Schema Markup for AI Search: Types, Examples and What Actually Works

Schema markup for AI search is structured data, written in the schema.org vocabulary, that labels the facts on a page (who published it, what product it describes, what it costs, who wrote it) so search engines and the AI systems built on top of them can read those facts without guessing. It is worth doing well, but it is not a secret AI switch: Google states that no special schema.org markup is needed to appear in AI Overviews or AI Mode, and no other AI engine documents a ranking boost from it.

Below: the formats, the 13 types most sites need (with Google rich result status checked in October 2026), what is documented about schema and AI engines, and how to add and validate it.

Key takeaways

• Schema markup is a shared vocabulary from schema.org, launched in 2011 by Google, Microsoft and Yahoo, used to describe entities on a web page in a machine-readable way.

• Google supports three formats (JSON-LD, Microdata and RDFa) and recommends JSON-LD where your setup allows it.

• Google says there is no special schema.org markup required for AI Overviews or AI Mode; standard SEO and structured data that matches visible text are what it asks for.

• Microsoft Bing has said publicly that schema markup helps its LLMs understand content. OpenAI, Anthropic and Perplexity have published nothing on how they use JSON-LD.

• FAQ rich results stopped appearing in Google Search on May 7, 2026, and HowTo rich results were deprecated in September 2023. The markup is still valid schema.org, but it no longer produces those visual results in Google.

• The types with live Google features in October 2026 include Article, Organization, Product, Review snippet, LocalBusiness, Breadcrumb (desktop only), Event, Video, Software app and Profile page.

• Every fact in your markup should also appear in the visible page text, because that is Google's guideline and the only form some AI fetchers are known to read.

What is schema markup?

Schema markup is code that describes the things a page is about using a standard vocabulary. The vocabulary lives at schema.org, a collaborative project started in 2011 by Google, Microsoft and Yahoo (Yandex joined later). It defines hundreds of types, such as Organization, Product and Event, and the properties each type can have, such as name, price or startDate.

With markup, you state facts instead of leaving them to inference: this is an Offer, the price is 49, the priceCurrency is EUR. Search engines use that explicit data to power rich results (stars, prices, event dates, breadcrumbs) and to understand entities, such as which logo and social profiles belong to your company.

JSON-LD vs Microdata vs RDFa

There are three ways to write schema markup. Google's structured data introduction says all three are acceptable when they are valid, and that it generally recommends JSON-LD.

In short: use JSON-LD, generate it from the same data that renders the page, and make sure every value in it is also visible to readers.

Schema markup for AI search: does it help, and what is documented?

This is the question most "schema for AI" articles answer with confidence they have not earned. Here is what the platforms themselves have said, as of October 2026.

Three practical conclusions follow. First, schema helps indirectly through the search indexes that many AI engines draw on: Google's index feeds AI Overviews and AI Mode, and Bing's index feeds Copilot and, in part, ChatGPT search. Better entity understanding in those indexes is a real benefit. Second, there is no evidence that adding a schema type causes an LLM to cite you; the 2023 GEO paper by Aggarwal et al. tested content changes such as adding statistics, quotations and citations, not markup. Third, when an assistant fetches a page live, it is not documented whether it reads the JSON-LD block or only the rendered text, so any fact that matters (prices, plan names, founding year) must be in the visible copy too.

In short: schema markup for AI search makes your facts explicit to the indexes AI engines rely on, which is useful, but nobody has shown it is a citation lever on its own. Treat it as hygiene that supports clear content, not a replacement for it.

For the specific case for FAQPage markup and LLMs, read our earlier piece on why FAQ schema matters for LLM visibility; note that Google has since retired the FAQ rich result itself, as covered below.

Types of schema markup that matter for SEO and AI search

Schema.org has hundreds of types, but most sites need a small set. For each type below: what it is, when to use it, its Google rich result status in October 2026 (from the Search Central search gallery and documentation updates log), and a minimal JSON-LD example. Each example goes inside a script tag like this one:

<script type="application/ld+json">{ ... }</script>

Organization

What it is: describes your company: legal name, logo, URL, contact details and the official profiles that belong to it (sameAs). When to use it: once, on your home page or about page. Google status: supported. Google says it can influence the logo shown in results and your knowledge panel, and there are no required properties. It is also the most useful type for entity clarity, because it ties your name to your domain and profiles.

{"@context":"https://schema.org","@type":"Organization","@id":"https://example.com/#org","name":"Example Co","url":"https://example.com","logo":"https://example.com/logo.png","sameAs":["https://www.linkedin.com/company/example"]}

WebSite

What it is: describes the site as a whole. When to use it: on the home page of the domain or subdomain root. Google status: used to indicate your preferred site name in results (required: name and url). The sitelinks search box that used to rely on it was discontinued in November 2024, according to MediaPost.

{"@context":"https://schema.org","@type":"WebSite","name":"Example Co","url":"https://example.com/","publisher":{"@id":"https://example.com/#org"}}

Article and BlogPosting

What it is: describes editorial content; BlogPosting and NewsArticle are subtypes of Article. When to use it: on every blog post, guide and news page. Google status: supported, with no required properties; Google recommends headline, image, datePublished, dateModified and author.

{"@context":"https://schema.org","@type":"BlogPosting","headline":"Schema markup for AI search","datePublished":"2026-10-26","dateModified":"2026-10-26","author":{"@type":"Person","name":"Jane Doe","url":"https://example.com/team/jane"}}

FAQPage

What it is: marks up a list of questions with one answer each. When to use it: only where the page genuinely has a visible FAQ written by you. Google status: retired. Google restricted FAQ rich results to well-known government and health sites in August 2023, then stopped showing them in Search entirely on May 7, 2026, and removed the documentation in June 2026 (Search Engine Journal). The markup is still valid schema.org and does no harm. Whether it helps LLMs is debated; our FAQ schema article covers that argument.

{"@context":"https://schema.org","@type":"FAQPage","mainEntity":[{"@type":"Question","name":"Does schema help AI search?","acceptedAnswer":{"@type":"Answer","text":"Indirectly..."}}]}

HowTo

What it is: step-by-step instructions with tools, supplies and steps. When to use it: on true instructional pages, if you want to describe the steps explicitly for other consumers of schema.org. Google status: deprecated. Google stopped showing HowTo rich results on mobile in August 2023 and on desktop on September 13, 2023 (Search Engine Land). Do not expect any Google feature from it; the value of a how-to page now comes from a clean numbered list in the visible HTML.

{"@context":"https://schema.org","@type":"HowTo","name":"Validate schema markup","step":[{"@type":"HowToStep","text":"Paste the URL into the Rich Results Test."}]}

Product

What it is: a product you sell, with name, image, brand and identifiers such as GTIN or SKU. When to use it: on individual product pages (not category pages). Google status: supported for product snippets and merchant listing experiences, which can show price, availability, ratings and shipping or return details. Google keeps extending it (for example Product.category was documented in July 2026).

{"@context":"https://schema.org","@type":"Product","name":"Trail Shoe X","sku":"TSX-42","brand":{"@type":"Brand","name":"Example"},"offers":{"@type":"Offer","price":"129.00","priceCurrency":"USD","availability":"https://schema.org/InStock"}}

Offer, AggregateRating and Review

What they are: supporting types nested inside Product, SoftwareApplication, Event and others. Offer holds price and availability, AggregateRating holds the average score and count, Review holds a single review. When to use them: whenever the page visibly shows a price or genuine reviews. Google status: review snippets are supported for types such as Product, Software App, Event, Recipe and Book. Pages using LocalBusiness or Organization markup are not eligible for stars on reviews they control about themselves, and since July 2026 the guidelines explicitly ban fake or undisclosed incentivized reviews.

{"@type":"AggregateRating","ratingValue":"4.6","reviewCount":"212"}

LocalBusiness

What it is: a subtype of Organization for businesses with a physical location: address, opening hours, geo coordinates, phone. Use the most specific subtype available, such as Dentist or Restaurant. When to use it: on each location page. Google status: supported (local business details in Search and Maps); name and address are required. Keep it consistent with your Google Business Profile.

{"@context":"https://schema.org","@type":"Dentist","name":"Example Dental","address":{"@type":"PostalAddress","streetAddress":"1 Main St","addressLocality":"Austin","addressCountry":"US"},"telephone":"+1-555-0100"}

BreadcrumbList

What it is: the page's position in your site hierarchy. When to use it: on every page below the home page, matching the breadcrumb trail readers see. Google status: supported, but since January 2025 Google documents breadcrumbs as desktop only because they no longer appear in mobile results.

{"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Blog","item":"https://example.com/blog"},{"@type":"ListItem","position":2,"name":"Schema markup"}]}

Person

What it is: an individual, used for authors, founders and experts, with jobTitle, worksFor and sameAs links to their profiles. When to use it: as the author of articles and on author or team bio pages. Google status: no standalone rich result, but Google supports ProfilePage markup with a Person as mainEntity for creator and author profile pages.

{"@context":"https://schema.org","@type":"ProfilePage","mainEntity":{"@type":"Person","name":"Jane Doe","jobTitle":"Head of SEO","worksFor":{"@id":"https://example.com/#org"}}}

SoftwareApplication

What it is: an app or SaaS product, with category, operating system, price and ratings. Subtypes include WebApplication and MobileApplication. When to use it: on your product or app landing page. Google status: supported as the Software app rich result. Google requires name, offers.price (use 0 if free) and either aggregateRating or review, so SaaS companies without genuine on-page ratings will not get the rich result, though the markup still describes the product.

{"@context":"https://schema.org","@type":"SoftwareApplication","name":"Example App","applicationCategory":"BusinessApplication","operatingSystem":"Web","offers":{"@type":"Offer","price":"0","priceCurrency":"USD"}}

Event

What it is: a scheduled event, online or in person, with start date, location, organizer and ticket offers. When to use it: on each event or webinar page. Google status: supported (event experiences in Search); name, startDate and location are required. Update eventStatus when an event is moved or cancelled.

{"@context":"https://schema.org","@type":"Event","name":"GEO Summit","startDate":"2026-11-18T09:00+01:00","location":{"@type":"Place","name":"Copenhagen","address":"Copenhagen, DK"}}

VideoObject

What it is: a video with title, description, thumbnail, upload date, duration and optionally a transcript or key moments. When to use it: on pages where a video is the main content or a prominent part. Google status: supported (video results, key moments). Google added the creator property and updated interactionStatistic in September 2026. Required: name, thumbnailUrl, uploadDate.

{"@context":"https://schema.org","@type":"VideoObject","name":"How to validate schema","thumbnailUrl":"https://example.com/thumb.jpg","uploadDate":"2026-10-01","duration":"PT4M30S"}

Which schema types are retired or limited in Google?

In short: build on the types with a live job to do (Organization, WebSite, Article, Product with Offer and reviews, LocalBusiness, Breadcrumb, Person, SoftwareApplication, Event, VideoObject). Keep FAQPage and HowTo only where they honestly describe the page, and stop expecting Google visuals from them.

Which schema types to use on which page

This is the mapping we recommend as a starting point. It is deliberately short: more types on a page do not mean more visibility.

How to add schema markup in six steps

1. Inventory your templates. List page templates (home, post, product, location), not individual URLs. Schema should be generated per template so it scales and stays correct.

2. Pick the types per template using the table above, and read the Google documentation page for each type to note required and recommended properties.

3. Generate JSON-LD from your data source. Pull price, dates, author and ratings from the same CMS fields or product database that render the visible page. Hand-typed JSON-LD is the main source of drift.

4. Connect entities with @id. Give your Organization a stable identifier such as https://example.com/#org and reference it as publisher, brand or worksFor everywhere else, so engines see one company, not twenty.

5. Validate before release (next section), then deploy to a small set of pages first.

6. Monitor after release in Search Console rich result reports and with a monthly crawl that checks the markup still matches the page.

Worked example: one connected graph for a SaaS pricing page

Here is a compact example of the pattern we prefer for schema markup for AI search and SEO alike: a single @graph that connects the company, the site, the product and the breadcrumb through @id references. Every value in it (plan name, price, currency) should also appear in the visible pricing table.

<script type="application/ld+json">{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://example.com/#org","name":"Example Co","url":"https://example.com","logo":"https://example.com/logo.png"},{"@type":"WebSite","@id":"https://example.com/#site","url":"https://example.com/","name":"Example Co","publisher":{"@id":"https://example.com/#org"}},{"@type":"SoftwareApplication","name":"Example Analytics","applicationCategory":"BusinessApplication","operatingSystem":"Web","publisher":{"@id":"https://example.com/#org"},"offers":{"@type":"Offer","name":"Growth plan","price":"199","priceCurrency":"EUR"}},{"@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Pricing","item":"https://example.com/pricing"}]}]}</script>

Before: the pricing page had no markup, the plan price lived only inside a JavaScript-rendered toggle, and the company name appeared in three spellings across the site. After: one connected graph, a server-rendered price table with the same values, and a single canonical company name. The bigger win is that page and markup now tell the same story.

How to validate schema markup

• Rich Results Test: shows which Google rich results a page or code snippet is eligible for, and flags missing required properties. It only reports on Google-supported features, so FAQ and HowTo no longer appear in it.

• Schema Markup Validator: checks any schema.org markup for syntax and vocabulary errors, regardless of Google support. Use it for types Google ignores.

• URL Inspection in Search Console: confirms Google actually found the structured data on the live, rendered page.

• Rich result status reports in Search Console: track errors and valid items across the site after deployment.

• A visible-text check: for key facts (price, plan names, author, dates), search the rendered page text. If a fact exists only in JSON-LD, add it to the copy.

Common schema markup mistakes

• Markup that contradicts the page. A price, rating or date in JSON-LD that differs from what readers see breaks Google's guidelines and gives AI engines conflicting facts.

• Self-serving review stars. Adding AggregateRating to your own Organization or LocalBusiness pages for reviews you control makes them ineligible for stars, and fake or undisclosed incentivized reviews violate Google's review guidelines.

• Marking up invisible content. FAQ answers or product details that exist only in the script block.

• Stale dates. dateModified updated automatically on every deploy without real content changes, or never updated at all.

• Duplicate, disconnected entities. Five Organization blocks from five plugins with different names and logos. Pick one source and use @id.

• Chasing retired features. Spending sprint time on FAQ or HowTo rich results that Google no longer shows.

• Treating schema as the AI strategy. Markup supports clear content and third-party authority; it does not replace them. See how to structure content for LLM extraction and how to get cited by AI search.

How to measure schema markup for AI search results

Because no AI engine documents a schema effect, the honest approach is to measure. Pick a group of pages, change only the markup (and the matching visible facts), and compare how often those URLs are cited for a fixed set of prompts before and after, against a control group you did not touch. Run the comparison over several weeks: AI citations are volatile, and in our volatility research single-day snapshots are close to meaningless.

In KIME, this means tracking a prompt set daily across engines and reading the citation analysis at URL level, so you can see whether the pages you changed are cited more often than the control pages. In our own tracking data (KIME tracking data, Jul 26 to Oct 26 2026, 30 prompts in the US, GB and DK), the single pages cited most were listicles, comparison guides and clearly structured explainers. That points to page format and content clarity as bigger factors than markup, which matches what the platforms document.

Schema markup for AI search is one part of on-page clarity. The sibling guides cover the others: meta titles, meta descriptions and headings, and the full AI search optimization process shows where markup fits in a team's plan.

Frequently asked questions

What is schema markup?

Schema markup is structured data that describes the entities on a web page, such as an organization, product, article or event, using the shared schema.org vocabulary. It is usually added as a JSON-LD script. Search engines use it to understand facts precisely and to power rich results like product prices, review stars, event dates and breadcrumbs.

Does schema markup help AI search?

Schema markup helps AI search indirectly, but there is no documented direct boost. Google says no special schema is needed for AI Overviews or AI Mode, and Microsoft has said schema helps its LLMs understand content. OpenAI, Anthropic and Perplexity publish nothing on it, so keep important facts in visible text and use markup to make them explicit.

What are the most important types of schema markup?

The most important types of schema markup for most sites are Organization, WebSite, Article or BlogPosting, BreadcrumbList and Person. Add Product with Offer and genuine reviews for ecommerce, SoftwareApplication for SaaS, LocalBusiness for physical locations, and Event or VideoObject where relevant. Use only the types that describe what the page actually shows.

Is FAQ schema still worth adding in 2026?

FAQ schema no longer produces rich results in Google, which stopped showing them on May 7, 2026. The FAQPage type is still valid schema.org and harmless when it matches a visible FAQ. Add it if your FAQ is real and you want the questions explicit for other consumers, but do not expect any visual change in Google Search.

How do I add schema markup to my website?

You add schema markup by placing a JSON-LD script in the page HTML, ideally generated by your CMS template from the same data that renders the page. Choose types per template, include required properties from Google's documentation, connect entities with @id, validate with the Rich Results Test, and monitor Search Console reports after release.

Which is better for SEO, JSON-LD or Microdata?

JSON-LD is better for most sites because it sits in one script block, is easy to generate and audit, and is the format Google recommends where possible. Microdata and RDFa are equally valid when implemented correctly, but they are woven into HTML attributes, so redesigns and template changes break them more often.

Is there special schema markup for AI Overviews or ChatGPT?

No, there is no special schema markup for AI Overviews, AI Mode or ChatGPT. Google states that no additional markup or AI text files are required for its AI features. OpenAI has not published any schema requirement for ChatGPT search. Standard schema.org types that match visible content are all any platform has asked for.

How do I test if my schema markup works?

Test schema markup with Google's Rich Results Test for eligibility in Google features, and the Schema Markup Validator for general schema.org errors. Then use URL Inspection in Search Console to confirm Google sees the markup on the live page, and check the rich result status reports for site-wide errors after deployment.

Related guides

• Why FAQ schema matters for LLM visibility

• How to structure content for LLM extraction

• How to get cited by AI search

• Headings for AI search

• Is your site agent ready?

• AI search optimization: the working process

Sources: Google Search Central: AI features and your website · Google: Introduction to structured data · Google: Structured data search gallery · Google Search Central: Documentation updates · Google: Organization markup · Google: Site names · Google: Article markup · Google: Review snippet · Google: Software app · Search Engine Journal: Google drops FAQ rich results · Search Engine Land: How-to rich results end · Search Engine Land: Bing and Copilot use schema for LLMs · MediaPost: Sitelinks search box discontinued · OpenAI: Crawlers · Aggarwal et al., GEO: Generative Engine Optimization · schema.org