How Can a Medical Practice Track Whether AI Assistants Recommend It?

A medical practice can track AI recommendations in two ways. First, run the same set of patient questions through ChatGPT, Gemini, Copilot and Google's AI Mode every month, logged out and more than once each, and record whether you are named and which sources are cited. Second, read the first-party reports: Bing Webmaster Tools AI Performance, Search Console's generative AI report and GA4's AI Assistant channel.
Neither gives a complete picture on its own. Prompt tests show what patients may see but vary from run to run; platform reports are consistent but only cover their own tools. Used together, they show the trend. This guide sets out the routine we suggest, which sits alongside our healthcare AI search optimisation work and our guide to getting your clinic recommended by ChatGPT.
Why is tracking AI recommendations difficult?
- Answers vary. The same question can produce different clinics on different runs, for different people and in different places.
- Personalisation. Signed-in accounts can draw on memory and past chats, which skews results for anyone who has asked about their own practice before.
- Location. Questions with 'near me' depend on where the assistant thinks the user is.
- No universal report. Google and Microsoft now report some AI data to site owners, but there is no ChatGPT equivalent of Search Console.
So the aim is not a precise score. It is a consistent, repeatable method that shows whether you are named more or less often, how accurately you are described and which sources assistants rely on.
How do you run a monthly AI prompt test?
1. Choose 15 to 25 prompts
Write the questions patients actually ask, in their words, and mix the types:
- Recommendation: 'Which private dermatologist in Bristol would you recommend for eczema?'
- Choosing: 'What should I look for in a private knee surgeon in Surrey?'
- Treatment questions: 'How much does private cataract surgery cost in London?'
- Brand checks: 'What do you know about [practice name]?' and 'Is [practice name] a good clinic?'
Take wording from enquiry emails, phone calls and the queries in Search Console. Keep the core list fixed so month-on-month comparisons are fair, and add any new prompts as a separate group.
2. Run each prompt logged out, more than once
- Use a private or incognito window and stay logged out where the tool allows it, or use a clean account with memory and chat history turned off.
- Run each prompt three times and record all three answers. A practice named in one run out of three is in a very different position from one named every time.
- Test from the same location each month and note it. Where you can, put the town in the prompt rather than relying on 'near me'.
- Cover ChatGPT, Gemini, Microsoft Copilot, Perplexity and Google's AI Mode, and note any AI Overview shown in ordinary Google results.
3. Log mentions and sources
Use a spreadsheet with one row per prompt, per tool, per run. These columns are enough:
- Date, tool and mode, for example ChatGPT with search or Google AI Mode.
- Prompt and location.
- Named: yes or no, and position in the answer.
- How you were described, and whether anything was wrong or out of date.
- Competitors named.
- Sources cited, with URLs, and whether your own site was one of them.
The sources column is often the most useful. If assistants keep citing a directory profile with your old address, or a competitor's comparison page, you know exactly what to fix or create. Our post on why a smaller rival gets recommended by AI tools explains how to read a competitor's footprint.
4. Summarise the month
Work out your mention rate: the share of runs in which you were named. Track it per tool and per prompt group, list every inaccuracy you found and what you did about it, and note any source that keeps appearing. Three numbers and a short list are more useful to a practice owner than fifty screenshots.
What does Bing Webmaster Tools AI Performance show?
Microsoft launched AI Performance in Bing Webmaster Tools as a public preview on 10 February 2026. It shows how often your content is cited as a source across Microsoft Copilot, AI-generated summaries in Bing and selected partner integrations.
- Total citations: how many times your pages were shown as a source in AI answers.
- Average cited pages: the daily average of unique pages from your site that were cited.
- Grounding queries: a sample of the phrases the AI used when retrieving your content.
- Page-level activity: which URLs were cited, and how that changes over time.
It is the closest thing to first-party citation data currently available, but it covers Microsoft's surfaces only and Microsoft's announcement does not mention clicks. If you have not used Bing Webmaster Tools before, you can import your site from Search Console rather than verifying it from scratch.
What does Search Console's generative AI report show?
Google introduced Search Generative AI performance reports in Search Console in June 2026 and says they reached all websites worldwide by 31 August 2026. They show how often URLs from your site appeared in AI Overviews, AI Mode and AI features in Discover, broken down by page, country, device and date.
The reports count impressions. Google's announcement does not mention clicks or queries, so use them to see which pages Google's AI features draw on and whether that is growing. Clicks from those features are still counted within the Web search type in the main Performance report, as Google's AI features documentation explains. Our guide to appearing in Google AI Overviews and AI Mode covers Google's side in more depth.
How do you track AI referrals in GA4?
GA4's default channel group now includes an AI Assistant channel for visits from chatbots such as ChatGPT, Gemini and Deepseek. GA4 sets the medium to 'ai-assistant' when the referrer is on its list of known AI assistants. To use it:
- Open Reports, then Acquisition, then Traffic acquisition, and look for the AI Assistant row.
- Add landing page as a secondary dimension to see which pages AI users arrive on.
- Check key events for that channel, such as enquiry form submissions, phone clicks and booking clicks.
- Remember that visits from Google's AI Overviews and AI Mode are counted as Organic Search, not AI Assistant.
- Expect some undercounting: visits from AI apps can arrive without a referrer and show up as Direct.
If your site uses consent mode, visitors who decline analytics will be modelled or missing too. Our guide to GA4 and consent mode for clinics explains the gaps, and our guide to whether answer engine optimisation brings in patient bookings covers what this data can and cannot prove.
Are paid AI visibility tools worth it?
Several SEO platforms now track prompts at scale, for example Semrush's AI visibility reports and Ahrefs Brand Radar. They run large prompt sets on a schedule and report mentions, citations and share of voice against competitors. The AI figures in our published case studies come from the Semrush AI Visibility Overview: Dr Carla Perna with 162 AI citations in 9 months and Gentle Procedures UK with 67 AI citations in 9 months.
These tools suit larger clinics, groups and anyone tracking many prompts, locations or competitors. For a single practice, the manual monthly test plus the free platform reports is usually enough to start. Before paying, check which assistants and countries a tool covers, how many prompts it runs, whether it repeats them, and whether you can export the raw answers and sources. Plans and prices change, so compare current pricing on each vendor's own site.
What should a monthly AI visibility report include?
- Mention rate by tool, compared with last month.
- Brand prompt results, and whether the description of your practice is accurate.
- Inaccuracies found and fixed.
- The sources cited most often, yours and other sites'.
- Bing AI citations and Search Console generative AI impressions.
- AI Assistant sessions and key events from GA4.
- Actions for next month.
Keep it to a page. If you also run SEO, fold it into your normal SEO performance reporting rather than creating a separate report.
What this means for your practice
You cannot see every AI answer about your practice, but you can track enough to act on. A fixed set of prompts, run monthly, logged out and repeated, tells you whether and how you are named. Bing's AI Performance report, Search Console's generative AI report and GA4's AI Assistant channel add consistent first-party data. Together they show the trend and, more usefully, what to fix.
Pulse Digital Health is a London healthcare marketing agency working with private clinics, consultants and healthcare groups. If you would like a snapshot first, our free AI visibility check runs an initial round of prompts and lists what we would fix.
Frequently asked questions
How do I know if ChatGPT recommends my clinic?
Ask it the questions your patients ask, logged out and several times each, and record whether your clinic is named and which sources are cited. Repeat the same prompts monthly.
ChatGPT does not give site owners a report of mentions, so a consistent manual test or a paid prompt-tracking tool is the only direct way to see it.
Why do I get different answers each time I ask?
AI assistants generate each answer afresh, and results change with the wording, location, account history and updates to the tools. That is why each prompt should be run more than once and logged. The share of runs in which you are named is more meaningful than any single answer.
Can Google Search Console show AI Overviews data?
Yes. The Search Generative AI performance reports show impressions in AI Overviews, AI Mode and Discover AI features by page, country, device and date. Clicks from AI features are included within the Web search type in the main Performance report.
Does GA4 track traffic from ChatGPT?
Partly. GA4's default channel group includes an AI Assistant channel for visits from known AI assistants such as ChatGPT and Gemini. Some app visits arrive without a referrer and are counted as Direct, and visits from Google's AI Overviews and AI Mode are counted as Organic Search.
What is Bing Webmaster Tools AI Performance?
A report, launched in public preview in February 2026, that shows how often your pages are cited in Microsoft Copilot, Bing's AI-generated summaries and selected partner integrations. It reports total citations, cited pages and a sample of grounding queries. Microsoft's announcement does not mention click data.
Do I need a paid tool to track AI visibility?
Not to start. A monthly manual test of 15 to 25 prompts, plus Bing Webmaster Tools, Search Console and GA4, is enough for most single practices. Paid tools help when you need to track many prompts, locations or competitors at once. Compare coverage and current pricing before you commit.
