Home › Knowledge base › Basics
Structured data for AI systems
JSON-LD per schema.org makes author, organisation, date and page type machine-unambiguous — instead of leaving them to be guessed from the body text.
Why the format matters
From plain running text an AI system can usually tell what a page is about — but not reliably who is responsible for the text, when it was last updated, or whether it is a guide, a product page or a news article. JSON-LD provides these details as an unambiguous data record in the page head, independent of the layout.
JSON-LD instead of Microdata
schema.org supports three syntaxes: Microdata and RDFa in the HTML markup, and JSON-LD as a separate script block in the form "script type application/ld+json". For new projects JSON-LD is the recommended option: the block sits independently of the visible HTML, can be maintained centrally and does not break when the template is rebuilt.
The types with the greatest benefit
For most pages a few types are enough: Organization (once per website, with name, URL, logo and sameAs references to official profiles), WebSite, and depending on the page Article or BlogPosting with author and datePublished. Shops add Product and Offer, guides HowTo, question pages FAQPage.
Linking author and organisation
The biggest lever for source selection is named authorship: author as a Person with a real name, linked via sameAs to an author page or a professional profile, and a publisher entry that points to the same Organization object as the rest of the website. This creates a connected graph instead of isolated snippets.
Common mistakes
Four mistakes come up again and again: invalid JSON (one comma too many and the whole block is ignored), marking up content that is not visible on the page at all, contradictory details across several blocks, and an @type that does not match the actual page content. When in doubt, prefer a few correct entries over many questionable ones.
Adding it and checking it
After adding it, verify with the Schema Markup Validator (validator.schema.org) and Google's Rich Results Test. citeglass reads the JSON-LD types found on your home page and lists them in the report — so you quickly see whether the markup arrives at all and whether organisation, author and date are covered.
Check your own site
citeglass shows in about 30 seconds where your website is unreadable for AI systems — free and without login.
Read on
General information, not legal advice. GEO is a young field — recommendations may change.