Guide
Structured data for AI search
Structured data states what a page is in a form a machine can read without inference. For AI search it helps a system decide what a page covers before deciding whether to use it, and it ties your brand to the entity a search engine already knows about.
Last updated
The types that earn their place
Most sites need a small number of types, applied consistently, rather than an exhaustive graph.
- Organization with sameAs linking your real profiles — this is how an engine ties your domain to the company it knows from elsewhere.
- Article with a named author and a genuine dateModified, for anything editorial.
- FAQPage for questions that are actually rendered on the page.
- Product and Offer for pricing, with the real numbers.
- DefinedTerm for glossary entries, which is the type that says 'this page defines one thing'.
- BreadcrumbList everywhere, so the site's shape is explicit.
The rule that keeps you safe
Mark up only what is genuinely visible on the page. Describing content a reader cannot see is against the guidelines and risks a manual action, and it is the single most common way sites get this wrong.
In practice that means the same array should feed both the rendered FAQ and the FAQPage node, so they cannot drift apart.
What it does not do
Structured data is not a ranking lever and it will not make a thin page citable. It removes ambiguity about what a page is; it does not improve what the page says.
It is worth doing because it is cheap, durable and occasionally decisive — not because it is transformative.
Questions
Do AI assistants read schema markup?
They lean on it to work out what a page is before deciding whether to use it, alongside the visible content. It is a supporting signal rather than the deciding one.
JSON-LD or microdata?
JSON-LD, in the page head. It sits apart from the visible markup, which makes it far easier to maintain and much harder to break when the layout changes.
Is more markup always better?
No. A small, accurate graph beats a sprawling one, and every node you add is another thing that can go stale and contradict the page.
Read next
Content that gets quoted
Retrieval selects passages, not documents. How to structure a page so an assistant can lift a complete answer from it.
How AI Overviews choose what to cite
What decides which pages appear in a Google AI Overview, how it relates to ordinary ranking, and what to do about it.
llms.txt, explained
What llms.txt is, what belongs in it, how it differs from robots.txt, and an honest account of what it does and does not do today.
How this works by industry
The sources and buyer questions differ by market.
Ready to Dominate AI Search?
Get started now
Allswap