Technical guide

Schema for AI: which structured data to use, and which you don’t need

Structured data (schema.org in JSON-LD format) describes your brand, products and pages so machines don’t have to guess. Google says no special schema is needed for its AI features, but correct Organization, Product or LocalBusiness markup keeps your facts consistent and makes you eligible for rich results. GeoRush checks and generates that JSON-LD.

Updated on

Detail of a curved glass and steel roof

What schema.org and JSON-LD are

Schema.org is a shared vocabulary for describing things on the web: organizations, products, people, articles, local businesses, events. JSON-LD is the most common format for adding that vocabulary to a page: a <script type="application/ld+json"> block with data the visitor doesn’t see, but which restates in an orderly way what they do see.

Example: your home page says “Example is invoicing software for freelancers, based in Austin”. The JSON-LD says the same thing in fields: name, description, url, logo, sameAs (your official profiles). A crawler doesn’t have to infer your official name or logo: you tell it.

In GEO (generative engine optimization) schema gets talked about a lot, sometimes with inflated promises. This guide separates what is documented from what is assumption.

What Google says (and what others don’t)

Google is the only major player that documents in detail how schema relates to AI. Its position, in its AI features documentation and in the generative AI optimization guide published in May 2026, is clear: there is no special schema.org markup you need to add to appear in AI Overviews or AI Mode, and structured data is not required for generative AI search. At the same time, Google recommends you keep using it as part of SEO, because it makes you eligible for rich results, and asks that your markup match the visible text on the page.

Google’s general rules also apply: JSON-LD is the recommended format, you shouldn’t mark up content users can’t see, you shouldn’t mark up misleading content (such as fake reviews), and correct markup does not guarantee any special result.

OpenAI, Anthropic and Perplexity document their crawlers, but we have found no official documentation from them explaining whether or how they use structured data. Any claim like “ChatGPT reads your schema and that’s why it recommends you” is an assumption, not a published fact.

What schema does for GEO (realistically)

BenefitDocumented?Comment
Rich results in Google (product, reviews, local business, etc.)Yes, by GoogleEach type has its own requirements, and display is not guaranteed.
Helping Google understand your organization’s admin details (name, logo)Yes, by GoogleYour marked-up logo can influence the one shown in results and the knowledge panel.
Appearing in AI Overviews / AI ModeGoogle says no special schema is neededWhat counts is being indexed and eligible to show with a snippet.
Helping ChatGPT, Claude or Perplexity understand youNot documentedPlausible as an aid to consistent data, but not publicly verified.
Entity consistency (same facts everywhere)Recommended practiceSchema forces you to settle on one official version of your facts.

The schema types worth reviewing first

TypeWhere it goesKey propertiesNote
OrganizationHome page or About page (not needed on every page)name, url, logo, description, sameAs, contactPoint, legalNameGoogle requires no property; it recommends adding all relevant ones. For stores, use OnlineStore.
LocalBusiness (and subtypes)Each location pagename and address (required by Google), geo, openingHoursSpecification, telephone, url, priceRangeUse the most specific subtype (Restaurant, Dentist, Hotel…).
Product + OfferEach product pagename, description, image, offers (price, currency, availability)Check Google’s requirements for merchant listings and product snippets.
SoftwareApplicationSoftware product pagename, offers.price and aggregateRating or review (required for the rich result), applicationCategory, operatingSystemNo real ratings, no rich result: don’t invent them.
Article / BlogPostingGuides, articles, newsheadline, author, datePublished, dateModified, imagedateModified helps show the content is current.
PersonAuthor or expert pagename, jobTitle, worksFor, sameAsMakes clear who stands behind the content.
BreadcrumbListEvery page with a hierarchyitemListElementDescribes the site structure.
FAQPagePages with visible FAQsmainEntity with Question and acceptedAnswerGoogle stopped showing the FAQ rich result on May 7, 2026. The vocabulary still exists in schema.org.

Example: Organization with sameAs

This is the most important block for brand consistency. sameAs links to pages that unambiguously identify your organization: official profiles, and Wikidata or Wikipedia if they exist. Replace every value with yours and make sure it matches what the page says:

organization.jsonld
<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Organization",
  "name": "Example",
  "legalName": "Example Software Inc.",
  "url": "https://www.example.com/",
  "logo": "https://www.example.com/logo.png",
  "description": "Invoicing software for freelancers and small businesses in the US.",
  "foundingDate": "2019",
  "sameAs": [
    "https://www.linkedin.com/company/example",
    "https://www.youtube.com/@example",
    "https://www.wikidata.org/wiki/Q00000000"
  ],
  "contactPoint": {
    "@type": "ContactPoint",
    "contactType": "customer support",
    "url": "https://www.example.com/contact",
    "availableLanguage": ["en", "es"]
  }
}
</script>

Example: local business

For businesses with a physical location, use the most specific subtype and repeat your Google Business Profile details exactly (name, address, phone, hours):

local-business.jsonld
<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Dentist",
  "name": "Example Dental Clinic",
  "url": "https://www.example.com/denver",
  "telephone": "+1 303 000 0000",
  "priceRange": "$$",
  "address": {
    "@type": "PostalAddress",
    "streetAddress": "1 Example St",
    "addressLocality": "Denver",
    "addressRegion": "CO",
    "postalCode": "80202",
    "addressCountry": "US"
  },
  "geo": { "@type": "GeoCoordinates", "latitude": 39.75222, "longitude": -104.99920 },
  "openingHoursSpecification": [{
    "@type": "OpeningHoursSpecification",
    "dayOfWeek": ["Monday", "Tuesday", "Wednesday", "Thursday", "Friday"],
    "opens": "09:00",
    "closes": "18:00"
  }]
}
</script>

Example: product with an offer

The price and availability in the markup must match the page. When prices change, update both at once:

product.jsonld
<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Product",
  "name": "Example Backpack 30L",
  "description": "Waterproof 30-liter hiking backpack.",
  "image": "https://www.example.com/img/backpack-30l.jpg",
  "brand": { "@type": "Brand", "name": "Example" },
  "sku": "EX-30L",
  "offers": {
    "@type": "Offer",
    "price": "89.00",
    "priceCurrency": "USD",
    "availability": "https://schema.org/InStock",
    "url": "https://www.example.com/backpack-30l"
  }
}
</script>

How to implement it, step by step

  1. 1

    Settle your official facts

    Before writing code, decide the single version of your name, description, category, logo, official profiles and contact details. That fact sheet is the base for everything else.

  2. 2

    Start with Organization

    Add an Organization block to your home page or About page, with sameAs pointing to your official profiles.

  3. 3

    Mark up your main offer

    Product, SoftwareApplication, LocalBusiness or Service depending on what you sell, on each product, service or location page.

  4. 4

    Mark up editorial content

    Article with author and dates on guides and articles; BreadcrumbList across the site.

  5. 5

    Validate

    Check syntax with the Schema Markup Validator and rich result eligibility with Google’s Rich Results Test.

  6. 6

    Keep it in sync

    Whenever a price, opening hours or an address changes, change the page and the markup together. Outdated markup contradicts your own site.

Common mistakes

  • Marking up data that isn’t on the page (prices, ratings, questions users can’t see). Google explicitly prohibits it.
  • Adding reviews of your own company to Organization or LocalBusiness: Google says that if the entity controls the reviews about itself, those pages are ineligible for star ratings.
  • Having several Organization blocks with different names or logos depending on the template.
  • Using a plugin that outputs generic markup without checking what it says.
  • Thinking schema replaces content: if the page doesn’t answer the question, markup won’t fix it.
  • Keeping FAQPage as the only reason for having FAQs: their value today is in the visible content, not the rich result.

How GeoRush helps

The AI technical section of the SEO and technical audit checks your site’s structured data alongside robots.txt and llms.txt, and includes JSON-LD, robots.txt and llms.txt generators. The brand identity feature stores your official facts so you use them the same way everywhere, and brand perception checks whether what assistants say matches them.

Is your site ready for AI?

The free check audits your site for AI (crawlers, structured data, llms.txt) and gives you an A–E grade with the first fixes.

Sources

Official documentation and primary sources checked on November 17, 2026. These policies change often: check the source before making important decisions.

Frequently asked questions

No. Google says there is no special schema.org markup you need to add for its AI features. The page needs to be indexed and eligible to show with a snippet; schema remains useful for rich results.

We have found no official documentation from OpenAI or Perplexity explaining whether or how they use schema.org. Using it is reasonable to keep your facts consistent, but don’t present it as a guarantee.

Organization, on your home or About page, with name, URL, logo, description and sameAs. Then the type that describes what you sell: Product, SoftwareApplication, LocalBusiness or Service.

Google stopped showing the FAQ rich result on May 7, 2026. The type still exists in schema.org, and visible FAQs remain useful as content.

Use the Schema Markup Validator for syntax and Google’s Rich Results Test for eligibility. Also check that every value matches the visible text.

Keep reading