Definition
Structured data is machine-readable markup, usually JSON-LD using the Schema.org vocabulary, that describes a page’s content to engines. It powers Google rich results, but has minimal measured effect on AI citations.
Structured data is code you add to a page to describe its meaning to machines — an article, an FAQ, a product, a breadcrumb. The recommended format is JSON-LD (a <script type="application/ld+json"> block) using the Schema.org vocabulary.
What it actually does
Its proven value is Google rich results: FAQ snippets, article cards, star ratings, breadcrumbs. What it does not meaningfully do is increase AI citations — Ahrefs tracked 1,885 pages adding schema across 2025–2026 and saw no real uplift on AI Overviews, AI Mode or ChatGPT.
Implement it correctly
Inline JSON-LD in the server-rendered HTML. Avoid injecting it after client-side hydration, and avoid microdata on collapsed accordions. Details in metadata LLMs actually read.
Frequently asked questions
Does structured data help with AI citations?
Barely. Controlled measurement shows schema markup has near-zero effect on whether AI engines cite a page. It remains valuable for Google rich results, but it is not a citation lever.
What is the best format for structured data?
JSON-LD, inline in the server-rendered HTML, using Schema.org types. Google recommends JSON-LD over microdata because it is decoupled from visual rendering.