Structured Data

When a person looks at a page, they immediately understand that the heading is a product name, the number is a price and the text below is a company address. For machines that inference is not always easy. Structured data labels the information already visible on the page and turns it into explicit statements such as "this is an organisation, this is its address, this page defines this term".

The shared vocabulary is schema.org, launched in 2011 by Google, Microsoft, Yahoo and Yandex. Of the available formats, JSON-LD, which Google recommends, is the easiest to maintain because it keeps the markup in a separate block from the HTML. Clean structured data supports both how a page appears in search results and how clearly its entities are identified in AEO and GEO work.

• Vocabulary: schema.org types and properties such as Organization, WebPage, Article, Product, BreadcrumbList and DefinedTerm.
• Format: JSON-LD, Microdata or RDFa; new implementations mostly use JSON-LD.
• Entity identifiers: Giving each entity a permanent address with @id and reusing the same identifier across pages.
• Consistency with visible content: Every piece of information in the markup should also be visible on the page, otherwise the data loses credibility.
• Validation: Regular checks with the Rich Results Test and the Schema Markup Validator.

This page is a good example. Its source code contains a JSON-LD block that describes the organisation, the website, this page, the term that is the page's main subject (DefinedTerm), the page's breadcrumb trail (BreadcrumbList) and the frequently asked questions below (FAQPage). The term is linked, with the same identifier, to the set of terms (DefinedTermSet) on the glossary's index page.

On an e-commerce site the same logic is applied to product name, price, availability and reviews; on an events site, to date, venue and ticket information. The rule is always the same: markup describes information visible on the page and does not add new information.

First the page types are identified and the schema.org type for each is decided. JSON-LD is then generated at template level, so hundreds of pages are marked up consistently from a single source. Recurring entities such as the organisation are defined once with a single @id, and other pages refer to that identifier.

Validation uses Google's Rich Results Test and schema.org's Schema Markup Validator. After launch, the enhancement reports in Search Console surface errors and warnings. These checks are built into the publishing process to make sure the markup is updated whenever the content changes.

Structured data is the markup you add; a rich result is the special presentation a search engine may show using that markup, such as stars, prices or breadcrumbs. Markup does not guarantee a rich result: the decision belongs to the search engine and the rules change over time. Since 2023, for example, Google has largely limited FAQ rich results to authoritative government and health websites. Markup still keeps its value for helping content be understood. Implementation is usually handled at template level as part of digital ecosystems work.