Skip to main content
GEO

Structured Data for AI Engines

How to use schema.org JSON-LD markup to make your content machine-readable for AI engines, improve rich results, and increase your citability in AI-generated answers.

Why Structured Data Matters for AI

Structured data is machine-readable metadata you add to your HTML that describes what your content is — not just what it says. While a human reading a page about WCAG can infer that the page is a technical guide with a publication date, an author, and a list of criteria, a machine needs explicit signals. Schema.org JSON-LD provides a standardized vocabulary for these signals.

For AI engines, structured data serves as an anchor for content extraction. When a language model or retrieval system encounters a page with explicit Article, FAQPage, or HowTo markup, it can extract structured facts more reliably than from unstructured prose. This increases the probability that your content is accurately cited and attributed.

Core Schema Types for Knowledge-Based Content

The most useful schema types for an accessibility knowledge base are:

  • Article / TechArticle — marks the page as an authored technical article with title, author, datePublished, dateModified, and description.
  • FAQPage — marks a page containing questions and answers; directly eligible for FAQ rich results in Google Search.
  • HowTo — marks step-by-step instructional content; eligible for HowTo rich results and strongly extracted by AI for procedural queries.
  • BreadcrumbList — marks navigation breadcrumbs; improves site structure signals for AI and displays in Google rich results.
  • WebSite — marks sitewide information; enables Sitelinks Search Box and improves brand recognition.
  • Person / Organization — marks author and publisher information; critical for EEAT signals.

Article Schema Implementation

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "TechArticle",
  "headline": "Structured Data for AI Engines",
  "description": "How to use schema.org JSON-LD to make content machine-readable for AI search engines.",
  "datePublished": "2025-09-01",
  "dateModified": "2026-04-01",
  "author": {
    "@type": "Person",
    "name": "Jane Doe",
    "url": "https://example.com/authors/jane-doe"
  },
  "publisher": {
    "@type": "Organization",
    "name": "inculva",
    "url": "https://inculva.com",
    "logo": {
      "@type": "ImageObject",
      "url": "https://inculva.com/logo.png"
    }
  },
  "mainEntityOfPage": {
    "@type": "WebPage",
    "@id": "https://inculva.com/guides/structured-data-ai-engines"
  }
}
</script>

FAQPage Schema

FAQPage schema is one of the most powerful types for GEO. AI systems are trained to recognize Q&A patterns and will preferentially extract from pages with explicit FAQ markup. Google also renders FAQ rich results with expanded answers directly in the SERP, dramatically improving click-through rate.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "FAQPage",
  "mainEntity": [
    {
      "@type": "Question",
      "name": "What is WCAG Level AA conformance?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "WCAG Level AA conformance means a website meets all Level A (30 criteria) and Level AA (20 additional criteria) success criteria. This is the standard required by most accessibility laws globally, including the EU Web Accessibility Directive and Section 508 in the US."
      }
    },
    {
      "@type": "Question",
      "name": "How do I test for WCAG compliance?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Testing for WCAG compliance requires a combination of automated tools (axe-core, Lighthouse, WAVE) and manual testing with keyboard navigation and screen readers (NVDA, JAWS, VoiceOver). Automated tools catch approximately 30-40% of accessibility issues; the remainder require human judgment."
      }
    }
  ]
}
</script>

HowTo Schema for Instructional Content

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "HowTo",
  "name": "How to Run an Accessibility Audit",
  "description": "A step-by-step guide to auditing a website for WCAG 2.2 compliance.",
  "totalTime": "PT2H",
  "step": [
    {
      "@type": "HowToStep",
      "position": 1,
      "name": "Run automated scan",
      "text": "Install axe DevTools in Chrome and run a scan on each page template. Export results as a spreadsheet."
    },
    {
      "@type": "HowToStep",
      "position": 2,
      "name": "Test keyboard navigation",
      "text": "Navigate every interactive element using Tab, Shift+Tab, Enter, Space, and arrow keys. Document any elements that are unreachable or have no focus indicator."
    },
    {
      "@type": "HowToStep",
      "position": 3,
      "name": "Test with screen reader",
      "text": "Use NVDA with Firefox and VoiceOver with Safari to navigate each page template. Check headings, forms, images, tables, and dynamic content."
    }
  ]
}
</script>

Validating Your Structured Data

Always validate structured data before deployment:

  • Google's Rich Results Test (search.google.com/test/rich-results) — checks eligibility for Google rich results and flags schema errors.
  • Schema Markup Validator (validator.schema.org) — validates against the schema.org specification.
  • Google Search Console → Enhancements — shows indexed rich result types and any errors detected by Googlebot.

Common mistakes to avoid: missing required properties, using deprecated schema types, nesting schema incorrectly, and adding fake or misleading information in schema fields (this violates Google's guidelines and can result in a manual penalty).

Accessibility and Structured Data Alignment

The best structured data implementations mirror the page's visible content — the schema description matches the visible meta description, the FAQ answers match the visible FAQ content, and the article title matches the visible h1. This alignment is both a Google requirement (schema must reflect what is on the page) and an accessibility principle (content should be consistent and predictable across representations).

Resources