JSON-LD

JSON-LD (JavaScript Object Notation for Linked Data) is a W3C standard for writing linked data as JSON. In SEO it is the most common way to add Structured Data: a <script type="application/ld+json"> block that describes what a page is about, using the schema.org vocabulary. Google calls it its recommended format, because it sits apart from the visible HTML and is easy to generate and maintain.

How it works

A JSON-LD block is a set of key-value pairs with a few special keywords:

• @context says which vocabulary you are using, usually https://schema.org.
• @type declares what the thing is: Organization, Person, Product, Article and so on.
• @id gives the entity a stable identifier so other blocks and pages can refer to the same node. See @id (Schema Node Identifier).
• @graph lets you list several connected entities in one block.

Properties such as name, author, offers or sameAs then describe the entity and link it to other entities. Because each node can point to others, a site’s JSON-LD forms a small graph that search engines and AI systems can read as facts.

How it’s used in SEO and marketing

• Making pages eligible for Rich Results such as product snippets, breadcrumbs, events and videos.
• Declaring your brand, people and products as entities, with sameAs links to Wikidata, LinkedIn and other profiles, to support Entity Disambiguation.
• Generating markup at scale from a CMS or product feed with templates or Python, then validating it automatically.
• Handing AI agents cheap, unambiguous facts such as prices, policies and opening hours.

Example

A minimal Organization block might declare "@type": "Organization", a "@id" such as https://example.com/#organization, the name, logo and URL, and a sameAs list of official profiles. Every Article on the site can then set its publisher to that same @id, so all pages point to one brand entity.

MLforSEO’s point in Consistency beats completeness: schema for machines, not results is that five stable fields that agree everywhere are worth more than fifty that contradict each other.

Limitations and things to watch

• Validators such as the Rich Results Test and the schema.org validator check syntax and eligibility, not whether your facts are consistent across pages.
• Markup must match visible content. Google can ignore or take action on misleading structured data.
• Google has retired several rich result types (How-to in 2023, FAQ in 2026, and others). The markup is still valid schema.org and other systems can read it, but it no longer earns that display in Google.
• Structured data helps understanding; Google says it is not a ranking factor in itself.

Learn it on MLforSEO

• Structured Data Automation with Python: Detect, Generate and Validate JSON-LD at Scale
• What Schema.org @id Does and Why Your Entity Graph Falls Apart Without It
• Consistency beats completeness: schema for machines, not results
• AI Search, LLMs, Entity SEO & Knowledge Graph Strategies for Brands
• Schema Entity Audit Template