Skip to content
foundry.

AI search

Schema markup for AI search: what helps and what doesn't

Labels for the facts you already show.

Schema markup for AI search is structured data, usually written in JSON-LD, that labels the facts on your page, such as your business name, address, services, prices and FAQs, so search engines can read them without guessing. It isn't a switch that gets you into AI answers: Google says AI Overviews need no special markup, and the assistants haven't said they read it.

But it helps the search engines AI tools draw on understand your pages, and it keeps your facts consistent. Add it to match what's visible, never instead of it.

Josef Winstone, Founder & Managing Director, Foundry Agency

By Josef WinstoneFounder & Managing Director · 7 min readUpdated

What is schema markup for AI search?

It's the same structured data used for normal SEO, applied with AI answers in mind. It tells machines, in a fixed vocabulary from schema.org, what each fact on your page is.

Your page might say "PERA Business Park, Melton Mowbray". A person knows that's an address. Schema says so outright: this is a LocalBusiness, this is its street address, this is its phone number, these are its opening hours. The usual format is JSON-LD, a small block of code in the page that Google recommends because it sits apart from the visible HTML.

Does schema help you appear in AI answers?

Indirectly. It helps search engines understand and trust your pages, and AI answers lean on those search engines. On its own, it won't get you named.

Google is clear that AI Overviews and AI Mode need no special markup: any page that's indexed and eligible for a snippet can be cited. Bing recommends structured data in its guidelines, and ChatGPT search leans on Bing. OpenAI, Anthropic and Perplexity haven't said whether their assistants read schema when they write an answer, and when an assistant fetches a page, it mostly reads the visible text.

So the honest position is this: schema is a strong supporting act. The words on the page do the main job. That's why we add it, and why we never treat it as the fix.

Which schema types matter most?

Seven types cover most small and mid-sized businesses. Use the ones that fit what the page actually is.

  • Organization: who you are, your logo, contact details and sameAs links to your official profiles. Sitewide.
  • LocalBusiness, or a more specific type such as Plumber or AccountingService: address, phone, hours and areas served.
  • Service: what a service page offers, who provides it and where.
  • Product with Offer: for shops, the product, price and stock.
  • Article: for guides and posts, with the author and dates.
  • FAQPage: the questions and answers on the page, word for word.
  • BreadcrumbList: where the page sits in your site.

What still matters on the page itself?

Structure. Schema labels the facts, but the HTML has to carry them first. These basics matter for search and AI alike.

  • One H1 per page that says plainly what the page is about.
  • H2s and H3s in order, never skipping a level, ideally phrased as the questions people ask.
  • A short, plain answer straight after each heading.
  • A unique title tag and meta description on every page.
  • Clean URLs that describe the page.
  • Key facts in the static HTML, not in images or text that only loads with scripts.

How do you add schema markup for AI search, step by step?

These are the steps we follow on every site. Most small sites take a day.

  1. List your page types: home, service, location, product, article, contact.
  2. Pick the schema type for each, using the most specific type schema.org offers for your business.
  3. Write down the facts once: name, address, phone, hours, services, areas, prices where you give them.
  4. Make sure every fact is visible on the page. If it isn't, add it to the page before you add it to the markup.
  5. Add Organization and LocalBusiness sitewide, with sameAs links to your Google profile and main directory listings.
  6. Add Service, Product, Article, FAQPage and BreadcrumbList on the pages they fit.
  7. Test each template with Google's Rich Results Test and the Schema Markup Validator.
  8. Generate the markup from the same content as the page if you can, so the two can't drift apart.

Is FAQ schema still worth adding?

Yes, but for understanding rather than looks. Google now shows FAQ rich results only for a small set of authoritative sites, so most businesses won't get the drop-down in results.

The markup still tells search engines which text is a question and which is the answer, which suits the way AI answers are built. Keep the questions real, the answers short, and the text identical to what's on the page.

What mistakes should you avoid?

Four come up on most sites we audit.

  • Marking up things that aren't on the page, such as reviews, prices or FAQs that only exist in the code.
  • Duplicate markup from a theme and a plugin, giving two different versions of your business.
  • Out-of-date facts: an old address or phone number in the schema long after the page was fixed.
  • Using the generic type when a specific one exists, such as LocalBusiness instead of Dentist.

What does LocalBusiness schema look like?

Here's a short example for a made-up plumbing business. Every fact in it is also visible on the page. Use our templates or generator for your own.

localbusiness.json
{
  "@context": "https://schema.org",
  "@type": "Plumber",
  "name": "Hollis Plumbing & Heating",
  "url": "https://www.example.co.uk/",
  "telephone": "01632 960000",
  "address": {
    "@type": "PostalAddress",
    "streetAddress": "1 Example Street",
    "addressLocality": "Leicester",
    "postalCode": "LE1 1AA",
    "addressCountry": "GB"
  },
  "areaServed": "Leicestershire",
  "openingHours": "Mo-Fr 08:00-17:30",
  "sameAs": ["https://www.example.com/hollis-plumbing-profile"]
}

How does Foundry handle schema?

On sites we build, schema is generated from the same content files as the pages: Organization and LocalBusiness sitewide, Service, FAQPage and BreadcrumbList on service pages, Article on guides. The markup can't say something the page doesn't.

On existing sites, we audit what's there, remove duplicates, fix stale facts and fill the gaps. Then we get on with the work that does most of the lifting: clear pages, crawler access, listings and coverage.

Our averages. Not our best days.

Our averages

  • 5days

    Website, kick-off to live

  • 20mins

    Average reply time

Schema for AI search FAQs

  • No. Google says no special markup is needed. Schema helps Google understand a page, as long as it matches what's visible, but the content does the main job.

  • OpenAI hasn't said. ChatGPT search leans on Bing, which supports structured data, so schema can help indirectly. When ChatGPT reads a page, it mainly uses the visible text.

  • JSON-LD. Google recommends it, it's easy to maintain, and it sits in one block rather than being woven through your HTML.

  • Wrong or misleading schema can. Marking up reviews or facts that aren't on the page breaks Google's guidelines. Accurate markup that matches the page is safe.

  • Not on its own. Google decides which rich results to show, and has narrowed several types. Add schema for clarity, not for a guaranteed look.

  • Run your pages through Google's Rich Results Test and the Schema Markup Validator. Search Console also reports errors for the types Google supports.

Related

More AI search guides

Want your facts read right? Get moving.

Tell us what you need. We'll reply within the hour (working hours), and most websites go live about 5 days after kick-off.

Book a callCall 01664 901025

We'll reply within the hour (working hours).

Or email hello@foundryagency.co.uk

Analytics cookies are off until you accept. We use them to see which pages help. Cookie notice