Structured Data

Structured data is machine-readable information added to a page that states what the page means — its type, its parts, and how they relate to each other — in a standardized vocabulary. Readers never see it; software reads it instead of inferring the same facts from prose.

Nine pages in this library explain a specific piece of markup. This one defines the umbrella they all sit under, because "structured data," "schema," and "JSON-LD" get used as if they were one thing, and the difference between them decides which claims about markup are even coherent.

What is the difference between structured data, schema.org, and JSON-LD?

They are three separate layers: a concept, a vocabulary, and a format. Collapsing them is what lets a vendor pitch "we'll add schema" without saying what is being described or why.

Google documents support for all three formats and recommends JSON-LD, which keeps the description in one self-contained block rather than threading attributes through the visible HTML. The format layer gets its own definition at JSON-LD.

What does structured data actually do?

Two documented jobs, both inside Google's own pipeline. Google's structured-data introduction describes markup as helping Google understand what a page's content is about, and as the mechanism that makes a page eligible for rich results — the enhanced listings Google grants for specific documented types (Google, structured data introduction).

A third job is real but less formally documented: entity disambiguation. Markup that names the publisher of a page, with a stable identifier reused across every page, gives any system resolving who is asserting this a machine-readable answer instead of a name-matching guess. That is the practical use we get most from, and the implementation is in Organization schema as your entity home.

Note what is not on the list: nothing about answer engines, nothing about being quoted, nothing about visibility outside Google's own result features.

Does structured data get you cited by AI answer engines?

No evidence says it does, and the one controlled test says it does not. Ahrefs tracked 1,885 pages that added schema between August 2025 and March 2026 against roughly 4,000 matched control pages and reported no major citation uplift on any platform (Ahrefs, May 2026). That is the null result any sentence connecting markup to citations has to carry.

The correlation that fuels the opposite claim is real and misleading at the same time: sites that add markup also tend to build crawlable structures, fast pages, and linkable authority, so cited pages disproportionately have schema without schema being the cause. The full accounting — including Google's own documented position on markup and its AI features — is in what the evidence shows about schema and AI citations.

We still ship markup on every page we publish, for the two documented jobs above and not for citations. Stating that against our own interest: we sell builds that include structured data, and we will not sell it as an AI-visibility lever, because the controlled test of record — Ahrefs' 1,885-page study, not one of ours — found nothing.

What rules govern a structured-data block?

One rule dominates the rest: markup must reflect the visible content of the page. Google's structured-data guidelines make the correspondence a requirement, which means a price, rating, author, or date that exists only in the markup is a violation rather than a clever shortcut. Everything else follows from it.

That last one is a production concern rather than a spec concern, and it is where most large sites actually fail.

Is structured data worth adding?

Yes, priced honestly: it is cheap once templated and it serves two documented jobs, which is exactly the level of investment the evidence supports. Generated from one source of truth and checked in the build, markup costs close to nothing per page after the first.

Against our own interest, the case for skipping it: if your stack makes markup expensive — a plugin tangle, hand-authored blocks per page, an agency line item billed monthly — cut the markup before you cut the writing. A page with clean, self-contained visible answers and no JSON-LD can be quoted anywhere text is quoted; a page with immaculate markup and mushy prose has optimized the layer nothing quotes. Where markup fits in the wider system, with each claim at its evidence tier, is mapped in our generative engine optimization guide.