Structured Data for AI Answers: What Matters and What Is Myth

Structured data attracts confident claims from both sides. One says schema markup is the key to being cited by ChatGPT. The other says AI made it pointless. The documentation supports neither. As of November 2026, the picture is simpler: structured data states your facts in a form machines can read without guessing, Google says its AI features need nothing special, and a few types do most of the useful work.

This guide sticks to what Google, OpenAI and schema.org actually say, and labels the rest as opinion.

What structured data does

Structured data is a shared vocabulary from schema.org, usually added to a page as JSON-LD, that labels facts: this text is the product name, this number is the price, this person wrote the article. Google's introduction to structured data explains how Search uses it to understand pages and to show rich results.

For AI answers the logic is indirect. A machine that can read your price, stock and ratings exactly has less to guess, and wrong guesses are how outdated prices and invented details end up in an answer about you.

What Google says about AI features

Google's AI features guide, last updated in December 2025, says there are no additional requirements to appear in AI Overviews or AI Mode, and no special schema.org structured data you need to add. Its advice is to follow normal Search best practice, including keeping structured data consistent with the text people can see.

Google's newer guide to generative AI search, updated in July 2026, goes further. It says structured data is not required for generative AI search, recommends it still because it helps pages qualify for rich results, and lists "overfocusing" on structured data among the things you do not need to do.

So no markup will put you into an AI Overview. If someone sells you one that will, they are claiming something Google's own documentation contradicts.

What OpenAI says about product data

OpenAI's help page on shopping in ChatGPT says product results are chosen by ChatGPT, are not ads, and draw on structured metadata from first-party and third-party providers, such as price and description, along with other third-party content such as reviews. It does not say that markup on your own pages is how that metadata is gathered. Our guide to ChatGPT shopping covers product feeds, which are the direct route OpenAI documents for merchants.

The types worth your time

Google's search gallery lists every type it supports. For a brand, a short list covers most of the value.

Product with Offer

For anything you sell, Product markup with an offer (price, currency, availability) is the first job. Google's product snippet and merchant listing guides list the properties, including shipping details and return policies. Google recommends setting one return policy for the whole business in Organization markup, with offer-level policies only for exceptions. Keep every value identical to the visible page.

Review and AggregateRating

Rating markup states review counts and averages. Note Google's review snippet rule: when a business controls the reviews about itself, its pages that use LocalBusiness or other Organization markup are not eligible for review stars. Product reviews on your own product pages are a different case.

Organization

Organization markup on your homepage states your name, logo, official profiles and contact points, and, as above, can carry your return policy. Google's Organization guide lists the properties. It keeps facts attached to the right company when names are similar.

Article

For editorial pages, Article markup with headline, author and dates makes authorship and freshness explicit. Pair it with a visible byline and date.

FAQ, without the rich result

In August 2023 Google limited FAQ rich results to well-known, authoritative government and health websites and cut back HowTo results, which it later dropped. In May 2026 it stopped showing FAQ rich results altogether, and removed the documentation in June. FAQPage remains a schema.org type, and a clear question-and-answer section is still useful to readers and to any machine reading the page. Just do not add the markup expecting a Google rich result.

The rest

Breadcrumb, Video, Event and many other types have their uses in Search. Google has also been retiring lesser-used types, starting with seven announced in June 2025. Add a type when the content exists and Google still supports it, not because a checklist says so.

Myths worth dropping

"Schema markup makes ChatGPT cite you." No AI company documents a mechanism like that as of November 2026. Citation follows from being a good source for the question, not from a script tag.

"There is a special schema for AI." Google says there is not, for its AI features. No schema.org type is required by any assistant we know of.

"Markup can stand in for content." Google's structured data policies say not to mark up content readers cannot see. Assistants quote prose; markup that has no matching text helps no one.

"Valid markup guarantees a rich result." The same policies say Google does not guarantee that structured data will show up, even when a page passes the test.

"You can block crawlers and stay visible through markup." A crawler that cannot fetch the page never reads the markup on it. Should you block GPTBot? covers which agents to allow.

How to check your markup

1. Syntax. Run key templates through Google's Rich Results Test or the Schema Markup Validator after every release.
2. Delivery. Check that the markup is in the HTML the server sends, not only injected by a script after the page loads. Many AI crawlers do not run JavaScript; the AI crawler readiness checklist has the details.
3. Agreement. Ask a few assistants plain factual questions about your product: price, availability, key specification. Wrong answers mean something in the public record disagrees with you. Start by making your page, your markup and your feeds agree.

A sensible order of work

1. Product with Offer on every page that sells something, matching the visible price.
2. Organization on the homepage, with your return policy if you sell online.
3. Ratings where genuine product reviews exist.
4. Article with author and date on editorial pages.
5. A validation check in every release, because broken markup fails without telling you.

The product page checklist pairs each markup item with the visible content it must match.

Check what the answers say

Markup makes your facts easy to read. Whether an assistant then states them correctly, and recommends you, shows up in the answers themselves. Vouch runs six-question buying conversations on ChatGPT, Claude, Gemini and Perplexity and records the sources each answer cites at every stage, so you can see where a wrong fact or a missing one costs you; the methodology explains how, and citation is defined in the glossary.

Vouch is invite-only for now: join the waitlist, and the invitation includes a 14-day free trial.