To measure whether AI assistants mention your practice, run a fixed set of patient-style questions through the assistants you care about on a regular schedule, and record what each answer says about you, including whether the facts are right. Then connect that to what your analytics and Business Profile data can show about visits and booking requests. Keep mentions, accurate mentions, visits, and new patients as separate numbers.
Practices have a measurement problem many businesses do not. An answer that names the practice but gives the wrong hours, a provider who has left, or an insurance plan the practice no longer accepts can send a patient the wrong way. And the tools that track website visitors can raise privacy questions on pages where patients book or log in.
This is a measurement article for practice owners and marketing leads, not legal advice. Our general guide to measuring AI search: mentions, citations, visits, and leads covers the definitions and reporting basics; this page applies them to a practice.
Build a question set from what patients ask
Start with the questions patients ask before booking, in their own words. Group them the way patients search: by concern and place ("physical therapist for a knee injury near Raleigh"), by access ("dentist taking new patients on Saturdays"), by coverage ("chiropractor that takes my plan"), and by name, covering the practice and each provider. Our guide to what patients ask before booking shows how to collect them without patient details.
Keep branded and unbranded questions in separate groups. An assistant describing your practice correctly when someone asks for it by name is not the same result as an assistant suggesting your practice when someone asks for help nearby. Mixing the two makes a report look better than the situation is.
Record accuracy, not just presence
Before the first round, prepare a current fact sheet: hours, locations, providers, services, plans accepted, and whether each provider is taking new patients. The person running the checks should not have to invent what counts as correct; the practice supplies it, and the fact sheet carries a date.
For each saved answer, record the question, date, assistant, mode, location context, whether the practice is named, and what the answer says about it. Then compare every stated fact with the fact sheet.
| Observation | What to record | Why it matters for a practice |
|---|---|---|
| Named | The answer mentions the practice or a provider | Presence only |
| Accurate | Every stated fact matches the current fact sheet | A wrong fact can send a patient to the wrong office or plan |
| Recommended | The answer proposes the practice for the stated need | Different from a passing mention |
| Cited | The source the answer names or links | Shows which page or profile carried the fact |
| Inaccurate | The wrong fact, quoted, and any linked source | Points to a page or listing to check |
An inaccurate answer may trace back to something the practice published and never updated, such as an old location page or a bio left up after a provider departed. Treat that as a hypothesis to check, since an assistant can draw on other sources too.
What Google and your analytics can show
Google Search Console's Generative AI performance report "shows data about how your site performs in generative AI features on Google Search," including impressions for AI Overviews and AI Mode. Google says a missing report "usually happens when your site hasn't received enough impressions in generative AI features on Google Search."
In Google Analytics 4, "AI Assistant is the channel by which users arrive at your site from sources like ChatGPT, Gemini, Deepseek, Copilot, or Grok," and the channel "excludes Google's AI Overviews and AI Mode," which are counted under Organic Search. OpenAI says "ChatGPT automatically includes the UTM parameter utm_source=chatgpt.com in referral URLs." Check what your own analytics actually receives before building a report on it.
Your Business Profile adds practice-level signals, as covered in what a practice's Google Business Profile should say. Its performance view counts calls, website clicks, and direction requests, and for bookings Google says "your bookings must be managed through a booking provider." If the practice books through a scheduling platform on another domain, confirm how that handoff is measured before presenting a visit and a booking as one path.
Check privacy before adding tracking
Measurement on a practice website has a HIPAA side. HHS guidance describes a tracking technology as "a script or code on a website or mobile app used to gather information about users or their actions." It says "Tracking technologies on a regulated entity's user-authenticated webpages generally have access to PHI," while tracking on many unauthenticated pages does not, and it treats tracking vendors that meet the business associate definition as business associates.
HHS also notes that on June 20, 2024, a federal court "issued an order declaring unlawful and vacating a portion of this guidance document." HHS says the vacated portion is the guidance as it applied where a technology connects an individual's IP address with a visit to an unauthenticated public page about specific health conditions or healthcare providers. Which parts still apply, and to which pages on your site, is a question for your privacy officer or counsel. The working rule for marketing is simple: no new pixel, chat widget, or conversion tag goes on appointment request forms, patient portals, or logged-in pages without that review.
A fictional monthly scorecard
Here is a hypothetical scorecard layout for a fictional orthodontic practice with two offices in Georgia. It shows what to collect, not results any practice achieved.
| Measure | Source | What to note |
|---|---|---|
| Unbranded questions | Saved answers from the fixed question set | Answers naming the practice, out of the total checked |
| Branded accuracy | Saved answers to questions by name | Which facts were wrong, and the likely source |
| AI Overviews and AI Mode impressions | Search Console generative AI report | Direction of change, with the report's limits stated |
| AI Assistant sessions | GA4 channel report | Landing pages and what visitors did next |
| Profile calls and website clicks | Business Profile performance | Monthly totals by office |
| New patient requests | Practice scheduling records | Count, with self-reported "how did you hear about us" answers kept separate |
The last row comes from the practice, not a dashboard. At the monthly review, the practice manager can also ask: "Which wrong fact did we see most, and where on our own site or profile does it still appear?" Record the self-reported answer without health details, and keep it apart from the tracked source. A patient may read an answer, remember the name, and call days later, and analytics may not connect those steps.
Measurement connects back to each of Hudson South's four steps: the Authority Asset gathers the approved facts the fact sheet comes from, Hudson Twenty-One records how the practice communicates across 21 dimensions, Market Intelligence researches the patient questions that make up the question set, and AEO Creation turns those inputs into answer-first content whose effect the scorecard tracks. For the wider set of practice pages, see content for medical, dental and wellness practices.
Start with a clear view of your competitors
On a free 30-minute call, Hudson South walks you through the gaps between your website and the practices patients compare you with.
Get your free competitor analysisSources
- Generative AI performance report (Search). Google Search Console Help. Read 2026-10-15.
- Default channel group. Google Analytics Help. Read 2026-10-15.
- Publishers and Developers - FAQ. OpenAI Help Center. Read 2026-10-15.
- Understand your Business Profile performance & insights. Google Business Profile Help. Read 2026-10-15.
- Use of Online Tracking Technologies by HIPAA Covered Entities and Business Associates. U.S. Department of Health and Human Services. Read 2026-10-15.