Schema Markup for AI: What to Implement and Why

Schema markup for AI is a way to hand machines a precise description of your business and content: who you are, what you sell, who wrote what, and how pages relate. The types that do the most work are Organisation, Article, Person, Product or Service, and FAQPage. The value comes from connecting them, not just adding tags.

Which schema types actually matter for AI search?

Not all schema is worth your time. One 2026 review found six types do most of the work: Article, FAQPage, HowTo, Organisation, Product and BreadcrumbList. Search Engine Land’s 2026 comparison of the best-positioned types for AI landed on a similar core set: Organisation, Article or BlogPosting, Person, Product or Service, and FAQPage.

Use this as your baseline. Organisation belongs on every site. Article or BlogPosting belongs on every article. Product or Service belongs on pages where you sell something. FAQPage belongs where you actually answer questions, and HowTo belongs where you describe a process. BreadcrumbList tells AI how the site is arranged and helps it attribute a page to its section.

If you are deciding where to start, put Organisation and Person in place first, then add the page-level types.

Why do AI systems need entity connections, not just keywords?

Search Engine Land’s 2026 analysis finds AI systems care most about entity definition, attribute clarity, and relationships such as offeredBy, worksFor, authoredBy and sameAs. When a page says ‘we offer bookkeeping’ without markup, the AI has to guess who ‘we’ is. With schema, the statement becomes explicit.

The properties that matter most are offeredBy, worksFor, authoredBy and sameAs. An Organisation block at the site root connects your name, logo and URL. A sameAs array connects that Organisation to its public profiles. Article schema with an author Person, a real bio URL and a publisher Organisation tells AI which human and which company stand behind the content.

This is why schema is different from keywords. Keywords suggest meaning. Schema states it. AI systems can cite a page more confidently when the entity behind it is unambiguous.

What does a minimal schema markup example look like?

Here is a minimal Organisation block in JSON-LD, the format most commonly recommended for AI-focused schema. Put it in the head of your homepage:

{
  "@context": "https://schema.org",
  "@type": "Organization",
  "name": "Your Business Name",
  "url": "https://www.example.com",
  "logo": "https://www.example.com/logo.png",
  "sameAs": [
    "https://www.linkedin.com/company/your-business",
    "https://www.facebook.com/your-business"
  ]
}

For an article, add the same author and publisher detail to the article page:

{
  "@context": "https://schema.org",
  "@type": "Article",
  "headline": "Example article title",
  "author": {
    "@type": "Person",
    "name": "Jane Smith",
    "url": "https://www.example.com/about/jane-smith"
  },
  "publisher": {
    "@type": "Organization",
    "name": "Your Business Name",
    "logo": {
      "@type": "ImageObject",
      "url": "https://www.example.com/logo.png"
    }
  },
  "datePublished": "2026-09-03",
  "dateModified": "2026-09-03"
}

Keep the markup truthful. Only mark up content that is visible on the page. If a box on your page says staff answer questions within a day, don’t mark it as FAQ unless the question and answer are actually there. That kind of mismatch teaches AI to ignore the markup.

Where should schema go and how do you check it?

JSON-LD goes in the head of each page. On a Squarespace, Wix or WordPress site, use the platform’s code injection or a plugin rather than editing theme files. One block per page, matched to the content on that page.

After publishing, run the URL through Google’s Rich Results Test and the Schema.org Validator. Both will tell you whether the markup parses. In Google Search Console, look at the page reporting to see if Google has detected the structured data. No rich result does not mean the schema is useless; AI systems can still read it. But a parse error means no one can.

If you want a wider check of how AI systems see your site, start with our guide to checking AI friendliness.

How does schema markup relate to an llms.txt file?

Schema and llms.txt answer different questions. Schema markup sits inside a page and tells AI what that page means. An llms.txt file sits at your site root and tells AI which pages exist and which matter for what. One is semantic, the other is a map.

They work well together. Schema gives an AI system confidence about your entity. llms.txt gives it a fast route to your best content. If you are prioritising, do the schema first, because it applies to every page and every AI system. Then add llms.txt if you want AI models to find your content without crawling the whole site.

We explain the difference in more detail in our post on what llms.txt is.

Common questions

What are schemas in artificial intelligence?

Schema is structured data that describes the parts of a page to machines. It uses a shared vocabulary, Schema.org, so Google and AI systems interpret the same fields the same way. Think of it as metadata with agreed meanings.

What is a schema markup example?

A simple one is an Organisation block: a @type, the business name, URL, logo and sameAs links to public profiles. Add Article blocks to posts with the author, publisher, headline and dates. The examples above show both.

How do I know if my website has schema markup?

View the page source and search for ‘application/ld+json’. If you find it, run the URL through the Schema.org Validator to see whether it parses. If you see no JSON-LD, there is no JSON-LD markup; older sites may use Microdata instead.

What are the types of schema markup?

Schema.org defines hundreds of types. For AI search, the ones that matter are Organisation, Person, Article or BlogPosting, Product or Service, FAQPage, HowTo and BreadcrumbList. Start with the first three and add the others where the content genuinely exists.

Is schema markup worth it for AI search?

Schema removes guesswork. It does not guarantee a position, but it gives AI systems clear statements about who you are and what your content covers. You can measure it by checking whether your marked-up pages appear for relevant queries.

If you want a second opinion on your own structured data, ask us through the site.