Structured data:what it really doesin AI search

What is structured data?

Google defines structured data as a standardised format for providing information about a page and classifying its content. A person sees at a glance that the figure under a product photo is the price and the line in the footer is the company address. A machine has to work that out from the text. Structured data does the work for it: you state explicitly that this is a product with a given name and price, or a business with a given address.

The shared vocabulary for this is schema.org, a collaborative project founded by Google, Microsoft, Yahoo and Yandex and developed today with a wide web community. It defines which types exist, such as Organization, Product or Article, and which properties belong to each. When marketers talk about schema markup, this is what they mean.

Google supports three formats: JSON-LD, Microdata and RDFa. It recommends JSON-LD because the code sits in its own script block, is not interleaved with the visible text and is therefore easier to maintain. For how structured data fits into a wider strategy for AI search, see our article on AI SEO.

What does Google say about structured data and rich results?

Rich results are search listings with extra elements, such as review stars, prices or a breadcrumb trail under the title. Structured data makes them possible but does not guarantee them. Google's general guidelines state plainly that a correctly marked-up page may still not appear as a rich result, even when the Rich Results Test reports no errors. The algorithm decides, based on the query, the device and the location, which presentation it considers best.

All required properties have to be present. Beyond that, Google advises supplying fewer recommended properties that are complete and accurate rather than many that are incomplete or wrong.

Which rich results exist keeps changing. Since 23 January 2025, Google shows the BreadcrumbList trail only in desktop results; on mobile, just the domain appears. Google had already limited FAQ rich results to well-known, authoritative government and health websites in August 2023. Since 7 May 2026 they no longer appear at all, and in June 2026 the Search Console report and Rich Results Test support followed. Existing FAQPage markup does no harm and can stay, but it no longer changes how your page looks in Google.

Why is structured data still worth the effort?

Because it removes ambiguity. Google explains that it uses structured data to understand the content of a page and to gather information about the world, such as the people, books or companies named in the markup. For a business, that means your name, address, logo and website are explicitly attached to one entity instead of being scattered through the text.

Google gets specific with the Organization type. According to its documentation, Organization markup on the home page can help Google understand your company's administrative details and disambiguate it from other organisations. Some properties do exactly that behind the scenes, others influence visible elements such as the logo shown in results or the knowledge panel. The sameAs property points to your profiles on other platforms, and Google notes that it can make general use of sameAs.

This matters most when names overlap. If several companies share a similar name, every unambiguous signal helps attribute statements to the right one. It is no guaranteed lever for AI answers, but a clean foundation that fits with everything else you do. How we review these foundations together is described on our GEO agency page.

One rule always applies: the markup has to match the visible content. Google says so explicitly in its AI features documentation, and its general guidelines name markup that does not represent the main content, or describes content hidden from users, as a reason rich results do not appear.

How do you test your structured data?

Two free tools complement each other. Google's Rich Results Test shows whether a page is eligible for rich results in Google, which types it detects and which required properties are missing. Google recommends it during development and, together with the URL Inspection tool in Search Console, names it as the way to catch most technical errors.

The Schema Markup Validator at validator.schema.org checks any schema.org markup, regardless of whether Google offers a rich result for it. It is based on Google's former Structured Data Testing Tool and is provided by Google as a service for the schema.org community. It extracts JSON-LD, RDFa and Microdata, shows a summary of the data it found and flags syntax errors.

After launch, Search Console takes over: the rich result status reports show when valid pages suddenly start throwing errors, for example after a template update. Whether your business actually shows up in AI answers is something these tools cannot tell you. How to measure that is covered in our article on measuring AI visibility.