and therapist discovery guide.

Four things AI platforms actually look at

Structured content

AI platforms can't read an image or interpret a vague "Services" page. They need content structured around specific specialties, specific locations, and specific questions. A page titled "Anxiety Therapy in Tacoma" with an FAQ answering "How do I know if I need therapy for anxiety?" is something an AI can cite. A generic services list isn't.

Citation consistency

AI platforms cross-reference multiple sources before recommending. Your name, address, phone number, and specialties should match across your website, Google Business Profile, Psychology Today, and other directory listings. Inconsistencies reduce AI confidence; consistency compounds it.

Review signals

Every recommended therapist in our testing had Google reviews. This isn't a coincidence — review count and average rating are among the few external trust signals AI tools can access. A therapist with 20 genuine reviews and a 4.8-star average is verifiably more trusted than one with zero reviews, regardless of actual clinical quality.

Schema markup

Schema is code that explicitly tells AI platforms what your content is about. LocalBusiness schema names your practice, location, and services. FAQPage schema labels Q&A content. MedicalBusiness schema identifies you as a healthcare provider. When schema is present, AI tools extract information with higher confidence than from unstructured text. It's the difference between handing someone a labeled filing cabinet and handing them a stack of loose papers.

Want the full research behind this?

We ran an 80-query study across ChatGPT, Claude, Perplexity, and Google's AI Overview — asking each to recommend therapists across five specialties in a mid-size metro. The results cover which therapists get recommended, why, and what the invisible therapists are missing.

Read the full analysis →

See where your practice stands — and whether your area is open

Two free, honest checks: how visible you are right now, and whether we still have capacity in your geography and specialty.