Therapist AI Visibility Audit: 12 Prompts to Test Whether ChatGPT, Perplexity, and Google AI Can Find You

88% of local businesses are completely absent from ChatGPT recommendations, according to Omni Eclipse's 2026 audit of 1,700 businesses across 32 industries and 3 countries. A parallel Surfeo study of 518 businesses across 20 countries found 91% score below 50 out of 100 on AI visibility — with the average sitting at just 38. Only 4 of those 518 businesses scored above 80 (Surfeo, 2026).

Therapy practices are not an exception to those numbers. They're typical of them. Most therapists have read about AI discoverability, suspected they might be invisible, and assumed the situation would sort itself out as AI matures. It doesn't. AI engines don't default to discovering you — they cite what's already indexed, structured, and cross-verified against multiple data sources. If that groundwork isn't laid, no client finds you through AI search.

The good news: finding out exactly where you stand takes 20 minutes and costs nothing. You need a ChatGPT account, a Perplexity account (free tier works), and access to Google. Run these 12 prompts. Record what comes back. By the end, you'll have a concrete picture of your AI visibility — and a prioritized list of what to fix.

An AI visibility audit tests three things, in order of dependency:

1. Entity recognition — Does AI know your practice exists as a distinct entity with accurate information?
2. Local discoverability — Does AI surface your practice when someone searches for therapy services in your area?
3. Recommendation strength — When AI is asked to recommend a therapist for your specialty in your location, do you appear?

These three layers are not the same. A practice can have entity recognition (AI knows it exists) but weak local discoverability (AI doesn't recommend it). A practice can show up in discoverability results but not be named when someone asks for help with a specific issue. Each failure has a different fix.

You'll test on multiple platforms because their data sources differ significantly. ChatGPT's local results pull primarily from Foursquare's Places API and Bing's search index. Perplexity runs a real-time web search and shows its citations visibly. Google AI Mode draws from its own index and Google Business Profile. A practice that's visible in one may be entirely absent from another.

Before you start: Open a spreadsheet or document with 12 rows (one per prompt) and 3 columns for ChatGPT, Perplexity, and Google. In each cell, record: Mentioned with accurate info, Mentioned with inaccurate info, or Absent.

Run these on ChatGPT. They test whether AI has indexed your basic practice information as a named entity — a prerequisite for being recommended in any other context.

Prompt 1: Basic identity
"Who is [your full name]? Are they a therapist?"

What to look for: Does AI recognize you as a distinct person? Does it have your correct credential, specialty, and location? If it confuses you with someone else, reports no information, or returns a confident but wrong answer (wrong city, wrong credential), your web presence lacks the E-E-A-T signals AI uses to build entity knowledge — specifically, your name, license type, and state aren't visible in structured form anywhere AI has indexed.

Prompt 2: Practice entity
"What is [your practice name]? What services do they offer?"

What to look for: Does AI know your practice as a distinct business with a correct address, phone number, and service list? Inaccurate information is sometimes worse than no information — AI presenting a wrong address or disconnected phone number actively misleads potential clients and signals data inconsistency across your listings.

Prompt 3: Data accuracy
"Does [your practice name] accept [a major insurance you actually take]?"

What to look for: Can AI answer a transactional question about your practice accurately? Incorrect answers here reveal that your insurance information isn't structured in a way AI can parse — either it's buried in paragraph text rather than marked up in schema, or it's absent from your listings entirely.

Run these on both ChatGPT and Perplexity. They test whether AI recommends you when someone searches for therapy in your area — which is how real prospective clients search.

Prompt 4: Core local search
"Find me a [your primary specialty] therapist in [your city]."

What to look for: Do you appear in the results? If not, name the practices that do. They're your benchmark for what AI-visible looks like in your market — and examining their websites will show you exactly what you're missing.

Prompt 5: Specialty + insurance filter
"I need a therapist in [your city] who specializes in [specialty] and accepts [insurance]. Who would you recommend?"

What to look for: Filtered queries like this surface fewer results and weight data accuracy more heavily. Practices with clear, structured insurance and specialty information consistently outperform here. If you appear on Prompt 4 but disappear on Prompt 5, your insurance data is the gap.

Prompt 6: Problem-first search
"I'm dealing with [anxiety / relationship issues / childhood trauma] and I'm in [your city]. Can you help me find a therapist?"

What to look for: Real clients search this way — describing their problem rather than naming a modality. AI handles these queries by matching the described problem to specialty pages and service descriptions on therapist websites. If your specialty page describes what you treat in plain language (not clinical jargon), you have an advantage here. If your services page is a list of modality acronyms without explanation, you don't.

Prompt 7: Telehealth filter
"Who offers online therapy in [your state] that specializes in [specialty]?"

What to look for: If you offer telehealth, your service descriptions and schema markup should say so explicitly. AI frequently fails to surface telehealth providers because the telehealth offering appears once in a FAQ rather than in structured schema. Run this prompt even if you're primarily in-person — if you do any telehealth, this reveals whether AI knows about it.

These prompts ask AI to make an explicit recommendation — the strongest test of whether AI trusts your practice enough to name you by name. Run on ChatGPT.

Prompt 8: Direct recommendation
"If someone in [your city] was looking for a therapist who specializes in [your primary specialty], who would you specifically recommend and why?"

What to look for: AI that recommends you by name and states a reason (your credential, specialty match, reviews) is the gold standard. AI that hedges ("I can't recommend a specific provider") or redirects to Psychology Today doesn't have enough indexed information about you to generate a confident recommendation. Our guide to how ChatGPT recommends therapists explains the underlying selection mechanism.

Prompt 9: Comparison
"Compare therapists in [your city] who offer [EMDR / CBT / couples therapy]. What are the differences between their approaches?"

What to look for: Does your practice appear in the comparison? Comparative queries reward structured content — comparison tables, question headings, and answer-ready first paragraphs. If you offer a specialty but aren't mentioned, your specialty page likely lacks the structured content AI pulls for comparative answers.

Prompt 10: Review-weighted recommendation
"What therapist in [your city] has strong reviews and also takes [insurance]?"

What to look for: Google Reviews function as a trust signal in AI recommendations. If you have more reviews than the practices AI names but you're still absent, that's a structured data problem — your reviews aren't being surfaced in a way AI can read. AggregateRating schema on your site connects your review count to your practice entity for AI engines.

Prompt 11: Perplexity with citations
On Perplexity, search: "therapist in [your city] specializing in [specialty] 2026"

What to look for: Unlike ChatGPT, Perplexity shows its source citations in a sidebar panel. If you appear in the answer, look at which domain it cited — probably your website, your Psychology Today profile, or a directory. If a competitor appears and you don't, look at what Perplexity cited for them. That domain is what's missing from your own presence. If Psychology Today is the only citation for therapists in your city, that's a signal that practice websites in your market aren't authoritative enough to cite directly.

Prompt 12: Google AI Mode
On Google, search: "therapist [specialty] [your city]"

What to look for: Does Google AI Mode produce a conversational answer with named providers? If it does and you're absent, the most common culprit is an incomplete Google Business Profile or missing MedicalBusiness schema on your website. Google AI Mode draws heavily from its own index and GBP — unlike ChatGPT, it doesn't rely on Foursquare. For a detailed breakdown of how this platform works, see our Google AI Mode guide for therapists.

Score each prompt: 2 points (appeared with accurate information), 1 point (appeared with inaccurate or incomplete information), 0 points (absent). Maximum: 24 points.

Score	AI Visibility Level	What It Means
20–24	Excellent	You're in the recommendation pool across platforms. Focus on maintaining accuracy and recency.
14–19	Moderate	AI knows you exist but doesn't consistently recommend you. Structured data and content gaps.
8–13	Weak	Visible in some contexts, absent in most. High-impact fixes are available.
0–7	Invisible	AI doesn't know you exist. Start with listings and schema markup — everything is upside from here.

Note where you scored 0 versus 1: partial credit usually means data exists but is inaccurate. That's a different fix from absence — you need to correct existing information rather than create new signals.

Entity recognition failures (Prompts 1–3): Your web presence lacks structured E-E-A-T signals. Add Person schema with hasCredential to your About page. Verify your Foursquare and Bing Places listings exist and have accurate NAP (name, address, phone). Publish your license type, number, and state on your website in plain text — not buried in a PDF or an image. AI uses these signals to confirm you're a licensed clinician and to distinguish you from the person with your name who writes software in another state.

Local discoverability failures (Prompts 4–7): You're either not in Foursquare (ChatGPT's primary local data source), not in Bing's search index (ChatGPT Search), or your specialty pages lack structured content. Claiming your Foursquare listing is the single highest-impact fix for ChatGPT local visibility. Submitting your sitemap to Bing Webmaster Tools addresses the ChatGPT Search gap. Our complete AI discoverability guide covers the full listings and technical checklist.

Recommendation failures (Prompts 8–10): AI knows you exist but doesn't trust you enough to name you specifically. This is a content authority gap. Your specialty pages need question-format headings, FAQ sections with FAQPage schema, and cited statistics from recognized sources. AI recommends content it can extract clean, direct answers from — three paragraphs of clinical narrative aren't it.

Cross-platform gaps (Prompts 11–12): Each platform uses different data sources. A Perplexity gap means your website content isn't authoritative enough to rank as a primary citation — look at what Perplexity cites for competitors. A Google AI Mode gap typically means your Google Business Profile or website schema is incomplete. Fix each platform's gap separately; they don't share data sources.

AI search is not a set-it-and-forget-it system. Your visibility changes as AI models update, as competitors optimize, and as your practice information changes — a new location, a new insurance panel, a new specialty. Running this 12-prompt audit monthly takes 20 minutes and tells you whether your changes are having an effect and whether any drift has crept in.

Keep a simple log: date, total score, and the most notable absence or inaccuracy from that run. Three months of data tells you more than any single snapshot. A practice that goes from invisible to appearing in 6 of 12 prompts — and is tracking it — knows what's working. A practice that assumes things are fine until a client says "I couldn't find you anywhere" doesn't.

The visibility gap between AI-optimized and unoptimized therapy practices compounds over time. AI models weight data that's been consistently accurate across multiple sources over months. Practices that started building these signals in 2025 and 2026 have a durable advantage over those who start in 2027 — not because AI rewards seniority, but because cross-platform NAP consistency and accumulated reviews take time to build.

Want your practice to handle this automatically? WebsiteTherapy manages the underlying infrastructure that makes this audit score well: 25+ schema types auto-generated on every page, Foursquare and Bing monitoring, structured specialty pages with question headings and FAQ sections, and a Google Reviews collection workflow. See what's included.

References

• Omni Eclipse. (2026). The 2026 AI Search Visibility Report: 88% of Businesses Are Invisible in ChatGPT. https://omnieclipse.ai/blog/ai-search-visibility-report-2026
• Surfeo. (2026). AI Visibility Report 2026 — 518 Businesses Audited Across 4 AIs. https://surfeo.ai/en/blog/ai-visibility-report-2026