What Is Structured Data? Its Role in AI Search and SEO, and How to Write It

“Apparently you need structured data to appear in AI search.” Hearing this, you may wonder whether to add something to your site.

This article starts with what structured data is, then covers its role in SEO, whether AI search needs it, how to choose types, how to write it and what to check afterward. It uses one product as a running example so you can decide without specialist knowledge. Information was checked on September 23, 2026.

What is structured data?

Structured data is markup that tells machines such as search engines what a page contains, using shared terms and a defined format. Google describes it as a standardized format for providing information about a page and classifying its content.[1]

People looking at a product page immediately see “white mug,” “¥3,300” and “in stock.” For a machine, it is harder to tell which number is the price and which is the product code.

So you add labeled information: “this is a product, its name is this, its price is this.” Most labels come from a shared vocabulary called Schema.org.[1]

A recipe page, for example, can describe ingredients, cooking time and calories as separate items.[1] A product page can describe the name, price and availability.

The SEO effect of structured data

Structured data can make a page eligible for enhanced search displays with prices or ratings, called rich results. It guarantees neither display nor rankings. Google states that even correct markup that passes testing may not appear in results.[2]

Reasons include Google deciding a plain result suits that search better, or the markup not reflecting the page’s main content.[2]

A manual action for markup that violates the guidelines removes eligibility for rich results. Google says this does not affect web search rankings.[2] It helps to think of structured data not as a ranking mechanism but as information that makes a page a candidate for certain displays.

To measure its effect, Google suggests comparing a few pages before and after adding it over several months.[1] Avoid pages whose traffic changes by season.

Is structured data necessary for AI search?

Structured data is not required to appear in Google’s AI Overviews or AI Mode. Google says there are no additional technical requirements and no special schema.org structured data is needed.[3]

Google does list, as one basic SEO practice, making sure structured data matches the visible text on the page.[3] If you use it, keep it accurate.

For ChatGPT search, OpenAI recommends allowing its search crawler, OAI-SearchBot, in robots.txt. That document does not mention whether structured data is needed.[4] Not being mentioned is not proof that it is unnecessary either, so avoid generalizing about other AI services.

If you are wondering how this differs from AI-specific files, see “Do You Need llms.txt? What It Can Do and What to Check Before Adding It.”

Types of structured data and how to choose

First decide what kind of page you have, then choose the matching type. Google publishes a list of the types it supports.[5]

Page type	Example type	Notes
Page where a product can be bought	Product	Price and availability[5]
Company overview or homepage	Organization	Name, logo, address[5]
Blog or news article	Article	Headline, author, dates[5]
Store or office information	Local business	Opening hours, address[5]
Page where users answer a question	Q&A	Different from a company-written FAQ[5]
Links showing page position	Breadcrumb	Site hierarchy[5]

As of September 23, 2026, FAQ and HowTo are not on this list.[5] Company details are covered in more depth in “What Is Organization Structured Data? Aligning Company Details and Official Links.”

You do not need the same markup on every page. Place structured data on the page that contains the information.[1]

Writing structured data: JSON-LD basics

There are three formats, and Google recommends JSON-LD in most cases. JSON-LD sits apart from the visible text, which makes it easier to maintain.[1] The others are Microdata and RDFa; any of them is fine for Google if written correctly.[1]

JSON-LD has three basics:

• @context: where the shared vocabulary lives, usually https://schema.org.
• @type: what the information is about, such as Product.
• Each property: the same value visible on the page.

Here is a short illustrative example. The values and URL are fictional, and it is not a complete version meeting every merchant listing requirement.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Product",
  "name": "White mug",
  "image": "https://example.com/mug.jpg",
  "offers": {
    "@type": "Offer",
    "price": 3300,
    "priceCurrency": "JPY"
  }
}
</script>

If you use WordPress, Shopify or a similar service and cannot edit HTML directly, you can use that service’s settings or an integrated plugin.[1]

Do not describe information that is not visible on the page. Even accurate information should not be marked up if readers cannot see it.[1]

Checks after adding structured data

After adding it, check the format with tools and the content with your own eyes. These are separate checks.

1. Use the Rich Results Test to check for critical format errors.[1]
2. After publishing, use URL Inspection in Search Console to confirm Google can find the page and its data.[1]
3. Have a person compare the values shown on the page with the values in the markup, because automated tests miss some errors.[2]
4. Keep watching Search Console’s rich result reports for errors caused by template changes and similar updates.[1]

Start by choosing which type in the table your main pages fall under. You can begin by accurately conveying only the information visible on that page.