Schema markup for AI search: what helps

Schema markup, or structured data, is code on a page that describes its content in a standard vocabulary: this is a product, this is its price, this is the company that makes it. Google explains that it gives explicit clues about the meaning of a page, and recommends the JSON-LD format (Google Search Central).

A typical example is a product page that carries a block of JSON-LD stating the product name, brand, price, currency and availability. The same facts are visible on the page; the markup restates them in a form machines can read without guessing.

No. Google states that there are no additional technical requirements for AI Overviews or AI Mode, and that you don't need special markup or new machine-readable files to appear in them (Google Search Central). A page needs to be indexed and eligible to show a snippet. That's the bar.

This is worth saying plainly because a lot of advice claims otherwise. Adding more schema types won't make Google more likely to link your page in an AI Overview. Google's own guidance for those features points to content: helpful, specific, people-first pages. Our guide to how Google AI Overviews choose sources covers what does matter there.

They haven't published guidance that structured data affects which sources they cite. Their documentation focuses on crawler access: allowing their bots in robots.txt so pages can be read. Our AI crawler list covers each one.

Without public guidance, any claim that schema "boosts" citations in these engines is a guess. What we can say is that they read the page's text, and the facts in that text are what they repeat. If a price lives only in markup and not in visible text, you're relying on something no engine has promised to use.

For rich results in classic search, and for stating your facts in one consistent, machine-readable place. Both are worth having for their own sake.

Rich results. Product markup can show price and availability in results. Article, breadcrumb and organization markup help Google present your pages correctly. These are documented features with documented requirements (Google Search Central).

A single source of facts. When your Organization markup names your official website, logo and social profiles, and your Product markup states the current price, you have one place where the facts are explicit. That discipline tends to fix inconsistencies elsewhere, because someone has to decide what the true values are.

Cleaner pages. Teams that add accurate markup usually tidy the visible content at the same time: one product name, one price, one address. That consistency is what helps every engine, with or without markup.

The ones that describe what your pages really are, and that Google documents. For most businesses that means a short list.

Two changes are worth knowing. Google stopped showing HowTo rich results in 2023 (Google Search Central Blog), and since 7 May 2026 it no longer shows FAQ rich results for any site. FAQPage markup is still valid and does no harm, but it won't change how your pages appear in Google.

Short, accurate and consistent with the rest of your site. Here is an example for a fictional company:

{
  "@context": "https://schema.org",
  "@type": "Organization",
  "name": "Ledgerly",
  "url": "https://ledgerly.example",
  "logo": "https://ledgerly.example/logo.png",
  "description": "Invoicing software for agencies and freelancers in the EU.",
  "sameAs": [
    "https://www.linkedin.com/company/ledgerly-example",
    "https://x.com/ledgerly_example"
  ]
}

Every value here should match something a visitor can see: the name in your header, the logo on the page, the description in your about section, the profiles linked in your footer. The sameAs list ties your site to your official profiles, which helps systems tell you apart from companies with similar names. Keep it to profiles you control and keep them current.

Use Product markup with an Offer on every product page, generated from the same catalog data as the visible price and stock. Google documents the properties it uses for product results, including price, currency, availability, shipping and returns (Google Search Central).

The value for AI visibility is indirect but real: when your catalog, your product pages, your markup and your merchant feeds all state the same price, any system that reads them gets the same answer. A common reason an assistant quotes the wrong price is that one of those sources drifted out of date, not that markup was missing. Fix the drift first.

The main rule is simple: markup must describe what's visible on the page, accurately. Google's guidelines say not to mark up content that users can't see, and not to use markup that misrepresents the page (Google Search Central).

Practical habits:

• Generate markup from the same data as the visible page, so prices and names never drift apart.
• Validate it with Google's Rich Results Test after every template change.
• Remove markup for things you no longer offer.
• Don't add types that don't fit the page just because a tool suggests them.

Wrong markup is worse than none. If your Product markup still says last year's price, you've published a wrong fact in the most machine-readable form possible.

Into the visible page. The facts AI engines repeat come from text they can read: clear answers, specific numbers, honest limits and consistent names. Our answer-first content guide has a page template for exactly this.

A useful split for a small team: spend an hour making sure Organization, Product and Article markup are accurate and generated from your real data. Spend the rest of the week on the pages that should answer your buyers' questions. That ratio reflects where the documented value is.

For AI answers, there's no documented loss. For classic search, you give up rich results that markup makes possible, such as product prices or breadcrumbs in the results. Whether that matters depends on your site: an online store loses more than a B2B service firm with ten pages.

If your team is small and has to choose, choose the visible page first. A clear product page without markup can be quoted. Perfect markup on a vague page gives an engine accurate data about a page it has no reason to use. When you have time for both, add the few documented types that fit, generate them from your real data and move on.

The one exception is accuracy. If markup already exists on your site, check that it's right, because wrong markup states a wrong fact in the clearest possible form.

Run your main templates through Google's Rich Results Test and look at the enhancement reports in Search Console. Fix errors first, then warnings that affect features you care about.

Then compare the markup with the page. Does the price in JSON-LD match the price a visitor sees? Does the Organization logo point to your current logo? Does the author on the article match the byline? Mismatches are the most common problem, and the easiest to fix.

Schema markup helps search engines read your facts and earns rich results, but Google says it isn't needed for AI Overviews or AI Mode, and other AI engines haven't said they use it to pick sources. Keep a few accurate, documented types that match your visible content, then put your real effort into clear, specific pages. Ossvy's content studio builds briefs for those pages from the prompts you're missing.