When ChatGPT decides whether to recommend a dentist in Charlotte or an estate attorney in Phoenix, it's not reading your website the way a human does. It's parsing patterns — and structured data is the cleanest, most unambiguous pattern you can hand it. A page without schema forces an AI system to guess: Is "Dr. Patel" a person, a clinic, or a brand? Is "5.0" a rating, a price, or a coincidence? A page with proper schema removes the guesswork entirely.
That's why structured data has quietly become one of the highest-leverage moves in AI search optimization. It's not glamorous. It doesn't drive traffic on its own. But it dramatically changes how confidently an LLM can identify, classify, and cite your business when someone asks for a recommendation.
Why Schema Matters More for AI Than for Google
Google has spent two decades getting good at inferring meaning from messy HTML. AI systems, by contrast, are pattern-matching across trillions of tokens — and structured data is the most reliable signal they get. When an LLM crawls your site (or pulls from a search index that crawled it), JSON-LD blocks give it:
- Unambiguous entity type — "Dentist," not "professional service."
- Canonical facts — name, address, phone, license number, credentials.
- Relationship data — which practitioners belong to which practice.
- Evaluation signals — aggregate ratings, review counts, awards.
Unstructured text can be misread. A bio that mentions "Dr. Lin trained at Harvard before joining the practice" might be parsed correctly — or it might end up attributing Harvard to the practice itself. Schema eliminates that ambiguity. For a deeper look at how this fits into the broader optimization picture, our generative engine optimization guide walks through the full stack.
The 6 Schema Types That Actually Move the Needle
Not all schema is created equal. After analyzing how AI systems cite local professionals, six schema types consistently correlate with stronger entity recognition and higher citation rates.
1. LocalBusiness (and Vertical Subtypes)
This is your foundation. But the mistake most practices make is using the generic LocalBusiness type when a more specific subtype is available. Schema.org provides vertical-specific subtypes that signal exactly what kind of professional you are:
- Dentist — for general and specialty dental practices
- Physician — for individual doctors or small practices
- MedicalClinic — for multi-provider medical groups
- Attorney — for lawyers and law firms
- LegalService — broader legal service providers
- RealEstateAgent — for individual agents
- RealEstateBroker — for brokerages
A dentist in Austin using Dentist instead of LocalBusiness makes it trivially easy for ChatGPT to surface them when a user asks for "dentists in Austin." The generic type forces inference; the specific type confirms it.
2. FAQPage — The Highest-ROI Schema for AI
If you only implement one schema type beyond LocalBusiness, make it FAQPage. AI systems extract question-answer pairs directly from this markup and frequently quote them in responses. When someone asks Perplexity, "How long does Invisalign take in adults?" — and your FAQ page has that exact question marked up with a clear answer — you've created a citation-ready chunk of content.
FAQPage schema works because it mirrors the structure of how people query AI:
- A user asks a natural-language question.
- The AI looks for question-answer pairs in its training data and retrieved sources.
- Structured FAQ content is the cleanest match.
- Your answer gets surfaced (often with attribution).
We've seen real estate agents in Denver double their AI mention rate simply by adding 8–12 FAQPage entries covering common buyer and seller questions. The questions weren't novel — they were the questions the agent already answered ten times a week on the phone.
3. Review and AggregateRating
AI systems use review signals to assess recommendation confidence. If two dentists are equally relevant to a query, the one with verifiable aggregate ratings is more likely to be named. AggregateRating embedded in your LocalBusiness schema gives the AI a clean number to anchor on: rating value, review count, best/worst rating scale.
This doesn't replace your presence on third-party platforms like Healthgrades, ZocDoc, or Avvo — it complements them. AI cross-references your self-declared rating with what it sees on independent sources. Consistency builds confidence; contradictions create doubt.
4. Person Schema for Practitioners
For solo practitioners or named providers within a practice, Person schema is essential. It links the individual entity to the business entity, which is how AI systems handle queries like "Who is Dr. Sarah Chen, the dermatologist in Seattle?"
Person schema should include:
- Full name and credentials (MD, DDS, JD, Esq.)
- Job title and specialty
worksForlinking to the LocalBusiness entityalumniOffor medical/law schoolmemberOffor professional associationssameAslinking to verified profiles (Doximity, Martindale-Hubbell, Super Lawyers, state bar pages)
The sameAs property is particularly important. It tells AI systems that the "Dr. Chen" on your site is the same "Dr. Chen" on Healthgrades and the state medical board — collapsing what could otherwise be treated as three separate people into one verified entity.
5. MedicalClinic / LegalService Specialty Fields
These types unlock specialty-specific fields that LLMs can parse directly. MedicalClinic supports medicalSpecialty (e.g., "Cardiology," "Pediatric Dentistry"), availableService, and acceptedInsurance. LegalService supports detailed service area and practice area declarations.
When a user asks Claude, "Family law attorneys in Tampa who handle high-net-worth divorces," the AI is trying to match three facets: location, practice area, and case type. A LegalService block that explicitly declares "Family Law" and "High-Net-Worth Divorce" as service types is doing the matching work for the LLM.
6. Article and BlogPosting
For every piece of content on your site — case studies, educational posts, market updates — Article or BlogPosting schema with proper author, datePublished, and dateModified fields signals freshness and authority. AI systems weight recent, authored content more heavily, particularly for time-sensitive topics like real estate market trends or evolving medical guidelines.
Author attribution also reinforces Person schema. If your managing partner publishes a blog post on estate planning and that post is marked up with BlogPosting + author=Person, you're stacking entity signals across multiple pages.
A Sample JSON-LD Block for a Dental Practice
Here's a working LocalBusiness block for a dental practice in Charlotte, combining several of the elements above:
{
"@context": "https://schema.org",
"@type": "Dentist",
"name": "Queen City Family Dental",
"image": "https://queencityfamilydental.com/images/office.jpg",
"@id": "https://queencityfamilydental.com/#dentist",
"url": "https://queencityfamilydental.com",
"telephone": "+1-704-555-0142",
"priceRange": "$$",
"address": {
"@type": "PostalAddress",
"streetAddress": "1450 Park Road",
"addressLocality": "Charlotte",
"addressRegion": "NC",
"postalCode": "28203",
"addressCountry": "US"
},
"geo": {
"@type": "GeoCoordinates",
"latitude": 35.1956,
"longitude": -80.8487
},
"openingHoursSpecification": [{
"@type": "OpeningHoursSpecification",
"dayOfWeek": ["Monday","Tuesday","Wednesday","Thursday","Friday"],
"opens": "08:00",
"closes": "17:00"
}],
"aggregateRating": {
"@type": "AggregateRating",
"ratingValue": "4.9",
"reviewCount": "287"
},
"sameAs": [
"https://www.healthgrades.com/dentist/queencityfamily",
"https://www.yelp.com/biz/queencityfamilydental"
],
"employee": {
"@type": "Person",
"name": "Dr. Maya Reyes",
"jobTitle": "DDS",
"alumniOf": "University of North Carolina School of Dentistry"
}
}
This block tells any AI crawler: this is a dentist, in Charlotte, at this address, with this rating, with this dentist on staff, verified across these external sources. Every field reduces ambiguity.
What Schema Can't Do
Schema is a clarifier, not a creator. It will not:
- Manufacture authority you haven't earned. If no third-party sources mention your practice, perfect schema won't get you recommended. AI systems still need corroborating signals from directories, reviews, and press.
- Override negative entity signals. If your aggregate rating across third-party platforms is 3.1 stars, declaring a 5.0 rating in your own schema creates a contradiction the AI will resolve against you.
- Substitute for real content. FAQPage schema on a page with three thin Q&As won't outperform a comprehensive resource with deep, useful answers.
Schema amplifies what's already there. If you want to understand how those underlying signals get weighted, see our breakdown of how ChatGPT, Claude, Gemini, and Perplexity recommend businesses.
Validation: Don't Ship Broken Markup
Before any schema goes live, validate it. The Google Rich Results Test is the industry-standard validator — paste in your URL or raw code and it will flag missing required fields, syntax errors, and warnings. Schema.org's own validator is useful for types Google doesn't fully support yet.
Common errors to watch for:
- Missing
@contextor@typedeclarations - Mismatched name/address/phone between schema and on-page content
- Aggregate ratings without corresponding reviews
- FAQPage with answers that are too short or contain HTML that breaks parsing
Re-validate after every major site change. A redesign that strips the JSON-LD blocks from your templates is one of the most common causes of sudden AI visibility drops.
How Schema Fits Into Your AI Visibility Score
Structured data is one input — a powerful one — into the broader question of whether AI systems recognize and recommend your business. To see how schema sits alongside citation density, review signals, and content authority, read our explainer on what an AI visibility score actually measures.
The pattern we see consistently: practices with strong third-party signals but no schema get cited inconsistently. Practices with strong schema but no third-party signals get ignored. The compound effect happens when both are in place — and that's when AI systems start naming you by default.
See Where Your Schema Stands
If you're not sure whether your current schema is helping or hurting your AI visibility, the fastest way to find out is to scan your business. RankCommander analyzes how ChatGPT, Claude, Gemini, and Perplexity see your practice — including the structured data signals they're pulling from — and shows you exactly which schema gaps are costing you citations. Run your free AI visibility scan now and get a clear picture of what to fix first.