Wayfinder AI
Access Compass
All Guides

Schema Markup for AI Search: High-Impact Types and Implementation

Guide to schema markup for AI discoverability—which types matter, JSON-LD examples, validation, and why it reduces hallucination.

Implementing AEO

Schema markup provides semantic context that AI agents need to understand your content reliably. For AI search, this is not about rich snippets or visual enhancements in search results. It is about structuring data so AI can extract correct information without guessing or hallucinating.

When an AI agent encounters raw text like "£99/month", it must infer this is a price, currency, and recurring cost. With proper schema, you tell it directly: "price": "99.99", "priceCurrency": "GBP". The distinction is unambiguous extraction versus probabilistic inference.

This guide maps schema types to AI discoverability impact. You will find practical JSON-LD examples ready to deploy, guidance on which types solve real problems versus which are nice-to-have, and validation methods to verify implementation works.

Why Schema Matters for AI

Schema is structured metadata that describes your content to machines. AI agents use it to understand content structure and context without parsing unstructured text.

The extraction problem. AI agents must parse unstructured content from millions of pages with different layouts. Schema makes this consistent. A product page with pricing in the header, body text, or sidebar all mean the same thing to a human but require different extraction logic for a machine. Schema standardises this.

Hallucination risk. Without schema, AI might misinterpret data. Consider a page listing "Free for 30 days, then £99/month." An AI might quote "This product is free" by cherry-picking the most prominent phrase. Schema clarifies the structure: trial period is temporary, full price applies after. This reduces citation errors.

Extraction at scale. AI agents process information across many sites. Schema allows consistent extraction regardless of layout variations. When every organisation marks up contact information identically, AI doesn't need custom logic for each site.

Citation confidence. When AI cites your content, schema gives it confidence the data is correct. Agents prioritise structured sources when multiple unstructured sources conflict. This is critical for factual queries where accuracy matters.

The key insight: schema does not help Google more than it helps AI agents. Google uses schema primarily for rich snippets. AI uses schema for understanding, extraction, and citation confidence across all content types.

High-Impact Schema Types

These five schema types have the highest impact for most sites. Implement them first.

1. Organisation (homepage, all sites)

What it does: Identifies your company, logo, contact info, and social profiles.

JSON-LD:

{
  "@context": "https://schema.org",
  "@type": "Organization",
  "name": "Your Company",
  "url": "https://yoursite.com",
  "logo": "https://yoursite.com/logo.png",
  "contactPoint": {
    "@type": "ContactPoint",
    "contactType": "Customer Service",
    "telephone": "+44-20-XXXX-XXXX",
    "email": "hello@yoursite.com"
  },
  "sameAs": [
    "https://twitter.com/yourcompany",
    "https://linkedin.com/company/yourcompany"
  ]
}

Impact: High. Helps AI understand who you are, how to contact you, and which social profiles represent your brand.

Where: Homepage <head>, deployed via template so it loads on every page.

2. Product (e-commerce, SaaS, product pages)

What it does: Names product, describes it, lists price, rating, and availability.

JSON-LD:

{
  "@context": "https://schema.org",
  "@type": "Product",
  "name": "Product Name",
  "description": "Product description",
  "image": "https://yoursite.com/product.jpg",
  "price": "99.99",
  "priceCurrency": "GBP",
  "availability": "https://schema.org/InStock",
  "url": "https://yoursite.com/products/product-slug",
  "aggregateRating": {
    "@type": "AggregateRating",
    "ratingValue": "4.5",
    "ratingCount": "123"
  }
}

Impact: High. AI can extract structured product information including price, availability, and reviews without parsing unstructured text.

Where: Product pages, in <head>.

3. Article / BlogPosting (content sites, blogs)

What it does: Names article, publication dates, author, and content type.

JSON-LD:

{
  "@context": "https://schema.org",
  "@type": "BlogPosting",
  "headline": "Article Title",
  "description": "Article summary",
  "datePublished": "2026-03-22",
  "dateModified": "2026-03-22",
  "author": {
    "@type": "Person",
    "name": "Author Name"
  },
  "image": "https://yoursite.com/article-image.jpg",
  "articleBody": "Full article text here..."
}

Impact: High. Helps AI date articles correctly and attribute authorship when summarising or citing content.

Where: Blog post pages, in <head>.

4. BreadcrumbList (all sites with hierarchy)

What it does: Shows navigation hierarchy for the current page.

JSON-LD:

{
  "@context": "https://schema.org",
  "@type": "BreadcrumbList",
  "itemListElement": [
    {
      "@type": "ListItem",
      "position": 1,
      "name": "Home",
      "item": "https://yoursite.com"
    },
    {
      "@type": "ListItem",
      "position": 2,
      "name": "Products",
      "item": "https://yoursite.com/products"
    },
    {
      "@type": "ListItem",
      "position": 3,
      "name": "Product Name",
      "item": "https://yoursite.com/products/product-slug"
    }
  ]
}

Impact: Medium-high. Helps AI understand site structure and navigation paths, which aids in contextual relevance when answering queries.

Where: Product/article pages, in <head> or body.

5. FAQ / HowTo (support, educational content)

What it does: Structures questions, answers, or step-by-step instructions.

FAQ JSON-LD:

{
  "@context": "https://schema.org",
  "@type": "FAQPage",
  "mainEntity": [
    {
      "@type": "Question",
      "name": "What is this product?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Answer text here"
      }
    }
  ]
}

HowTo JSON-LD:

{
  "@context": "https://schema.org",
  "@type": "HowTo",
  "name": "How to do X",
  "step": [
    {
      "@type": "HowToStep",
      "position": 1,
      "name": "Step 1",
      "text": "Step 1 description"
    }
  ]
}

Impact: Medium. Helps AI extract Q&A content and procedural instructions when users ask how-to questions or seek specific information.

Where: FAQ pages and tutorials, in <head>.

Medium-Impact Schema Types

These types are worth implementing if you have the relevant content, but are not critical for AI discoverability.

LocalBusiness. If you operate physical locations, this schema helps AI find opening hours, address, and phone number. Essential for local search queries.

Event. If you host events, this schema conveys dates, times, locations, and ticket prices. AI can surface events in response to "when" or "where" queries.

Recipe. For cooking sites, this helps AI extract ingredients, instructions, cooking times, and nutritional information.

NewsArticle. If you publish news content, this is similar to BlogPosting but signals the content is timely news. Use for breaking news or current affairs.

JobPosting. If you are hiring, this schema helps AI find job listings with details including title, company, salary, and location.

Priority: Implement these if your content type matches. Otherwise, focus on the high-impact types first.

Schema You Can Likely Skip

Some schema types are nice-to-have but rarely critical for AI discoverability.

Review/AggregateRating. Useful for Google rich snippets where star ratings appear visually. Less critical for AI agents because they can read review text directly from the page.

Thing/WebPage. Too generic. Prefer specific types like Product, Article, or Event that provide meaningful context.

Markup for rich snippets. Optimised for Google Search display. AI agents do not use these visual enhancements.

Custom properties not in schema.org. Stick to official schema.org types. Custom properties create extraction problems for AI agents that expect standardised structures.

Real talk: If you are unsure whether a schema type helps AI discovery, it probably does not. Focus your resources on the high-impact types listed above.

Implementation Best Practices

Follow these practices to implement schema correctly.

Format: Use JSON-LD (not RDFa or Microdata)

JSON-LD is the industry standard. It is easiest to implement as pure JSON in a script tag and is most AI-friendly. Avoid legacy formats like RDFa or Microdata unless you have specific compatibility requirements.

<head>
  <script type="application/ld+json">
    {
      "@context": "https://schema.org",
      "@type": "Organization",
      ...
    }
  </script>
</head>

Placement: In <head> section

Most search engines and AI agents expect structured data in the <head>. This ensures it loads before the body content and is processed first.

Validation

Both should pass without errors. Warnings indicate non-critical issues. Errors indicate broken markup that will not parse correctly.

Keep it accurate

Do not hallucinate properties. Do not add properties that do not match actual page content. Update schema when content changes. Inaccurate schema damages AI trust and reduces citation confidence.

Tool-based generation (if you have time)

Many frameworks have schema-generation libraries. Next.js offers the next-seo package. Generic Node.js libraries include schema-org. These reduce manual errors but require maintenance.

Common mistakes

  • Using outdated schema type definitions (check schema.org regularly)
  • Nesting properties incorrectly
  • Missing required fields (each type has required properties)
  • Adding schema that does not match actual content

Testing Your Schema

Verification is essential. Implement once, verify it works, monitor over time.

Automated testing. Use Google Rich Results Test or Schema.org validator to confirm markup parses correctly. These tools catch syntax errors and missing required properties.

Manual verification. Ask an AI assistant to extract information from your page. If it extracts pricing, dates, or contact details correctly, your schema is working. This tests what actually matters: can AI understand your content?

Monitoring. Check quarterly that schema has not been accidentally removed during site updates. CMS migrations, template changes, and developer deployments frequently break structured data.

Schema implementation is not a one-time task. It is part of ongoing content maintenance. Add schema validation to your release checklist to prevent regressions.

Implemented schema? Run a Compass audit to verify your site is discoverable by AI agents.