Short answer
No vendor publishes which profile fields an assistant reads, so treat any list that claims to know as marketing. What is documented is narrower and more useful: the fields that identify you, the ones assistants cross-check against your site, and the handful that go stale and turn into wrong answers.
Key takeaways
- Name, address, phone, hours, category and services are the fields that assistants can check against other sources. Disagreement is what costs you the mention.
- Google documents name and address as the only required properties for LocalBusiness structured data. Everything else is recommended, which is not the same as optional in practice.
- Google’s own structured data guidance does not cover medical subtypes. schema.org does: Dentist, Physician, MedicalClinic.
- Hours and phone are the fields that rot fastest, and they are exactly the ones patients act on.
Every month somebody publishes a list of the Google Business Profile fields that AI "prioritises". None of them can know. Google does not publish it, and nobody has run a controlled test at a scale that would settle it. What can be said honestly is narrower, and it is enough to work from.
What is actually documented
For structured data on your own site, Google documents exactly two required properties on a LocalBusiness: name and address. Telephone, opening hours, geo coordinates, price range, URL and ratings are listed as recommended. Recommended means Google will accept the markup without them, not that they do nothing for you.
Worth knowing for health practices: that guidance uses examples like restaurants and day spas and does not cover medical subtypes. schema.org itself does have them, including Dentist, Physician and MedicalClinic, and using the specific subtype is still the right call. Just do not expect a Google rich result to appear because you did.
The fields that decide whether a model can describe you
An assistant assembles a description by cross-checking sources. Every field below exists in at least three places: your website, your Business Profile, and the directories. When they agree, you are one clear entity. When they disagree, the model reaches for a competitor it can describe without hedging.
| Field | Why it matters | How it usually goes wrong |
|---|---|---|
| Business name | The string a model matches across sources | Keyword stuffing, or a trading name the website never uses |
| Address | One of the two properties Google requires | Suite number present in one listing, missing in another |
| Phone | What a patient acts on immediately | An old tracking number left in a directory nobody manages |
| Hours | Quoted directly in answers, and acted on | Holiday hours never updated, so the answer is confidently wrong |
| Primary category | How you get shortlisted at all | Too broad, or chosen once at setup and never revisited |
| Services | The vocabulary a patient question matches | Lists services the website does not mention anywhere |
| Practitioner credentials | Weighed heavily where expertise matters | Spelled three ways across three profiles |
Why disagreement costs more than a missing field
A blank field is a gap. A contradiction is a reason to distrust the whole record. If your site says you are open until six, the profile says five, and a directory says four, a model has no way to choose, and the safest thing it can do is describe a practice whose sources agree with each other.
This is not theoretical. In rater8’s April 2026 survey of 992 US adults, 66% of the patients who had used AI to research a provider had met incorrect information, most often addresses, phone numbers, insurance and hours, and 60% did not check it anywhere else.
A 30-minute audit worth doing this quarter
- Put your website contact page, your Google Business Profile and your three most visible directory listings side by side in one document.
- Compare name, address, suite, phone, hours, category, services and credentials line by line. Write down every disagreement, including punctuation.
- Pick the version that is true, then change every other source to match it exactly rather than approximately.
- Fix holiday and seasonal hours in the same pass, and put a recurring reminder in whoever owns the calendar.
- Re-check in three weeks. Some directories overwrite edits from their own upstream data, and you want to know which ones do that.
What this will not do
Consistent data does not make a new practice outrank an established one, and it does not create demand. It removes the reasons a model has to leave you out or describe you wrongly. That is worth doing precisely because it is cheap, dull, and almost nobody does it.
Questions people ask
Which Google Business Profile fields does AI actually use?
Nobody outside Google can answer that honestly, because it is not published. What is documented is which properties Google requires in LocalBusiness structured data, name and address, and which it recommends. Treat any confident list of AI-prioritised fields as marketing.
Does my medical practice need a specific schema type?
Use the most specific schema.org type that fits: Dentist, Physician or MedicalClinic rather than plain LocalBusiness. Google’s own structured data guidance does not cover medical subtypes, so expect clearer machine parsing rather than a visible rich result.
How often should we update the profile?
Hours whenever they change, including holidays, and everything else on a quarterly check. Hours and phone rot fastest and are the fields patients act on immediately, which makes them the most expensive ones to get wrong.
Do photos and posts help with AI visibility?
There is no published evidence either way, and it would be wrong to claim otherwise. They help human visitors decide, which is a good enough reason. Do not let them displace the audit of fields that assistants can cross-check.
What if a directory keeps reverting my correction?
Some listings pull from upstream data aggregators or from your payer directories, and they overwrite manual edits. Fix the upstream source, then the listing, then re-check in a few weeks. Note which ones behave this way so the next person is not surprised.
Sources
- Google Search Central, local business structured data documentation (accessed 20 September 2026): name and address are the only required properties; telephone, hours, geo, priceRange, url and ratings are recommended; medical subtypes are not covered.
- rater8, 2026 Patient Choice Report (April 2026, n=992 US adults, vendor survey): 66% of AI users met incorrect provider information, most often address, phone, insurance and hours; 60% did not verify it.
Published . Last reviewed .
Want this in your business?
Book a call and we will scope exactly which part of your operation an agent should take first.
Book a call with the team