Home  /  Blog  /  Search & AI Visibility  /  What Structured Data Is Actually For (and What It Isn't)
Search & AI Visibility

What Structured Data Is Actually For (and What It Isn't)

Structured data (JSON-LD / schema.org) tells machines what your page is about and makes it eligible for rich results. That's real and worth doing. What it isn't: a proven lever for AI citations — or a place to fake ratings.

SPOKETarget: structured data seoPublished 2026-06-19

What is structured data?

Structured data is machine-readable markup — almost always the schema.org vocabulary expressed as JSON-LD — that states what a page's content is: an Article, an Organization, a Person, a Product, a BreadcrumbList. Instead of leaving a machine to infer meaning from prose, you declare it.

It sits in the page's head or body as a small JSON block. It changes nothing a human sees — its entire audience is software.

What it actually does

Two proven jobs. First, eligibility for rich results: breadcrumbs, article treatments, and other SERP features only appear when the matching, accurate schema is present (presence makes you eligible — it doesn't guarantee display). Second, entity understanding: it disambiguates your organization and author and helps connect them in the knowledge graph.

This estate ships Organization, Person, BlogPosting, and BreadcrumbList JSON-LD — generated from canonical data so the markup can't drift from what's on the page. Those are the real, defensible uses.

What it does NOT do

It is not a proven AI-citation lever. There's no controlled evidence that adding schema lifts your odds of being cited in AI Overviews, ChatGPT, or Perplexity — citation tracks answer quality, not markup (see the cornerstone). And it actively backfires when misused: AggregateRating or Review schema on pages that aren't eligible for it is a documented demotion flag.

So the failure mode isn't too little schema — it's inaccurate schema bolted on to fake signals.

Which types are worth adding — and how do you test them?

Lead with the types that earn rich results or anchor your entities: Organization and Person (who you are, for the knowledge graph), BreadcrumbList (navigation in the result), Article/BlogPosting (article treatments), and Product, Recipe, or FAQPage only where the page genuinely qualifies. Skip speculative markup with no eligible display.

Then test it: Google's Rich Results Test shows which rich results a URL is eligible for, and the Schema Markup Validator catches syntax errors. Validating before you ship is how you stay clear of the inaccurate-schema failure mode — the markup either earns a real feature or it shouldn't be on the page.

How to use it well

Mark up what is genuinely on the page, accurately, using only eligible types; keep the markup in sync with the visible content; and never assert ratings, reviews, or facts the page doesn't actually carry. Generating it from the same source as the page — as we do — is the cleanest way to guarantee it never drifts into a false claim.

Accurate schema for its real jobs, paired with genuinely good content, is part of our Search & AI visibility practice — and it pairs with shipping llms.txt as hygiene rather than as a lever.

The teardown, in your inbox

One system at a time.

Email only. Mechanism teardowns when they ship — not on a schedule, no listicles. Unsubscribe anytime.