Schema Markup Validator — Check Your JSON-LD Structured Data

A schema markup validator fetches a page, extracts its JSON-LD structured data and checks it against schema.org type definitions — whether required properties are present, whether entities are linked rather than duplicated, and whether values are the right shape. This one also scores completeness 0-100 and gives the fix for each finding. It validates markup, not the truth of what the markup claims.

JSON-LD is a JSON-based format for embedding structured data in a page, placed inside a script tag with type "application/ld+json". It is the format Google recommends, because it sits in one block separate from the visible HTML rather than being woven through the markup like microdata or RDFa.

Google’s Rich Results Test answers one question: is this page eligible for a Google rich result. This validator answers a broader one: is the entity graph complete and internally consistent — publisher identity, sameAs links, @id addressability, authors as real entities. Those matter for machine-readability whether or not a rich result exists. Use both; they check different things.

Google has stated that structured data is not a ranking factor in itself. What it does is make the page unambiguous to a parser and make it eligible for rich results, which can change how the listing looks and how often it is clicked. Treat any claim of a direct ranking boost from schema as unsupported.

No. Google is explicit that valid structured data makes a page eligible for a rich result but does not guarantee one. Whether a rich result appears depends on the query, the page’s quality and Google’s own judgement. Eligibility is the part you control; display is not.

No. Google removed the FAQ rich result from Search on 7 May 2026 and retired the documentation on 15 June 2026. FAQPage remains valid schema.org, leaving the markup in place causes no problem, and it still exposes clean question-answer pairs to any parser — but it will not produce a search feature. This validator reports it as a note, not a defect.

HowTo no longer produces a rich result on any Google surface — it was removed from mobile and desktop in 2023. The type is still valid schema.org and still describes a procedure in machine-readable form, so it is not an error to keep, but expect no search feature from it.

sameAs links your entity to external pages that describe the same entity — an official LinkedIn company page, a GitHub organisation, a Crunchbase or Wikidata entry. It is the property that turns a name and a URL into something a parser can corroborate against sources it already knows. It is the single most useful property for entity disambiguation.

@id gives an entity a stable identifier so it can be referenced instead of repeated. Declare your Organization once with "@id": "https://example.com/#organization", then write "publisher": {"@id": "https://example.com/#organization"} everywhere else. Without it, the same company described on 200 pages can read as 200 separate objects.

An object. A string names an author; an object identifies one. Person with name, url, jobTitle and sameAs can be resolved to a real, checkable individual, which is what an expertise signal requires. A bare string carries nothing beyond the characters in it.

There is no limit, and multiple blocks are valid. What matters is that entities across them are linked by @id rather than repeated in slightly different forms. Many sites use one block per concern — Organization, WebSite, the page entity, breadcrumbs — which is fine as long as the @id references tie them together.

No. It reads the server-rendered HTML, so structured data injected later by client-side JavaScript or a tag manager is invisible to it — as it is to any parser that does not execute JavaScript. If you inject markup that way, verify it separately in a tool that renders the page.

It detects them and tells you they are present, but it validates JSON-LD only. JSON-LD is the format Google recommends and the one that is practical to check reliably. Mixing formats on one page is allowed but harder to keep consistent.

Invalid markup is usually ignored rather than penalised — a block that does not parse simply counts for nothing. What can cause harm is markup that misrepresents the page: ratings, prices or reviews that a visitor cannot see are a structured data policy violation and can attract a manual action.

No major AI search vendor has published structured data as a retrieval or citation input, so any specific claim about it is unverified. What can be said honestly: structured data states who published a page, who wrote it and what it is about in an unambiguous machine-readable form, which removes guesswork for any consumer that chooses to read it. That is a reason to get it right — not evidence of a citation lift.