Schema That Actually Gets You Cited by AI in 2026

TL;DR

AI engines disambiguate entities through structured data. The minimum schema set in 2026 is Organization on every page, Person for any named author, Article/BlogPosting on posts (with author, datePublished, dateModified), Product on pricing pages, LocalBusiness for local-serving businesses, BreadcrumbList on inner pages, and VideoObject when videos exist. Use JSON-LD, not microdata. Link entities via @id. Validate with Google’s Rich Results Test. Skip llms.txt — Google publicly confirmed it does nothing for Search and only 0.1% of AI-bot traffic touches it.

---

Why schema matters more in the AI-search era

Classic Google could understand a page from the words alone. AI engines extract entities — the business, the author, the product, the location — and they prefer explicit signals over inferred ones. Schema is the explicit signal.

The 2026 data on schema impact:

• Pages with Article schema (with author + dates) are cited roughly 3× more often than the same pages without it
• Businesses with LocalBusiness schema rank measurably better in map-pack queries than competitors without
• Product schema with proper price + availability data is required for inclusion in Google AI Overviews on shopping queries
• Anonymous content with no author or organization signal is the single largest reason small business sites get skipped by AI citation

Most small business sites we audit have zero structured data. That’s the gap.

The minimum schema set for 2026

Organization (every page)

Identifies the business behind the site. Should appear sitewide via the header or footer template. Minimum properties: name, url, logo, sameAs (linking to your social profiles), contactPoint for phone, and an @id you’ll reference from other schemas.

{
  "@context": "https://schema.org",
  "@type": "Organization",
  "@id": "https://codecraftstudio.ca/#organization",
  "name": "CodeCraft Web Studio",
  "url": "https://codecraftstudio.ca/",
  "logo": "https://codecraftstudio.ca/logo.png",
  "sameAs": [
    "https://www.linkedin.com/company/codecraftstudio",
    "https://www.instagram.com/codecraftstudio"
  ]
}

Person (any named author)

Disambiguates the human behind the content. Required for AI engines to attribute expertise. Minimum: name, url (to an author page), jobTitle, worksFor referencing the Organization @id, sameAs linking to LinkedIn.

Article / BlogPosting (every blog post)

Marks up the post itself. Must include author (referencing Person), publisher (referencing Organization), datePublished, dateModified, headline, and image. The date fields are where most sites slip — Schema must match the visible “Updated” date on the page.

Product (pricing pages)

For service packages or physical products. Include name, description, offers with priceCurrency “CAD” and price, brand referencing Organization. Product schema on service packages is what lets AI engines cite “CodeCraft’s Launch package is $2,400 CAD” with confidence.

LocalBusiness (local-serving businesses)

Required for any business with a service area. Use the most specific subtype available — Plumber, HVACBusiness, Restaurant, Bakery, WebDesignCompany (if applicable). Include address, geo (lat/lng), openingHoursSpecification, telephone, priceRange, and areaServed.

BreadcrumbList (inner pages)

Helps engines understand site hierarchy and surfaces breadcrumb rich results. One line per breadcrumb, in order.

VideoObject (any page with embedded video)

YouTube is 23% of Google AI Overviews citations. If you have videos, mark them up — name, description, thumbnailUrl, uploadDate, contentUrl or embedUrl, duration in ISO 8601 format.

FAQPage — yes, even after the May 2026 change

Google killed the FAQ rich result on desktop in May 2026 and on mobile shortly after for most sites. That doesn’t mean FAQPage schema is dead — it means the visual rich result is dead.

FAQPage schema is still parsed by:

• Bing (which retains FAQ rich results)
• Perplexity (uses structured Q&A in answer synthesis)
• ChatGPT search (cites FAQ pairs)
• Google AI Overviews (extracts from FAQPage markup even without visual rich results)

Keep FAQPage schema wherever you have real Q&A content. Don’t add fake FAQs just to chase the schema — Google penalizes thin/manipulative FAQ markup.

Why @id linking matters

The single most-skipped schema tactic is using @id URIs to link entities together. Without @id linking, every Person on the site looks like a different Person. With it, AI engines understand that the author of post A is the same Person as the author of post B and is associated with the same Organization sitewide.

The pattern:

• Organization: "@id": "https://yoursite.ca/#organization" (defined once, in the sitewide template)
• Person: "@id": "https://yoursite.ca/about/team/vikas/#person"
• In Article schema, reference them: "publisher": { "@id": "https://yoursite.ca/#organization" } and "author": { "@id": "https://yoursite.ca/about/team/vikas/#person" }

This is how you build a coherent entity graph instead of disconnected JSON-LD blocks.

JSON-LD, not microdata

Microdata (the itemscope itemtype attributes embedded in HTML) was the older way to mark up schema. JSON-LD is the modern way and what every engine prefers in 2026:

• Easier to maintain — one block in the <head> per type, not scattered through markup
• Doesn’t pollute the HTML
• Google explicitly recommends JSON-LD over microdata
• Easier for AI engines to extract programmatically

If your site is on WordPress, use Rank Math, Yoast, or hand-rolled JSON-LD in the theme. If you’re on Shopify, the platform emits some Product schema by default — you’ll need to add Organization, Article, and LocalBusiness via theme code or apps. If you’re on Webflow, the Custom Code embed in page settings is where it lives.

Validating your schema

Three tools, in order of importance:

1. Google’s Rich Results Test — paste your URL or code, see what Google detects and whether it’s valid. This is the ground truth.
2. Schema.org Validator (validator.schema.org) — checks against the schema.org spec rather than Google’s narrower rich-result requirements. Useful for verifying schema that Google doesn’t surface but AI engines still parse (like FAQPage).
3. Bing Webmaster Tools URL Inspection — Bing’s own schema validator. Worth checking because Bing’s rich-results coverage is broader than Google’s.

If a tool throws errors, fix them. Warnings (missing optional properties) can be ignored if the page is otherwise complete.

The llms.txt non-story

In 2024 there was a proposal for an llms.txt file (similar to robots.txt but for AI engines) that would tell LLMs which content to index and how. The 2026 reality:

• Google publicly stated llms.txt does nothing for Search. No production weight, no ranking impact.
• Otterly AI’s 2026 traffic analysis found 84 of 62,100 AI-bot requests touched llms.txt — that’s 0.1%.
• No major AI vendor (Google, OpenAI, Anthropic, Perplexity, Meta) honors it in production retrieval as of 2026.

Skip it. The work that does matter — Organization, Person, Article, Product schema — actually moves citation outcomes. llms.txt is the SEO equivalent of a “this site is mobile-friendly” badge from 2012.

Copy-paste starter snippets

Article schema (blog post)

{
  "@context": "https://schema.org",
  "@type": "BlogPosting",
  "headline": "Schema That Actually Gets You Cited by AI in 2026",
  "image": "https://yoursite.ca/hero.jpg",
  "datePublished": "2026-06-30",
  "dateModified": "2026-06-30",
  "author": {
    "@type": "Person",
    "@id": "https://yoursite.ca/about/team/vikas/#person",
    "name": "Vikas Saini"
  },
  "publisher": {
    "@id": "https://yoursite.ca/#organization"
  }
}

LocalBusiness schema

{
  "@context": "https://schema.org",
  "@type": "WebDesignCompany",
  "name": "CodeCraft Web Studio",
  "image": "https://codecraftstudio.ca/logo.png",
  "telephone": "+1-905-XXX-XXXX",
  "priceRange": "$$",
  "address": {
    "@type": "PostalAddress",
    "addressLocality": "Oakville",
    "addressRegion": "ON",
    "addressCountry": "CA"
  },
  "areaServed": ["Oakville", "Mississauga", "Burlington", "Milton"]
}

Product schema (service package)

{
  "@context": "https://schema.org",
  "@type": "Product",
  "name": "Launch Package — Shopify Website",
  "description": "5-page Shopify build for small businesses",
  "brand": { "@id": "https://codecraftstudio.ca/#organization" },
  "offers": {
    "@type": "Offer",
    "price": "2400",
    "priceCurrency": "CAD",
    "availability": "https://schema.org/InStock"
  }
}

Frequently asked questions

How long until schema changes show up in AI citation?

30–60 days for Perplexity, 60–120 days for ChatGPT and Google AI Overviews. Schema is necessary but not sufficient — you also need content quality and clear authorship for citation lift to materialize.

Will Google penalize me for using schema “wrong”?

Yes, if the schema is manipulative — fake reviews, FAQ markup with no visible FAQ on the page, Product schema for things you don’t sell. Google’s spam detection on structured data has improved sharply. Accurate schema only.

What about Speakable schema?

Originally for voice search. Use it if you have content meant to be read aloud, but voice-search-specific traffic is small. Not a priority for most small businesses.

Do I need Review schema if my reviews live on Google?

Don’t fake first-party Review schema using Google reviews — that’s a policy violation. AggregateRating is fine if you genuinely collect reviews on your own site. For Google reviews, the LocalBusiness profile is the citation surface.

Should I add Event schema for events I’m hosting?

Yes. Event schema surfaces in Google search and AI engines pull from it. Include name, startDate, location, organizer, and offers if it’s paid.

What’s the difference between Article and BlogPosting?

BlogPosting is a subtype of Article. For blog posts, use BlogPosting. For more formal long-form content (whitepapers, reports), use Article. Engines treat them similarly.

Does too much schema slow my site down?

Negligibly. JSON-LD adds 2–5KB per page, deferred and parsed by engines, not blocking render. Performance is a non-issue.

Can I have multiple schema blocks on one page?

Yes. Most pages should have at least Organization + WebPage, plus the relevant content type (Article, Product, LocalBusiness, etc.) and BreadcrumbList. Each in its own JSON-LD <script> block is fine.

What to do next

If you’re starting from zero, the highest-ROI move is Organization + Person + Article schema across your blog and main pages. That alone closes the biggest gap most small business sites have. We implement and validate schema as part of our SEO marketing service, usually bundled into an initial audit and cleanup.

If you want a quick check of what schema your site currently emits, send us your URL and we’ll run it through Rich Results Test plus the schema.org validator and send back what you have, what’s broken, and what’s missing. No pitch.

Related guides

• How to Get Recommended by ChatGPT
• What is GEO? Get Quoted by AI Search
• Technical SEO Issues We Find on Every Site
• Optimizing for Perplexity (And Why It’s Different)