Schema Markup for AI Overviews: What Actually Works in 2026

Schema markup isn't new. It's been a ranking signal and featured snippet driver for years. What changed in 2024 is how heavily Google's AI Overviews lean on structured data when deciding which pages to cite. Schema isn't a nice-to-have anymore — it's one of the clearest signals you can send an LLM about what your page is and what it answers.

Here's what the citation data shows and how to implement each schema type correctly.

Which Schema Types Appear Most in AI Overview Citations

SearchEngineLand and independent SERP analysts tracking AI Overview citations through 2024–2025 found three schema types dominate:

FAQPage is the most frequently cited. Pages with FAQ schema get their question-and-answer pairs extracted and used almost verbatim in AI Overview text. The structured Q&A format maps directly to how AI Overviews present information — a question is asked, an answer is synthesized. Pages that already structure content this way in machine-readable form win disproportionately.

HowTo is heavily cited for procedural queries ("how to do X" searches). When you have sequential steps, HowTo schema lets Google extract step names, descriptions, and images as structured objects rather than inferring them from prose.

Article schema is the baseline. It establishes authorship, publication date, update date, and content type. AI Overviews use the dateModified field to assess freshness. Without Article schema, Google has to infer all of this from on-page signals — and it often infers wrong.

How to Implement FAQPage Schema

FAQPage schema tells Google that a section of your page contains explicit question-and-answer pairs. Here's the minimum viable implementation:

{
  "@context": "https://schema.org",
  "@type": "FAQPage",
  "mainEntity": [
    {
      "@type": "Question",
      "name": "What is schema markup?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Schema markup is structured data added to a webpage that helps search engines and AI systems understand the content's meaning and context."
      }
    },
    {
      "@type": "Question",
      "name": "Does schema markup improve AI Overview citations?",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "Yes. Pages with FAQPage schema are cited in AI Overviews significantly more often than equivalent pages without structured data."
      }
    }
  ]
}

Implementation rules that matter:

The name field should match your H2 or H3 heading text exactly, or very closely. Google cross-references schema with on-page content.
The text in acceptedAnswer should be a complete, self-contained answer. Don't assume the reader has context from the surrounding paragraph.
Keep answers under 300 words. Longer answers get truncated in AI Overview display.
Only include questions that are actually answered on the page. Mismatches between schema and content degrade trust signals.

Free tool

Schema markup generator

Generate valid JSON-LD in 60 seconds — no signup.

Try it

How to Implement HowTo Schema

HowTo schema is for step-by-step content. The key is mapping your actual numbered steps directly to schema steps objects:

{
  "@context": "https://schema.org",
  "@type": "HowTo",
  "name": "How to Add Schema Markup to a WordPress Site",
  "description": "A step-by-step guide to adding structured data to WordPress pages without a plugin.",
  "step": [
    {
      "@type": "HowToStep",
      "name": "Open your theme's header.php",
      "text": "Navigate to Appearance > Theme Editor and open header.php."
    },
    {
      "@type": "HowToStep",
      "name": "Paste your JSON-LD script",
      "text": "Add your JSON-LD schema block before the closing </head> tag."
    }
  ]
}

Don't create HowTo schema for content that isn't actually procedural. Google's quality raters flag mismatched schema type as a negative signal.

How to Implement Article Schema

Article schema is the foundation. Every content page should have it, and it should be accurate:

{
  "@context": "https://schema.org",
  "@type": "Article",
  "headline": "Schema Markup for AI Overviews: What Actually Works in 2026",
  "author": {
    "@type": "Organization",
    "name": "RankCrab Team"
  },
  "datePublished": "2026-05-24",
  "dateModified": "2026-05-24",
  "publisher": {
    "@type": "Organization",
    "name": "RankCrab",
    "url": "https://rankcrab.com"
  }
}

The dateModified field is read by AI systems to assess freshness. Keep it accurate. A page with a stale dateModified will be deprioritized even if the visible text says "updated."

The Entity-Clarification Angle: sameAs Links

One of the highest-leverage schema tactics for AI citation is using sameAs to connect your brand or page topic to a known knowledge graph entity.

{
  "@context": "https://schema.org",
  "@type": "Organization",
  "name": "RankCrab",
  "url": "https://rankcrab.com",
  "sameAs": [
    "https://www.wikidata.org/wiki/Q12345678",
    "https://www.linkedin.com/company/rankcrab"
  ]
}

sameAs tells the LLM: "This entity is the same as the known entity at this URL." When Google's knowledge graph already has information about an entity and your sameAs link connects to it, your content inherits some of that entity's established authority.

For personal brands and authors, link to a Wikipedia page, a LinkedIn profile, or a verified social account. For companies, a Wikidata entry is the strongest possible sameAs target — if you don't have one, creating a Wikidata entry is worth the effort.

Common Schema Mistakes That Hurt AI Citations

Duplicate FAQ schema across multiple pages. If you have identical FAQ markup on ten pages, Google reads them as duplicates and the citation signal is diluted. Each FAQ schema block should be unique to its page.

Schema that doesn't match the visible content. If your FAQ schema says "What is X?" but the page doesn't actually answer that question, you're sending a false signal. This triggers quality filters.

Missing dateModified in Article schema. Freshness matters more in AI search than traditional search. A page without dateModified is harder for Google to assess, and the default assumption is stale.

Using plugins that generate generic boilerplate. Many WordPress SEO plugins add Article schema automatically but fill in incorrect values (wrong datePublished, wrong author, missing sameAs). Review your auto-generated schema — don't assume it's correct.

Combining Schema Types on One Page

You can and should use multiple schema types on a single page. A comprehensive guide might have Article schema (for the whole page), FAQPage schema (for the Q&A section), and HowTo schema (for a step-by-step section).

Stack them as separate JSON-LD blocks in <script> tags, or combine them into one block with multiple @type values. Google handles both correctly.

For a broader look at schema's role in GEO, see the schema markup guide and the AI search hub.