How to Audit Your Clinic's AI Search Visibility: A Step-by-Step Checklist

Why standard SEO tools miss AI visibility entirely

According to BrightEdge Generative AI Research (2024), 84% of health-related queries in Google now return an AI Overview block. Yet most clinic marketing teams have no idea whether their practice appears in those answers — because the tools they rely on were never built to measure it.

Google Search Console shows organic clicks and impressions. Rank trackers show positions in the ten blue links. Neither tells you whether ChatGPT, Perplexity, or Google's own AI Overview cites your clinic when a patient asks "which cardiologist in [city] should I see?"

The gap is real and measurable. A clinic can hold a #3 organic ranking for a high-intent keyword and still be completely absent from every AI-generated answer on that topic. That absence is invisible in any standard SEO dashboard. This is the core problem that an AI visibility audit is designed to solve.

Since May 2024, Google Search Console has included a dedicated AI Overviews filter under Search type, showing impressions and clicks attributed specifically to AI Overview appearances. Most clinic accounts have never opened it. When they do, the numbers are often startlingly low — not because the clinic has bad content, but because AI systems cannot find, parse, or trust the structured data they need to generate a citation.

The 4 things an AI visibility audit measures

An AI visibility audit is not a technical SEO crawl with a different name. It measures four specific things that determine whether an AI system can find your clinic, understand who you are, and include you in a generated answer.

1. Citation frequency

Does your clinic's name actually appear in AI-generated answers? Citation frequency is the most direct measure of AI visibility. A clinic with strong citation frequency shows up when patients ask about relevant specialties, symptoms, or locations — regardless of whether they use your clinic's name in the query.

2. Named entity recognition

Named entity recognition (NER) determines whether AI systems correctly identify your clinic as a distinct entity — not just a web page. AI models that use RAG (Retrieval-Augmented Generation) pull information from their training data and live web sources. If your clinic name, address, specialties, and physicians are not consistently structured and cross-referenced, the AI may ignore the entity entirely or conflate it with another practice.

3. Schema completeness

Schema markup is the structured vocabulary that tells AI crawlers what type of entity a page represents. Clinics without MedicalOrganization, Physician, or FAQPage schema give AI systems nothing formal to work with. Data from Clingeo's benchmark of 334 medical site audits shows that fewer than 15% of clinic sites have complete Physician Schema on their doctor profile pages.

4. Crawl access

AI systems send their own bots: GPTBot (OpenAI), ClaudeBot (Anthropic), and PerplexityBot. Many clinic websites accidentally block these crawlers through overly broad robots.txt rules written years ago for spam bots. If the AI cannot read your pages, it cannot cite you — full stop.

Step 1 — Manual testing baseline: 15 queries across 3 platforms

The fastest way to get a real signal on your current AI visibility is to run a structured set of manual queries. This takes 60–90 minutes and gives you a documented baseline before you touch a single file on your website.

Run a minimum of 15–20 representative queries across at least three AI platforms: ChatGPT, Perplexity, and Google AI Overview (via Google Search). Spread those queries across four categories:

• Branded — "[Clinic name] reviews", "is [clinic name] good for [specialty]"
• Specialty-based — "best endocrinologist in [city]", "where to get a skin check in [city]"
• Local — "private GP near [neighbourhood]", "walk-in clinic [postcode area]"
• Symptom-based — "doctor for persistent back pain [city]", "anxiety therapy clinic [city]"

Record results in a spreadsheet with these columns: Query | Platform | Result type | Clinic mentioned? | Citation type | Notes. For each result, classify the citation as one of three levels: full citation (clinic named with link or specific details), partial mention (name appears but without context or accuracy issues), or absent (clinic not referenced at all).

At the end of this step, calculate your raw citation rate: number of queries where you appeared divided by total queries run, across each platform separately. This number is your starting AIV score proxy — and for most clinics running this for the first time, it is lower than expected.

Step 2 — Technical access audit: robots.txt and llms.txt

AI crawlers cannot cite pages they cannot read. Before investing time in content changes, confirm that the AI bots your clinic wants to reach are actually allowed on your site.

Open your robots.txt file at yourdomain.com/robots.txt. Check for rules referencing GPTBot, ClaudeBot, PerplexityBot, and Google-Extended. A Disallow: / rule under any of these user agents completely blocks that AI system from reading your content.

Also check for blanket wildcard blocks — rules like User-agent: * followed by broad disallow paths that may inadvertently cover key service pages. This is one of the most common mistakes in clinic websites built on older CMS templates.

Next, check for an llms.txt file at yourdomain.com/llms.txt. This is a newer standard (analogous to robots.txt but designed specifically for large language models) that helps AI systems understand your site structure without crawling every page. The Clingeo benchmark found that fewer than 8% of medical websites have a correctly formatted llms.txt. If yours is missing, that is a quick, high-impact fix — you can learn how to set it up as part of a broader technical GEO audit.

Step 3 — Schema coverage scan

Schema markup is the primary signal AI systems use to understand what an entity is and what it does. Without it, AI models rely on free-text parsing — which is slower, less reliable, and more likely to produce errors or omissions in generated answers.

Use the Google Rich Results Test (search.google.com/test/rich-results) and the Schema Markup Validator (validator.schema.org) to scan your priority pages. Start with three page types in this order:

• Homepage — should include MedicalOrganization with name, address, telephone, and opening hours
• Physician profile pages — should include Physician (or MedicalSpecialty) with credentials, specialty, and affiliation
• Service pages — should include MedicalProcedure or HealthAndBeautyBusiness as appropriate, plus FAQPage where a FAQ block exists

The minimum viable schema set for any clinic is: MedicalOrganization on the homepage, Physician on each doctor's page, and FAQPage on any page containing a question-and-answer section. If those three types are missing or malformed, fixing them alone will meaningfully improve how AI systems represent your clinic in generated answers.

Step 4 — Google Search Console AI Overview check

Google Search Console has included AI Overview attribution data since May 2024. Most clinic accounts have either missed this or not understood what it shows.

To find it: in GSC, go to Search Results under Performance. Click the "Search type" filter at the top and select AI Overviews. The report shows impressions and clicks attributed specifically to appearances within the AI Overview block — separate from standard organic results.

What to look for: if you have consistent organic rankings but near-zero AI Overview impressions on the same queries, that is a strong signal that your content is not being selected for AI synthesis. Common causes include missing structured data, thin paragraph structure that does not answer questions directly, and content that lacks the authoritative signals (author credentials, citations, date freshness) that AI systems use when selecting sources.

Cross-reference this data with your manual testing results. Pages with high organic impressions but zero AI Overview impressions are your highest-priority optimisation targets. Understanding what content formats that AI cites helps you prioritise which of those pages to rewrite first.

The audit checklist at a glance

What to check	Tool	Pass criteria
Citation frequency (manual)	ChatGPT, Perplexity, Google AI Overview	Clinic cited in ≥30% of relevant queries per platform
AI bot crawl access	robots.txt (manual inspection)	GPTBot, ClaudeBot, PerplexityBot not blocked
llms.txt file present and valid	Browser check at /llms.txt	File exists, correctly formatted, includes key page list
Schema markup coverage	Google Rich Results Test, Schema Markup Validator	MedicalOrganization + Physician + FAQPage present, no errors
Google AI Overview impressions	Google Search Console (AI Overviews filter)	Impressions present on target queries; not zero across all

Automating the audit with dedicated tools

Manual testing gives you a baseline. It does not scale. Running 20 queries across three platforms every month is feasible; running 100 queries across five platforms weekly is not — at least not manually.

Dedicated AI visibility tools automate the measurement layer. Semrush AI Narratives (launched 2024/2025) monitors brand mentions in ChatGPT and Perplexity responses at scale. BrightEdge tracks AI Overview visibility across large keyword sets. These are enterprise-grade tools built for agency workflows.

Clingeo is purpose-built for medical practices. It calculates an AIV Score — a citation frequency index tracked across ChatGPT, Perplexity, Gemini, and Google SGE — using a set of target queries defined for each clinic's specialty and location. The score is benchmarked against 334+ clinic audits, so you can see not just your raw number but where you stand relative to comparable practices. Clingeo also surfaces schema gaps, robots.txt/llms.txt issues, and content recommendations in a single report.

Use manual spot checks alongside any automated tool. AI platforms update their models and retrieval logic frequently; a tool may lag behind a model change that affects your citation frequency. Running a 15-query manual check quarterly confirms that automated scores reflect real-world AI outputs.

For a deeper dive into the technical side of making your site readable by AI systems, the technical GEO audit guide covers robots.txt configuration, llms.txt structure, and schema implementation in detail.

Ready to see your clinic's current AIV Score? Run a free audit at Clingeo.

FAQ

How long does an AI visibility audit take?

A manual baseline audit — 15–20 queries across three AI platforms plus a robots.txt and schema check — takes 60–90 minutes. An automated tool like Clingeo generates a full report in under 30 minutes without manual query testing.

My clinic ranks well on Google. Does that mean I'm visible in AI search?

No. Organic rankings and AI citation frequency are measured separately and often do not correlate. A clinic ranked #1 organically for a keyword can have zero AI Overview impressions for that same query if its content structure, schema, or crawl access does not meet what AI systems need to synthesise an answer.

What is an AIV Score?

The AIV Score (AI Visibility Score) is a metric developed by Clingeo that measures how frequently a clinic is cited across ChatGPT, Perplexity, Gemini, and Google SGE for a defined set of target queries. It is expressed as a percentage and benchmarked against the Clingeo dataset of 334+ medical practice audits.

Which AI platforms should I test in a manual audit?

At minimum: ChatGPT (GPT-4o), Perplexity, and Google AI Overview. These three have the highest share of health-related AI queries and use different retrieval mechanisms, so testing all three gives a more complete picture than any single platform alone.

How often should I repeat the audit?

Run a full audit quarterly. If you make significant changes to your website content, schema, or robots.txt, run a spot check — 15 queries across two platforms — within two weeks of the change to verify the update had the expected effect.

What is generative engine optimisation (GEO)?

Generative engine optimisation (GEO) is the practice of structuring a website's content, schema, and technical setup to increase citation frequency in AI-generated answers — the equivalent of SEO, but for AI search rather than traditional ranked results. An AI visibility audit is the first step in any GEO programme.