Google Drops Healthcare Review Snippets — What It Really Signals

A health-tech platform we audited in August 2026 — a Delhi NCR physician-discovery service with 28 location pages — noticed their star-rating strips had vanished. Perfect AggregateRating markup. Verified clean through Google Rich Results Test. Still no ratings showing in Search.

Their dev team's first instinct was a schema bug. We ran the diagnostic and found something more clarifying. Four of their 28 pages had machine-readable MedicalOrganization + Physician entity stacks with proper sameAs anchors pointing to hospital registry URLs and medical council profiles. Those four pages were appearing in AI Overview responses for "neurologist in Delhi NCR" and "diagnostic lab Gurgaon." The other 24 pages had four-star ratings. And zero AI citations.

That gap — between star display and AI citation — is what Google's healthcare review snippet removal is actually telling you.

What exactly changed with Google's healthcare review snippets?

Review rich snippets are the visual star ratings that appear under search results — the "4.8 ★ (312 reviews)" line you'd see under a clinic or doctor listing. Until May 2026, healthcare businesses could display these by implementing AggregateRating schema markup on their pages. Google would read that structured data and render the stars in Search results.

Starting late May 2026, the stars stopped appearing. Barry Schwartz at Search Engine Roundtable confirmed the pattern on October 7, 2026: review snippets aren't showing in doctor and dentist searches outside of the local pack. Andrea Badder from Schema App — a structured data consultancy — had tracked two waves of visibility drops across their healthcare clients: one starting late May 2026, a second in early August 2026.

The team at Schema App ran multi-typing tests. They tried adding Product schema alongside the healthcare entity markup. Nothing brought the stars back. Google hasn't issued an official explanation — which isn't unusual for display-format changes. But the absence of a statement makes the signal louder, not quieter.

Why did Google remove review snippets from healthcare specifically?

Healthcare is a YMYL vertical — "Your Money or Your Life" in Google's own Quality Rater Guidelines. That label means Google applies the highest editorial standards because incorrect health information, or fake credibility signals, can cause real harm to real people.

Star ratings are easy to inflate. Fake reviews are a documented problem across Yelp, Google Maps, and Healthgrades. A clinic with 4.9 stars built on 12 reviews from staff members isn't a trust signal. It's noise. Google made a judgment call: in healthcare, the risk of amplifying gamed social-proof signals outweighs the UX benefit of showing stars.

We've watched this pattern play out before. Hotels got a similar treatment when Google launched its own hotel-search experience — third-party star snippets were replaced by Google's own verified booking data. Products faced review-gating pressure after fake product reviews became a widespread manipulation vector. Healthcare is the latest vertical where Google decided it didn't want to amplify a signal it can't authenticate. The parallel to Google's site reputation policy crackdown isn't accidental — both moves reduce reliance on third-party signals that can be manufactured without Google's ability to verify them.

What does Google use instead of star ratings to evaluate healthcare trust?

Schema App's post-mortem pointed directly at the answer: the correct pivot isn't to recover the snippet. It's to build the machine-readable entity layer that AI Overviews actually use to evaluate and cite healthcare sources.

Specifically, that means correct entity typing in your structured data. MedicalOrganization for a hospital or clinic. Physician for individual doctors. MedicalSpecialty for the conditions and disciplines the practice covers. availableService for specific procedures. And — critically — sameAs anchors: external URLs that point to verifiable records outside your own website.

Those external records could be a hospital's government registration listing, a doctor's profile on the Indian Medical Council website, a Wikidata entity, or a recognized health directory. When an AI grounding pass encounters a Physician schema block with a sameAs pointing to a verifiable external record, it gets confirmation the entity is real. That's a machine-confidence signal. An AggregateRating block with 4.8 stars, by contrast, tells the AI nothing it can confirm — it can't verify whether those reviews are real or gamed.

This is also why we've seen our AI SEO strategy recommendations for healthcare-adjacent clients focus on entity verification over display optimization since early 2026. The star-rating moment had always been a display layer in classic Search. AI Overviews operate on a different layer entirely. For the full breakdown of which schema types and fields feed AI grounding passes — not just for healthcare but for every B2B vertical — our technical AI SEO pillar covers the stack end to end.

Why AI Overviews don't need star ratings to trust your healthcare brand

Here's the part most SEO coverage is missing when they write about this change.

Star-rating snippets were always a human-trust signal. They worked because people read them. A 4.9-star rating under a doctor's name made a person feel more confident before clicking. That's a UX play — a visual reassurance mechanism optimized for human cognition.

AI Overviews don't read pages the way humans do. They run a grounding pass: an extraction process that looks for structured, verifiable, machine-readable data that can be synthesized into a response without hallucination risk. Stars and review counts don't survive that process cleanly — they require the AI to trust that the underlying reviews are authentic, which it can't verify from markup alone. An entity-correct JSON-LD block with verifiable external references does survive the grounding pass. It gives the AI something it can confirm, cite, and include in a synthesis response.

Schema App put it well: "Rich results are still valuable when Google awards them. But earning a particular rich result is becoming a smaller part of what an advanced Schema Markup strategy needs to accomplish." That's not a consolation-prize statement. It's a fundamental reframe of what structured data is for. For the last decade, many healthcare and B2B brands optimized their schema to win SERP decorations — the visual UI elements that make blue links look more appealing. That job is shrinking. The new job is building the schema layer that gets you cited in AI Overviews.

We can't fully predict which healthcare queries will shift to AI Overview-dominant responses over the next 12 months — the AI Overview expansion is still accelerating and the coverage is uneven by query type. But based on our Q3 2026 audits, we're already seeing AI Overview responses dominate informational healthcare queries ("what is rheumatoid arthritis treatment" and "best diagnostic lab Delhi") at rates above 60%. Entity-correct pages win those slots. Star-only pages don't appear at all.

What should healthcare businesses do right now?

Five steps, roughly in order of impact. You don't need to do all of them at once — but step 1 and 2 together move the needle fast.

1. Verify your entity type declaration. Open Google's Rich Results Test and paste in your homepage or a representative clinic page. Check that the parsed schema type is MedicalOrganization, Hospital, MedicalClinic, or Physician — not just Organization or LocalBusiness. The type determines which AI Overview slots your page is eligible for. Generic types won't match healthcare entity sub-queries.
2. Add sameAs anchors to verifiable external records. For each entity, add at least two sameAs URLs: a government or professional registry link (hospital registration, medical council) and a Wikidata or recognized health directory entry. These external anchors are the verification layer that AI grounding passes look for. A page with zero sameAs is an unverified claim. In our Q3 2026 audit, every page that earned an AI Overview citation had at least two external sameAs anchors — government registry or medical council links, not third-party directories. A page with two verified external records is a confirmed entity.
3. Add medicalSpecialty and availableService fields. Healthcare queries are almost always specialty-scoped ("neurologist", "laparoscopy", "MRI scan"). Declaring the specialty and available services in your structured data is how AI Overviews match your entity to the right sub-queries. Without these fields, your hospital page might be entity-correct but still invisible for specialty-specific patient searches.
4. Build Physician-level entity blocks for individual practitioners. Patients search for doctors by name and specialty, not just clinics. Each practitioner on your site deserves their own Physician schema block with name, medicalSpecialty, worksFor (pointing to the parent MedicalOrganization), and a sameAs to their medical council profile. Individual-practitioner entity blocks generate individual AI Overview citation slots — they're the highest-specificity unit in healthcare AI SEO.
5. Run a 30-day AI Overview citation audit. After updating your schema, submit the changed URLs in Google Search Console for re-indexing. Then run your top 10 target patient queries (e.g., "best cardiologist in [city]", "[procedure] near me") in Google incognito weekly for 30 days. Log whether your pages appear as AI Overview sources. That citation rate is your new healthcare visibility KPI — the practical replacement for the star-rating snippet you've lost.

Is your healthcare site ready for the AI Overview era?

Run through this before your next schema update — we've used this exact sequence on every healthcare schema engagement since Q2 2026. The first five items are blocking; we've never seen an AI Overview citation on a page that missed any of them.

• Entity type correctly declared: MedicalOrganization, Hospital, MedicalClinic, or Physician — not just Organization
• At least two sameAs URLs per entity pointing to external verifiable records (government registry, Wikidata, medical council)
• medicalSpecialty declared at organization AND individual practitioner level
• availableService fields enumerate specific procedures or treatments as structured data (not just prose)
• Individual Physician blocks for all practitioners, each with worksFor reference to parent organization
• FAQPage JSON-LD covering the top 5 patient-intent long-tails (symptoms, appointment, treatment, cost, timing queries)
• GPTBot, PerplexityBot, ClaudeBot, and OAI-SearchBot allowed in robots.txt
• llms.txt includes your key healthcare service page URLs
• Each H2 section passes the 40–80 word extractable-answer test — a clean definition an AI can lift and cite
• AggregateRating still present as a secondary field (don't remove it — it feeds local pack and third-party platforms)

Review-snippet strategy vs entity-first AI SEO: side by side

Here's what the shift looks like in practice across every dimension that matters for a healthcare brand's search visibility in 2026 and beyond.