Schema markup for AI search: what it can and cannot do

Structured data helps Google understand what a page is about and can make it eligible for rich results. It is not a documented way into AI answers: Google says structured data is not required for AI Overviews or AI Mode, and that there is no special schema.org markup to add for them.

The other makers say less. The crawler and publisher pages from OpenAI, Anthropic and Perplexity that we read on 23 September 2026 do not mention structured data at all. Add it for what Google documents it does, and do not expect more.

What structured data is

Schema.org is a shared vocabulary for describing things on a page: organizations, products, articles, people. It was founded by Google, Microsoft, Yahoo and Yandex, and says that as of 2024 more than 45 million web domains used it. Google accepts it as JSON-LD, Microdata or RDFa, and recommends JSON-LD as the easiest to maintain. Our glossary entry on structured data covers the basics.

What Google says it does

• Explains the page. Google calls structured data explicit clues about a page's meaning. It also uses it to gather information about the world, such as the people and companies named in the markup.
• Makes rich results possible. Correct markup makes a page eligible for enhanced displays such as product snippets with price and ratings.
• Tells your organization apart. Organization markup can help Google understand your administrative details and disambiguate your organization in search results. Some properties can influence which logo appears and your knowledge panel. The glossary entries on entities and the Knowledge Graph explain why that matters.
• Carries links to your other profiles. Google says it can make general use of the sameAs property. Schema.org defines it as a page that unambiguously indicates the item's identity, such as its Wikipedia page or official website.
• Counts as permission. Google's max-snippet rule limits how much of a page AI Overviews and AI Mode can use, but Google says the limit does not apply to content you supply as in-page structured data.

What Google says it does not do

• It is not required for AI features. Google's guide to its generative AI features lists "overfocusing on structured data" among things you do not need to do. It recommends keeping structured data for rich results.
• It guarantees nothing. Google says valid markup makes a feature possible but does not guarantee it will show.
• FAQ markup no longer earns a rich result. Google stopped showing the FAQ rich result on 7 May 2026 and removed its documentation in June 2026. Whatever else FAQ markup does, it will not produce that display.

What the other engines document

Nothing, on structured data itself. What their pages and outside measurements do show is how their crawlers read a page, and that affects where your markup must live.

Vercel and MERJ found that the crawlers from OpenAI, Anthropic and Perplexity do not run JavaScript. They noted that data in the initial HTML response, such as JSON, may still be indexed. Google, by contrast, says it can read JSON-LD that JavaScript injects into a page.

So markup added through a tag manager or a client-side script can reach Google but not the crawlers that run no JavaScript. If you want any other crawler to have a chance of reading it, put the JSON-LD in the HTML your server sends. JavaScript and AI crawlers explains how to check.

Markup worth adding

Here is a minimal Organization block for a home page, using properties from Google's documentation. Every value must match what the page shows.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Organization",
  "name": "Acme",
  "url": "https://www.example.com",
  "logo": "https://www.example.com/images/logo.png",
  "description": "Acme is an invoicing app for freelancers.",
  "sameAs": [
    "https://www.linkedin.com/company/acme-example",
    "https://github.com/acme-example"
  ]
}
</script>

The description names the category next to the product, the same habit that gives engines less room to misread a name. Google recommends placing Organization markup on the home page or on one page that describes the organization, and says it need not go on every page.

Rules that keep markup from backfiring

• Mark up only what readers can see. Google's guidelines say markup must be a true representation of the page and must not describe hidden content.
• Do not misrepresent who you are. The same guidelines bar using markup to misrepresent ownership, affiliation or primary purpose.
• Leave review stars off your own organization. Pages using Organization or LocalBusiness markup are not eligible for review stars when the business controls the reviews about itself.
• Do not import ratings. Google says not to aggregate reviews or ratings from other websites. Since July 2026 its guidelines also bar fake reviews and incentivized reviews that are not disclosed.

Breaking these can bring a structured data manual action, which removes the page's eligibility for rich results.

How to test it

1. Confirm the JSON-LD is in the HTML your server sends:

   curl -s https://www.example.com/ | grep -c "application/ld+json"
   

   A count of 0 means crawlers that run no JavaScript cannot see it.

2. Print the block and compare every value with the visible page:

   curl -s https://www.example.com/ | grep -A 15 "application/ld+json"
   

3. Run the page through Google's Rich Results Test, then check it with URL Inspection in Search Console.

4. To judge the effect, follow Google's own advice: compare a few pages before and after, using several months of Search Console data.

What to do first

• Add Organization markup, with sameAs links to your official profiles, to your home or About page.
• Move any JSON-LD injected by a tag manager into the server-rendered HTML.
• Stop adding FAQ markup in the hope of FAQ rich results.
• Check every marked-up value against the visible page.
• Read how to structure a page for AI answers, because the visible page matters more than the markup.