Add schema.org JSON-LD for Agents
Mark up your entities with schema.org JSON-LD so an agent can extract structured, typed facts in one parse, not just ambiguous prose.
What is JSON-LD?
JSON-LD (JSON for Linking Data) is the W3C's JSON format for linked data; version 1.1 has been a W3C Recommendation since 16 July 2020 (W3C, 2020). On a web page it sits in a <script type="application/ld+json"> block that uses the shared schema.org vocabulary, one of the three formats schema.org supports beside Microdata and RDFa (Schema.org, 2026). Instead of leaving an agent to infer that a string is a price, a date or an author, you state it as typed data: a Product with an offers.price, an Article with a datePublished, an Organization with a sameAs link to its other identities.
Why does JSON-LD matter for agents?
JSON-LD matters because typed data is unambiguous where prose is not: an agent lifts exact facts in one parse, which lowers Cost of Retrieval and the chance of a wrong answer. Google recommends JSON-LD over Microdata and RDFa as the easiest format to implement and maintain at scale (Google, 2025), and its guidelines require the markup to be a true representation of the visible page content (Google, 2026). It is the structured-data backbone of the quality dimension, and sameAs ties your entity into the wider knowledge graph, mirroring the Lexicon's DefinedTerm pattern.
How do you implement JSON-LD?
Implement JSON-LD by choosing the right type, emitting valid markup in the initial HTML, and validating it.
- Pick the schema.org type that matches the page's primary entity (
Article,Product,Organization,FAQPage,HowTo) and fill its required properties. - Emit it as a
<script type="application/ld+json">block in the page head, server-side; use an@graphwhen one page carries several linked entities and addsameAsto connect identities. - Validate with a schema.org-aware validator such as the Schema Markup Validator and confirm the JSON parses and the types resolve.
The Agent-Readiness Audit's json-ld check (7 of 100 points) passes when the first application/ld+json block in the raw HTML parses as JSON; it reads the initial HTML only, never a script-rendered DOM. This page is live proof: its @graph carries WebPage, HowTo, FAQPage and BreadcrumbList nodes, and the questions below are the visible twin of that FAQPage node.
JSON-LD — frequently asked questions
Which structured-data format should you use for agents?
JSON-LD. Google recommends it as the easiest format to implement and maintain at scale (Google, 2025), and it lives in one script block instead of being spread through the markup, so an agent parses it in one pass.
Do FAQ rich results still appear in Google Search?
No: Google stopped showing FAQ rich results in Search on 7 May 2026 and removed the documentation on 15 June 2026 (Google Search Central, 2026). FAQPage markup still hands agents a typed question-and-answer list, provided the same text is visible on the page.
Can you inject JSON-LD with JavaScript?
Google states it can read JSON-LD injected by JavaScript (Google, 2025), but most agents fetch raw HTML without executing scripts, and the Agent-Readiness Audit reads the initial HTML only. Emit the block server-side.
Sources
- W3C: JSON-LD 1.1 — A JSON-based Serialization for Linked Data (W3C Recommendation), 2020. w3.org
- Schema.org: Getting started with schema.org, 2026. schema.org
- Google Search Central: Introduction to structured data markup in Google Search, 2025. developers.google.com
- Google Search Central: General structured data guidelines, 2026. developers.google.com
- Google Search Central: Search Central updates (FAQ rich result deprecation, May 2026), 2026. developers.google.com
Related: schema.org for agents · the quality dimension · the Agentic Web Lexicon · audit your site
