JSON-LD

JSON-LD is placed inside a <script type="application/ld+json"> tag, completely separate from the HTML markup. This decoupling means developers can add, modify, or remove structured data without touching the DOM or risking layout shifts. Unlike Microdata or RDFa, which inline attributes directly into HTML elements, JSON-LD keeps data cleanly isolated.

• No interference with CSS or JavaScript rendering
• Easier to inject via tag managers or server-side logic
• Simplifies debugging and validation

Every JSON-LD document begins with an @context key that maps short property names to full IRI (Internationalized Resource Identifier) vocabularies. The most common context is https://schema.org, which tells consuming applications that the properties follow Schema.org definitions.

• @context resolves ambiguity across vocabularies
• Enables mixing multiple schemas in one block
• Supports custom extensions for proprietary data

The @type property explicitly declares what kind of entity is being described—such as Organization, Product, Event, or Article. This classification is critical for search engines to trigger rich results like knowledge panels, carousels, and recipe cards.

• Google recognizes over 800 Schema.org types
• Nested types enable complex entity relationships
• Incorrect typing can disqualify a page from rich results

The @id property assigns a unique IRI-based identifier to each entity, enabling cross-referencing within and across documents. This transforms isolated data points into a connected knowledge graph that search engines can traverse.

• Link an Organization to its WebSite and Person entities
• Reduces redundancy by referencing entities instead of duplicating
• Foundational for building a semantic web of linked data

JSON-LD supports two syntax modes: a flattened node-based approach using @graph arrays, and a nested approach where entities are embedded directly within parent objects. Nesting mirrors the natural hierarchy of data and is the most common pattern for web pages.

• @graph is preferred for complex, multi-entity datasets
• Nested syntax is more readable for simple page-level markup
• Both produce identical RDF triples when parsed

Google explicitly recommends JSON-LD over Microdata and RDFa for Schema.org markup. Their documentation states that JSON-LD is the preferred format for all search features, including rich results, merchant listings, and the knowledge graph.

• Easier for Googlebot to parse and validate
• Supported across all Google rich result types
• Backed by W3C as an official web standard