Structured data is machine-readable metadata embedded in a web page that describes the entities, relationships, and facts the page is about. It does not replace the visible HTML content — it accompanies it as a structured mirror that search and AI engines can extract reliably without natural-language parsing.
The dominant vocabulary is Schema.org, a shared hierarchy of types (Article, Product, FAQPage, HowTo, Organization, WebSite, and more) used by Google, Bing, ChatGPT, Perplexity, and most major engines. The recommended serialization format is JSON-LD embedded in a script tag — easier to author and easier to keep in sync with the rendered DOM than Microdata or RDFa.
For SEO, structured data enables rich results: review stars in SERPs, recipe cards, FAQ accordions, event listings, breadcrumbs, and more. Rich results dramatically lift click-through rates even when ranking stays the same.
For AEO, structured data lowers the extraction cost for AI engines. A page with FAQPage schema gets its Q&A pairs lifted verbatim; a page with HowTo schema gets its steps extracted as a numbered list; a page with Article schema reliably surfaces author and date. Pages without structured data force the AI engine to do an LLM-style natural-language extraction, which is more expensive and less confident — they get cited less.