Paste a JSON-LD block or a whole page and get five passes of checks, sorted into errors, warnings and info. It works on unpublished code, right in your browser.
Your files are processed in your browser, on your own device. Nothing is uploaded to a server.
This schema validator checks JSON-LD the fast way. Paste a single markup block, or paste a whole page and it pulls out every application/ld+json block it finds. It then runs five passes over the code and sorts what it finds into errors, warnings and info, so you fix the breaking problems first.
Errors, warnings and info are kept apart on purpose. An error means the block breaks or lacks something its type demands. A warning means it works and could be stronger. Info is context, and the clearest case is the note attached to FAQ and HowTo blocks: their rich results are retired at Google for most sites, so valid markup there still earns no Google feature.
Three lines are worth keeping straight, and this page prints all three. Valid is not the same as eligible, and eligible is not the same as shown. A clean run here proves the code parses and carries what its type asks for. It does not prove Google will grant a rich result.
Google Rich Results Test is the final word on Google eligibility, because it renders the page with Googlebot and sees what a crawler sees. This validator is the fast first pass. It runs on code that is not live yet, behind logins and on staging pages a URL test cannot reach.
Two formats get a partial answer. If pasted HTML carries Microdata or RDFa, the validator detects them and reports them as present. It does not validate them, and it says so in the findings instead of staying quiet.
Most markup problems are found before launch, and URL tests cannot help there. Staging sites sit behind logins. Templates spit out markup no single page shows in full. A plugin update can change output across a whole site overnight. In each case the code is what you have, and pasting the code is the test.
It extracts every application/ld+json block from the HTML and checks each one. The rest of the page is ignored, apart from a note if Microdata or RDFa shows up in the markup.
The finding names the line and column in plain language. Usual suspects at that spot are a missing comma, a trailing comma after the last item, or a quote that was never closed.
Because it arrived as a string with a currency symbol inside it. Prices in schema are plain numbers, like 89.00, with the currency carried by its own three letter code property.
Not promised. Clean means valid. Eligibility is a second bar, and showing is a call Google makes alone. Run the live page through the Rich Results Test for the eligibility answer.
It detects Microdata and RDFa in pasted HTML and reports them as present. It does not validate them. JSON-LD is the format it checks in depth.
No. The checks run in your browser, on your device. That is why the tool works on unpublished code in the first place.
Same rules as this one: free, in your browser, with nothing uploaded.
See if a page repeats a phrase a healthy amount.
Title and description tags, written and copied clean.
A correct robots file without the guesswork.
A fast on-page check with a clear fix list.
Read the title, description and social tags of any page.
The tags that decide how a shared link looks on social media.
See the social preview tags a page already carries.
Card tags so a link posted on X shows its picture and title.