Schema Markup for AI SEO

Schema markup is structured data (JSON-LD) that explicitly tells search engines and AI engines what a page contains. For AI SEO, four schema types carry most of the weight: FAQPage (for question-answer pages), Article (for every page), HowTo (for procedural pages), and BreadcrumbList (for site hierarchy). Adding schema correctly is a one-time technical setup, not an ongoing task, and can be done by AI tools in a single prompt.

Why schema matters more for AEO than for traditional SEO

AI engines read your page twice: once as raw text for extraction, and once as structured data for classification. Schema is the structured-data pass.

A page without schema: the AI engine has to infer "this is a Q&A page about X" from the prose, which works but adds uncertainty.

A page with FAQ schema: the AI engine KNOWS "this is a Q&A page, these are the questions, these are the answers". Uncertainty drops. Citation probability rises (schema.org).

Classical Google SEO cared about schema mainly for rich results (those expandable boxes in search results). AEO cares about schema for the classification signal itself.

The four schema types that matter for AEO

In order of impact.

1. FAQPage schema

Used on any page answering multiple related questions. The schema wraps each question-answer pair explicitly.

{
  "@context": "https://schema.org",
  "@type": "FAQPage",
  "mainEntity": [{
    "@type": "Question",
    "name": "What is AI SEO?",
    "acceptedAnswer": {
      "@type": "Answer",
      "text": "AI SEO is the practice of structuring..."
    }
  }]
}

Google's rich result test validates this in 30 seconds. Every AEO page should have it or Article schema minimum (developers.google.com).

2. Article schema

Used on every single page regardless of content type. Fields that matter most:

  • headline (the page title)
  • author (a real person with a linked author page if possible)
  • datePublished (the original publication date)
  • dateModified (last substantive update)
  • description (short page summary)
  • mainEntityOfPage (the canonical URL)

The author field is worth paying attention to. An author with a visible bio and other content attributed to them reads as more credible than a missing or generic author.

3. HowTo schema

Used on procedural pages ("how to set up X"). Breaks the steps into discrete data points the AI can parse.

{
  "@context": "https://schema.org",
  "@type": "HowTo",
  "name": "How to add FAQ schema to a WordPress page",
  "step": [{
    "@type": "HowToStep",
    "position": 1,
    "name": "Install Rank Math",
    "text": "Install the Rank Math plugin and activate it."
  }, {
    "@type": "HowToStep",
    "position": 2,
    "name": "Enable FAQ schema",
    "text": "Open the page editor, add the FAQ block, and enter each question-answer pair."
  }]
}

Most AEO pages are not HowTo pages, so this is optional. For pages that ARE procedural, it is worth adding.

4. BreadcrumbList schema

Used on every page to declare the hierarchy. Example for a spoke page:

Home > AI SEO for Business Owners > What is AI SEO?

Helps AI engines understand site structure and topical clustering. Small impact individually, real impact as a cumulative signal.

Adding schema without writing code

Three ways a non-technical business owner adds schema.

  1. AI coding tools. Claude Code generates JSON-LD from page content in one prompt. Paste the schema into the page's HTML head. Validate in Google's rich result test
  2. CMS plugins. WordPress plugins like Rank Math or Yoast SEO auto-generate Article schema and offer FAQ schema input fields. Webflow and Framer have similar plugins or native support
  3. Schema generators online. Merkle, TechnicalSEO, and Schema.org have free generators where you fill in fields and get the JSON-LD to paste (developers.google.com)

For most non-technical owners, option 2 (CMS plugin) is the easiest. For owners using Claude Code, option 1 is faster.

Common schema mistakes that cost citations

Four mistakes to avoid.

  1. Schema content that does not match page content. The schema says the answer is "X" but the page actually says "Y". AI engines detect the mismatch and downweight the page
  2. Missing dateModified. The schema says datePublished: 2023-01-15 with no modification date. The page looks stale even if the content is current
  3. Generic or missing author. author: "Admin" or no author at all. Looks low-credibility
  4. Stale schema after content updates. The page gets updated but the schema block is not re-generated. The mismatch breaks trust

The fix for all four: regenerate schema whenever content changes, using the same AI tool that generated it originally.

The 30-minute schema audit

For an existing site with no schema discipline:

  1. Open 5 random pages and view source. Look for application/ld+json blocks
  2. Paste each block into Google's Rich Results Test. Note which validate and which error
  3. Identify the most common issues (missing dateModified, generic author, wrong type)
  4. Write one AI prompt that fixes the pattern across all pages and test it on 3 sample pages before rolling out

30 minutes tells you whether schema is a small cleanup or a full rebuild.

Schema for multi-language or multi-region sites

If the business serves multiple regions or languages, schema needs to declare that.

  • inLanguage field in Article schema
  • Separate URLs per language (not URL parameters) with proper hreflang tags
  • Region-specific schema fields where relevant (Address, AreaServed for local businesses)

Most small businesses are single-language and single-region so this is not relevant. For the ones that are multi-region, getting this right compounds citation share across markets.