Medical Schema Markup: A Practical Guide

What schema does for a healthcare site in 2026

Google supports a specific list of structured data features for rich results, and the list keeps shrinking. On 7 May 2026 FAQ rich results stopped appearing in Google Search, with a notice added to Google’s FAQ structured data documentation. Most medical types on Schema.org have never produced a rich result.

So why do I still build it on every healthcare site? Because structured data is the cheapest way to remove doubt. It states which organisation publishes the page, which physician reviewed it, which location offers which service, and which public register confirms a clinician’s identity. Search systems and AI answer engines that are deciding whether to trust a health source are resolving exactly those questions.

The types that matter

Build a graph, not snippets

The most common problem I find is schema pasted page by page by a plugin: five different “organisation” objects with slightly different names, locations without addresses, and authors who are just a text string. Search engines can’t join those into one picture.

The fix is a graph with stable identifiers:

1. Give the organisation one @id (for example your domain plus #organization) and declare it once, sitewide.
2. Give each location an @id on its own page, and point parentOrganization at the organisation’s @id.
3. Give each clinician an @id on their profile page and reference that @id wherever they are an author or reviewer.
4. On content pages, reference those @ids rather than repeating the data.

That is how this site is built, and it is the first thing I check in a healthcare SEO audit.

sameAs to registers: the strongest trust signal

A Person node that says “Dr Jane Smith, cardiologist” is a claim. A Person node with sameAs links to the doctor’s public register entry (the NPI registry in the US, the GMC register in the UK, the AHPRA register in Australia or a provincial college in Canada) is a claim a machine can check. Add the practice’s regulator profile, such as a CQC page, to the organisation node in the same way.

This is the schema work that pays off most in AI search, which is covered in my AI SEO for healthcare service.

Labs and test pages

Diagnostic labs have a type built for them: DiagnosticLab with availableTest pointing to MedicalTest entities. In my lab website benchmark, only 3 of 30 lab homepages used MedicalTest. Each test page should carry a MedicalTest node and each collection site should list what it offers. The detail is in how pathology labs rank on Google Maps.

Errors I fix most often

• Self-serving reviews. Google’s review snippet documentation says reviews a business publishes about itself are not eligible for review rich results. Aggregate ratings pulled from testimonials into your own LocalBusiness markup add risk and no benefit, and in some countries showing testimonials is itself restricted.
• Markup for things that aren’t on the page. FAQs in schema that no visitor can read, services a location doesn’t offer.
• Conflicting NAP. The address in schema differs from the Google Business Profile or the footer.
• Authors as strings. “author”: “Admin”. Use a Person @id.
• Medical codes guessed. Only add codes (ICD, SNOMED, CPT) if someone qualified has checked them.

Validation workflow

1. Validate syntax with the Schema.org validator and eligible features with Google’s Rich Results Test.
2. Crawl the site and extract every JSON-LD block, then check that each @id resolves to one consistent entity.
3. Compare schema addresses and hours against each Google Business Profile.
4. Recheck after every CMS or plugin update, because plugins overwrite graphs.

If you want your graph checked, book a strategy call.