Free Schema Markup Validator

You get a list of every JSON-LD block on the page, each item in it, and what is wrong with each item, in plain words. Errors are things Google's documentation says must be there or must be written a certain way; warnings are things it recommends, or that search engines will ignore.

• JSON syntax errors, with the line and column where the code stops being valid JSON: a trailing comma, a missing comma, single quotes, a line break inside a string.
• A missing or wrong @context, and items with no @type.
• Each @graph expanded into its items, and @id references followed to the node they name on the same page.
• Missing required properties, and the "one of" rules, such as a Product needing an offer, a review or a rating.
• Value formats: ISO 8601 dates, numeric prices, ISO 4217 currency codes, full URLs and image URLs.
• Types and properties that schema.org does not define, which search engines skip. Typos like adress or a lowercase product show up here.

URL mode reads the page's HTML once, up to 3 MB, the way a crawler does before running any JavaScript. Paste mode takes up to 60 KB of JSON-LD or HTML and fetches nothing. One check covers up to 500 blocks.

Google's Structured Data Testing Tool is gone. Google announced in July 2020 that it would retire it in favour of the Rich Results Test, and later handed the general-purpose validation part to schema.org, which now runs it at validator.schema.org. Many guides still link to the old tool, which is why the name keeps coming up.

That split left two tools that answer different questions. The Rich Results Test asks whether a page can get a Google rich result, and renders the page with JavaScript first. The schema.org validator checks markup against the whole schema.org vocabulary, with no opinion on Google's features.

This validator sits between them. It checks the vocabulary like the schema.org validator, and checks the types behind Google's most used rich results against Google's published required and recommended properties. It is not a Google tool, and meeting Google's documented requirements makes a page eligible for a rich result; it does not guarantee one. Google decides what to show.

The table lists the requirements this checker applies, taken from Google Search Central's structured data documentation and last compared with it on 2026-10-25. Subtypes use the nearest type below: a Dentist is checked as a LocalBusiness, a NewsArticle as an Article.

Recommended properties come next. They never block eligibility, but Google uses them where it can, for example a product's price and availability in its snippet. Some recommendations apply only in some cases, such as a menu for a restaurant, and the checker lists those as notes rather than warnings.

No. Google stopped showing How-to rich results in September 2023, and stopped showing FAQ rich results on May 7, 2026, after limiting them to well-known government and health sites in 2023. Both documentation pages have since been removed.

The checker still validates FAQPage and HowTo markup against the last published rules, because the markup is valid schema.org that other search engines and AI answer tools can read. It reports these items as valid but not eligible, so you are never told to expect a rich result that no longer exists.

The checker finds microdata (itemscope and itemtype attributes) and RDFa (typeof attributes) and lists the types it finds, but validates JSON-LD only. Google reads all three formats, and recommends JSON-LD because it sits in one script block instead of being spread across the page's HTML, which makes it easier to keep correct as templates change.

If a page uses microdata, check it in Google's Rich Results Test, or move it to JSON-LD and check it here.

A check that could not read everything says so and never reports a clean page. That happens when a site blocks or rate-limits our checker, when it is too slow to answer within the time limit, when the page is larger than 3 MB, or when it holds more blocks than one check covers.

A site that blocks checkers may still serve Googlebot normally, so a refusal says nothing about what Google sees. Paste the page's HTML instead, or use Search Console's URL Inspection, which shows the page as Google rendered it.

Markup that JavaScript adds after the page loads is not in the HTML this checker reads. Google renders JavaScript, so it may see markup that a URL check here does not; when a page looks like an empty app shell, the result says that too.