Yoast Schema Aggregation
A feature in Yoast SEO that collects a WordPress site’s per-page schema.org graph into one deduplicated response served from a single REST endpoint, aimed at AI systems rather than page-by-page crawlers. T3 — a vendor feature page, promotional, with no measurement behind its central claim. Volatile: supported types and packaging change per release.
What it changes
Yoast already emitted a connected schema graph per page (@id-linked pieces: WebPage, WebSite,
Organization, Person, Article, Product, and — with the matching plugins — Event and
Recipe). Aggregation keeps that output and changes its delivery: instead of a consumer fetching
500 pages and reassembling the graph, it fetches one WordPress REST endpoint and gets the whole thing,
with repeated entities merged — the same author or product mentioned across many pages collapsing to
one node. It’s a toggle, with no re-authoring; existing indexability and privacy settings still apply,
and only public content is exposed.
The developer-facing half is the interesting part: Yoast presents the endpoint as an open API surface and points at Microsoft’s NLWeb as the sort of thing you’d build on it — a conversational interface over the site’s own graph. That reframes site structured data from a search-engine signal into a queryable local knowledge base.
The move worth naming: markup leaves the page
Classic schema markup is embedded in the document it describes; the graph is implicit, spread across
the site, and reconstructed by whoever crawls it. Aggregation makes the graph explicit and
addressable at one URL. That deduplication step is doing knowledge-representation work — deciding
that two mentions are one entity is entity resolution, the problem sameAs and stable URIs exist to
handle in rdf.
So the same trajectory shows up here as in entitymap and in ../search-marketing-wiki’s
llms-txt: a separate machine-facing artifact at a predictable location, published for machines
rather than inferred by them. See structured-data-for-ai for the pattern and the evidence
against it.
The claim, and what backs it
The page’s premise is stated flatly: “Your content cannot be visible in AI if it is not understood
correctly,” with the feature ensuring structured data is “complete, connected, and ready for accurate
interpretation by AI systems.” No named AI system is documented as consuming the endpoint, no
before/after visibility measurement is offered, and the claim is in direct tension with Google’s own
first-party guidance that structured data “isn’t required for generative AI search”
(structured-data-for-ai, and ../search-marketing-wiki’s google-ai-optimization-guide).
Recorded as a vendor position. The engineering — one endpoint, deduplicated entities, standards-based output — is real and describable; the benefit is asserted.
Related
schema-org · structured-data-for-ai · entitymap · merchant-listing-structured-data · rdf · yoast · synthesis