JSON-LD (JSON for Linked Data) is a JSON-based format for embedding structured data in a web page. Search engines read it to learn what the page is about, such as a product, an article, or an organization. The vocabulary usually comes from schema.org.
How is JSON-LD built?
The W3C published JSON-LD 1.1 as a Recommendation on July 16, 2020. Every JSON-LD document is also a valid JSON document. Three keywords carry the meaning. @context maps terms to unambiguous identifiers, @type names the type of an object, and @id identifies a node so that statements can be linked.
On web pages JSON-LD sits in a <script type="application/ld+json"> element. Google allows it in the head or the body, and it can also be injected with JavaScript. That separates it from Microdata and RDFa, which put attributes into the visible HTML.
Why does Google recommend JSON-LD?
Google supports JSON-LD, Microdata, and RDFa and recommends JSON-LD. Google calls JSON-LD the easiest solution for site owners to implement and maintain at scale, which makes it less prone to user errors. The data must match the visible content and belong to the page it sits on. Markup on blank pages is not allowed.
Whether a page then gets a rich result depends on the type and its required properties. Google recommends validating markup with the Rich Results Test.
| Format | Where does the markup go? | Relation to visible content | Google's position |
|---|---|---|---|
| JSON-LD (this term) | in a <script type="application/ in the <head> or <body> | separate from the visible text; nested data is easy to express | recommended as easiest to implement and maintain at scale, less prone to user errors; also read when injected with JavaScript |
| Microdata | as attributes on HTML elements, usually in the <body>, also possible in the <head> | nested within the HTML content | supported; equally fine for Google as long as the markup is valid and implemented per the documentation |
| RDFa | as attributes on HTML elements, in the <head> and <body> | attributes on elements with user-visible content | supported; equally fine for Google as long as the markup is valid and implemented per the documentation |
All three formats are syntaxes for the same vocabulary: according to schema.org, its types and properties can be expressed in Microdata, RDFa and JSON-LD. Google supports all three unless the documentation for a specific feature says otherwise. Google recommends JSON-LD but calls all three formats equally fine as long as the markup is valid. Sources: Google Search Central: Introduction to structured data markup in Google Search (last updated 2025-12-10); schema.org: Data model. Last checked: .
Example
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "Organization",
"name": "Example Inc.",
"url": "https://www.example.com/",
"logo": "https://www.example.com/logo.png"
}
</script>
JSON-LD and AI assistants
A test by searchVIU (2025) checked whether AI services use information from structured data. None of the five services tested (ChatGPT, Claude, Gemini, Perplexity and Google AI Mode) reported a price that appeared only in a test page’s JSON-LD, while ChatGPT and Gemini did find prices shown in the visible text. Google says AI Overviews and AI Mode need no special markup and expects structured data to match the visible content.
Common mistakes
- The markup describes content that is not visible on the page.
- Required properties are missing, so the type does not qualify for a rich result.
- The JSON is invalid, for instance through a trailing comma or typographic quotation marks.
FAQ
What is the difference between JSON and JSON-LD? JSON-LD is JSON with additional keywords such as @context and @type that tie the data to a vocabulary.
How can I validate JSON-LD? With Google’s Rich Results Test for Google features and with the Schema Markup Validator for general schema.org syntax.
More on the searchVIU blog
Sources
- Introduction to structured data markup - Google Search Centraldevelopers.google.com
- JSON-LD 1.1 - W3C Recommendation, 16 July 2020w3.org
- Schema Markup and AI in 2025: What ChatGPT, Claude, Perplexity & Gemini Really See – searchVIU blog (2025)searchviu.com
- AI features and your website – Google Search Centraldevelopers.google.com
Last updated: