Schema Markup for AI Search: What Actually Matters in 2026

Why schema is the single highest-leverage AEO investment

Most websites have plain HTML. The browser renders it, humans read it, and search engines do their best to interpret it. Schema markup is the structured data layer that turns "we hope Google understood that" into "Google definitely understood that, in the exact format AI engines extract from."

For AI search specifically, schema is more important than it is for traditional Google rankings. AI engines like ChatGPT, Perplexity, and Google AI Overviews extract structured data directly into their answers. A page with proper FAQPage schema is dramatically easier to cite than the same content as plain HTML.

This post covers the five schema types that actually move AI search citations in 2026, with what to add, common mistakes, and how to verify each.

1. Organization (or LocalBusiness for local)

Establishes who you are as an entity. Required for branded AEO (the AI engine accurately describing your business when asked by name).

Place it once on your homepage. Required fields: name, url, logo. Strongly recommended: description, sameAs (array of LinkedIn, Twitter, GitHub, etc.), areaServed, founder. For local businesses, use LocalBusiness instead of Organization and add: address (PostalAddress block), telephone, openingHours, geo (GeoCoordinates block), priceRange.

Common mistake: Organization schema with only name and url, missing sameAs links. The sameAs array is what lets AI engines confirm you are the same entity across LinkedIn, Twitter, your GBP, etc. Without it, the entity is ambiguous.

2. FAQPage

The single highest-leverage schema type for AEO. FAQPage schema is extracted directly by AI engines and used in their answers. A page with 5 to 8 well-structured FAQs in proper schema format is dramatically easier for AI to cite than the same content as plain HTML.

Place it on every page where you have FAQs: homepage, service pages, product pages, top blog posts. Each Q&A should be 40 to 60 words for the answer (the optimal length for AI engine extraction). The questions should be the actual questions your customers ask, not marketing-speak.

Common mistake: FAQPage schema with Q&A that does not match the visible page content. Google specifically penalizes this. The schema should mirror what is on the page, not extend or replace it.

3. Article

For every blog post, news article, and long-form content page. Tells AI engines this is editorial content, who wrote it, when it was published, and what it is about.

Required: headline, datePublished, author, publisher. Strongly recommended: dateModified (signals freshness on every redeploy), image, articleBody, mainEntityOfPage. The author field should be a real person or organization with their own URL or LinkedIn profile, not "Admin" or "Editor."

Common mistake: Article schema without dateModified. Google uses dateModified as a freshness signal; without it, the post looks stale by default.

4. Product

For ecommerce. Required for any product to appear in Google Shopping and increasingly required for AI shopping recommendations (Shopify ChatGPT integration, Google AI shopping).

Required: name, image, description, sku, offers (with price, priceCurrency, availability). Strongly recommended: aggregateRating, review, brand, gtin (UPC/EAN/ISBN if applicable), category. Most Shopify themes ship with basic Product schema; the recommended fields require theme customization.

Common mistake: Product schema without offers.priceCurrency. AI shopping engines typically filter out products with missing currency.

5. BreadcrumbList

Site hierarchy. Tells search engines and AI engines how your pages are organized relative to each other. Required for breadcrumb navigation to appear in Google search results.

Place it on every page that has breadcrumb navigation (typically every page except the homepage). The position field starts at 1 (homepage) and increments through the hierarchy. Skip the current page from the list. Schema is for path, not destination.

Common mistake: BreadcrumbList with item.id pointing to non-canonical URLs (with tracking parameters or different protocols). Always use the canonical URL.

Two more worth adding

Beyond the core five, two more schema types are worth adding for specific use cases.

HowTo schema for step-by-step guides. Each step gets its own HowToStep block with a name, text, and optional image. AI engines extract HowTo schema for "how to" queries. Common in DIY, recipe, and tutorial content.

VideoObject schema for any embedded or hosted video. AI engines like Google AI Overviews increasingly cite video answers. VideoObject schema (with name, description, thumbnailUrl, uploadDate, contentUrl) makes the video citable.

The five most common schema mistakes that hurt AI search

Beyond the per-type mistakes flagged above, five recurring mistakes show up across nearly every site we audit:

1. Schema in microdata or RDFa instead of JSON-LD. Older themes and CMS plugins sometimes output schema as inline microdata attributes on HTML elements. Google supports it but the AI engine extraction pipeline strongly prefers JSON-LD. If your schema is microdata, convert it. The work is one-time.
2. Schema that does not match visible page content. Putting Q&A in FAQPage schema that does not appear in the visible HTML is treated as cloaking by Google and gets penalized. Always mirror the schema to what is on the page; never extend beyond it.
3. Copy-pasted boilerplate Q&A. Generic FAQ blocks like "What payment methods do you accept" pasted across every product page do not earn citations. AI engines deduplicate boilerplate and skip pages whose schema is mostly common to other pages. FAQ should be specific to the page.
4. Missing required fields. Organization schema without name or url, Product schema without offers.priceCurrency, Article schema without datePublished. These pages still validate as "valid" but the AI engine cannot use them. Use validator.schema.org's "Recommended Properties" section in addition to "Required Properties."
5. Invalid types for the page content. Product schema on a service page, LocalBusiness schema on a national franchise homepage, Article schema on a category landing page. Type mismatch is one of the easiest ways to confuse an AI engine. Match the schema type to the actual page intent.

All five are easy to catch with the validation tools above. The fix for each is usually less than 30 minutes once identified.

How to verify schema is working

Three free tools, run after every schema change:

1. Google's Rich Results Test (search.google.com/test/rich-results). Shows what Google extracts. Run it on your URL after publishing.
2. validator.schema.org. Shows whether your JSON-LD is technically valid. Catches missing required fields, type mismatches, syntax errors.
3. Schema App Schema Markup Validator. Gives detailed warnings beyond strict validation. Useful for catching deprecated fields and best-practice issues.

If all three pass, your schema is in good shape. If any fail, fix the highlighted issues before redeploying.

A minimal Organization schema you can copy

Here is the smallest viable Organization schema block. Adapt the values to your business and paste into your homepage <head> as a JSON-LD <script>:

{
  "@context": "https://schema.org",
  "@type": "Organization",
  "name": "Your Business Name",
  "url": "https://yourdomain.com",
  "logo": "https://yourdomain.com/logo.png",
  "description": "One-sentence description of what you do and who you serve.",
  "sameAs": [
    "https://www.linkedin.com/company/your-business",
    "https://www.instagram.com/your-business"
  ],
  "areaServed": {
    "@type": "Country",
    "name": "Canada"
  }
}

For local businesses, swap "Organization" for "LocalBusiness" and add: address (PostalAddress block), telephone, openingHours, geo (GeoCoordinates block with latitude and longitude), and priceRange. The pattern stays the same; the additional fields make the entity unambiguous for local AI search.

Validate the result at validator.schema.org before deploy. The validator catches typos, missing required fields, and type mismatches in seconds.

The order to add schema in if you are starting from scratch

If you are adding schema from scratch, this order works for most sites:

1. Organization or LocalBusiness on the homepage (10 minutes)
2. BreadcrumbList site-wide (10 minutes if templated)
3. Article on blog posts (template change, 30 minutes)
4. Product on ecommerce SKUs (theme work, 1 to 2 hours depending on theme)
5. FAQPage on top traffic pages (one page at a time, 15 minutes per page)

You do not need to do all five at once. Start with Organization and BreadcrumbList, validate, ship. Add the others over the next few weeks. Each addition compounds.

What we ship in our Optimization engagements

Our SEO + AEO Optimization ships the schema work as part of the deliverable.

Our SEO + AEO Optimization ships the five core schema types plus the broader SEO and performance baseline directly on your specific theme. Done-for-you, From $2,500, in 1 to 3 weeks.

What to do this week

Three concrete steps:

1. Run your homepage through Google's Rich Results Test. Note what schema is detected and what is missing (5 minutes).
2. Run validator.schema.org on the same URL. Note any errors (5 minutes).
3. Pick one schema type from the list above that is currently missing on your site. Add it. Validate. Ship (15 to 60 minutes).

Or, if you want a complete schema audit, run the free Website Scorecard first, then look at SEO + AEO Optimization for the done-for-you engagement. Schema is one of six categories the scorecard grades.