Structured data: what it is and how it works in search

Open the page source of a recipe, product or news page that shows star ratings, cook times or extra images in Google's results, and you'll find a block of code the browser never displays to anyone. That block is structured data: markup written for search engines, not for readers.

Structured data is a standardized format for labeling the content on a page so a search engine can tell what that content means. It's added as in-page code, usually JSON-LD, spelling out specific details such as a recipe's ingredients and cook time, or an article's author and publish date.

How structured data works

Structured data has to describe the content of the exact page it sits on. Google's documentation is direct about the two ways people get this wrong: don't build a blank page just to hold structured data, and don't describe something that isn't actually visible to a reader on that page, even when the information is true.

Most of the vocabulary for this markup comes from schema.org, an open standard, but Google Search Central's own documentation decides how Google Search actually uses it. Schema.org defines more properties than Google requires. Those extras can still help other search engines or tools, but Google's documentation is what determines whether a given property does anything inside Google Search.

Google accepts three formats for structured data, and none of them outranks the others. JSON-LD, Microdata and RDFa are, in Google's words, equally fine for Google as long as the code is valid and built the way that particular feature's documentation specifies. JSON-LD is the format Google recommends, mostly because it lives in its own script tag rather than being woven into the visible HTML, which makes nested data easier to write and lets Google keep reading it even when it's injected into the page by JavaScript. Microdata nests data inside the HTML itself, using tag attributes, usually in the body of the page, though it can appear in the head too. RDFa works the same way, through HTML tag attributes, and commonly shows up in both the head and the body.

A minimal example for a blog post, using the Article vocabulary, looks like this:

{
  "@context": "https://schema.org",
  "@type": "BlogPosting",
  "headline": "How to Price a SaaS Free Trial",
  "image": ["https://example.com/photos/16x9/photo.jpg"],
  "datePublished": "2026-03-01T08:00:00-05:00",
  "dateModified": "2026-03-04T10:00:00-05:00",
  "author": [{
    "@type": "Person",
    "name": "Ada Lovelace",
    "url": "https://example.com/authors/ada-lovelace"
  }]
}

Once markup like this is valid, Google can match the page to a more specific search feature. A correctly marked-up recipe page, for instance, can be found by ingredient or cook time, not just by its title. Some SEO tools fold a structured data check into a wider audit: SEOAgent's AI visibility check, for example, includes JSON-LD in its quotability checklist, alongside robots.txt, title, meta description, H1 and sitemap.

To check your own markup, run it through the Rich Results Test while you're building it. After the page is live, use the URL Inspection tool to confirm Google actually found the structured data on that URL, and keep watching the Rich result status reports afterward, since a templating change or a serving issue can quietly break markup that used to validate fine. For more guidance sourced the same way, straight from Google's own documentation, see our rundown of 15 SEO tips from Google's own documentation.

Examples

Google documents every rich result type it currently supports on one page, organized into categories such as ecommerce, jobs, and food and drink. A handful come up most often for a SaaS site, a startup blog, or an agency's own pages:

• Article (using the Article, NewsArticle or BlogPosting type; Google's guidance was last updated 8 September 2026): shows a cleaner headline, larger images, and publish or update dates for a news, sports or blog post. There's no required property, just the ones that apply to your content, among them headline, image, author, datePublished and dateModified. If more than one person wrote the piece, list each author in a separate author entry instead of combining names into one field.
• Organization: placed once, on the home page or an About page, it gives Google administrative detail about who runs the site, such as a logo, address and contactPoint. It can influence which logo shows up in a knowledge panel.
• Product: price, availability and review ratings for an item you sell.
• Review snippet: a short excerpt or rating pulled from a review, usable on a book, recipe, movie, product, software app or local business page.
• Breadcrumb: shows a page's position in your site's hierarchy inside the search result itself.
• Local business: details such as open hours, ratings and directions, shown in Google's knowledge panel, with actions to book an appointment or order.
• Event: an interactive result listing an organized event, such as a concert or an art festival, with its time and place attached.
• Job posting: an interactive result job seekers browse directly in Google, which can show a company's logo, reviews, ratings and job details.

That's a fraction of the full list, which also covers recipes, course listings, datasets, software apps and more, each with its own documented properties.

When to use structured data (and when not)

Google's own case studies make a concrete argument for bothering with this. Rotten Tomatoes added structured data to 100,000 pages and measured a 25% higher click-through rate compared to pages without it. The Food Network converted 80% of its pages to enable search features and saw a 35% increase in visits. Rakuten found that users spend 1.5 times longer on pages with structured data than on pages without it, with 3.6 times higher interaction on AMP pages carrying search features compared to AMP pages that don't carry them. Nestlé measured an 82% higher click-through rate on pages that displayed as rich results compared to pages that didn't.

None of that happens just by adding code. Google's general structured data guidelines, last updated 10 July 2026, split into two kinds of requirements. The technical ones are things a validator can catch: use one of the three supported formats, and don't block the page from Googlebot with robots.txt, a noindex tag, or any other access control. The quality ones are softer and won't show up in an automated test: the markup has to truthfully represent content a reader can actually see on the page, it has to include every required property for that type, since anything missing a required property isn't eligible, and it has to sit on the page it describes rather than somewhere else. Google's own examples of irrelevant markup include a sports live-streaming site labeling its broadcasts as local events, and a woodworking site labeling its instructions as recipes.

Reviews and ratings carry their own warning: Google says users prefer recipes and products with genuine reviews and star ratings from actual users, and that ratings which aren't genuine can lead to a manual action. A structured data manual action removes a page's eligibility to appear as a rich result. According to Google, it does not change how that page ranks in ordinary web search. The Manual Actions report in Search Console shows whether this has happened on your site. Even clean, valid markup is not a guarantee of anything: Google states plainly that it does not guarantee that features built on structured data will actually show up in search results.

One of the easiest ways to lose eligibility by accident is indexability. A page that's set to noindex, or blocked in robots.txt, gets no credit for any structured data sitting on it, since Google's access guideline requires these pages to stay fully reachable by Googlebot. A site audit that checks indexability and noindex status across every page, which is part of what SEOAgent's audit covers, catches that kind of accident before it quietly costs you a rich result.

If you want to know whether any of this is worth the effort on your own site, rather than taking Google's case studies on faith, Google recommends testing it directly. Pick a few pages that already have several months of history in Search Console, that won't be thrown off by the time of year, and that get read often enough to produce a meaningful signal. Add the structured data, confirm it's valid and that Google has found it using the URL Inspection tool, then watch the Performance report, filtered by that URL, for a few months before drawing a conclusion.

Once your markup validates and your most important pages are eligible for a rich result, the next thing worth checking is the rest of your site's technical health, which is exactly what our guide to what to ask for and expect from an SEO audit service covers.

Key takeaways

• Structured data is in-page code, most often JSON-LD though Microdata and RDFa work too, that labels a page's content so Google can understand it well enough to show a rich result.
• Google treats the three formats as equally valid; what matters is that the markup is correct and built the way that specific feature's documentation describes.
• Rich results aren't automatic: a type needs every required property it has, plus content that's genuine, complete and actually visible on the page.
• A structured data manual action removes a page's eligibility for rich results, but Google says it does not change how that page ranks in ordinary web search.
• Google's own case studies, like Rotten Tomatoes' 25% click-through increase across 100,000 pages, show rich results can lift engagement, but testing your own pages before and after is the only way to know the effect on your site.

FAQ

Does structured data improve rankings?

Google's documentation does not describe structured data as a ranking factor. What it does say is that a structured data violation triggers a manual action that removes a page's eligibility to appear as a rich result, and that this specifically does not affect how the page ranks in ordinary web search. The benefit Google documents is about click-through rather than ranking position: Nestlé, for instance, measured an 82% higher click-through rate on pages that displayed as rich results compared to pages that didn't.

Which structured data format should I use, JSON-LD, Microdata or RDFa?

Google accepts all three as equally valid, provided the markup is valid and built the way that specific feature's documentation describes. Google recommends JSON-LD in particular because the code sits in its own script tag, separate from the visible page content, which makes nested data easier to write and lets the markup be added or updated dynamically through JavaScript without touching the visible HTML.

How do I check that my structured data is actually working?

Run your markup through the Rich Results Test while you're building it, since it validates the code and can preview how a feature might look. Once the page is live, use the URL Inspection tool to confirm Google has actually found the structured data on that specific URL, and keep checking the Rich result status reports afterward, because a templating change or a serving issue can break markup that used to be valid.

Does every page on my site need structured data?

No. Google's own guidance depends on the type: Organization markup is recommended once, on the home page or a single About page, rather than on every page, while markup like Article belongs only on the specific page it describes. The general rule Google gives is to put structured data on the page it actually describes, and to repeat the same markup across every duplicate of that content if duplicates exist, rather than placing it only on the canonical version.