Glossary · AI Search & Prompting

Structured Data Formats

Structured data formats organize values and relationships under explicit syntax and rules so systems can exchange and interpret information consistently.
Back to glossary

What are structured data formats?

Structured data formats are standardized ways to represent fields, values, types, and relationships so software can parse them consistently. On web pages, common structured-data syntaxes include JSON-LD, Microdata, and RDFa. These syntaxes can express vocabularies such as Schema.org. In broader data work, JSON, XML, CSV, relational tables, and other formats may carry structured records.

Format and vocabulary solve different problems. JSON-LD describes how the data is written. Schema.org supplies terms such as Organization, Product, name, and offers.

How the web formats differ

JSON-LD places linked data in a script block, separate from the visible HTML. It is often easier to generate and maintain from CMS fields. Microdata adds attributes directly to HTML elements. RDFa also uses HTML attributes and supports RDF relationships. All three can represent valid structured data when implemented accurately.

Search feature support and platform guidance may favor one syntax for a particular use. Technical validity alone does not guarantee a rich result.

How to choose a format

Start with the consuming system, supported vocabulary, update workflow, and source of truth. Prefer a format the team can keep synchronized with visible content. Validate syntax, required properties, URLs, identifiers, dates, and relationships. Avoid creating several independent copies of the same price, author, or event data.

Document the template owner and retest after CMS or page changes.

Example

An event page stores name, start date, location, image, status, and ticket URL in the CMS. The template renders those values for people and generates Event markup in JSON-LD from the same fields. Editors update one source. Automated checks flag missing dates and invalid URLs before publication. The format makes the facts readable to machines without becoming a hidden alternate version of the event.

Format versus data model

A file can be syntactically structured and still have a weak model. JSON with inconsistent names, missing identifiers, and ambiguous relationships is hard to use. A sound data model defines entities, properties, types, constraints, and relationships before choosing the transport or markup format. On the web, the visible page, CMS fields, vocabulary, and JSON-LD should express one maintained model rather than four loosely connected copies.

Set up once

See what Surface can do for your team.

Get a walkthrough