Structured data7 min read

JSON-LD is how AI answer engines know what you are

JSON-LD states your page's facts in a machine-readable form, so an answer engine can cite specifics — price, rating, author, availability — rather than guessing from prose.

Why it matters more for AI than for search

Classic search used structured data to decide whether you deserve a rich result. Answer engines use it differently: it is the difference between paraphrasing your page and quoting a fact from it.

A model reading prose has to infer that $49 is your monthly price. A model reading an Offer node with price: 49 and priceCurrency: USD knows it. Inference gets hedged; facts get stated.

The types that carry weight

Organization and WebSite establish identity — who you are, what you are called, where else you exist. Product with a complete Offer covers price and availability. FAQPage maps question to answer directly, which is exactly the shape an answer engine wants. Article with an author supports attribution.

Completeness beats breadth. One Product node with price, availability, SKU, brand and image is worth more than six half-filled types.

The @graph mistake

Multiple related entities usually go in one block under @graph. The catch: @context sits at the document root, and children inherit it. A validator that checks each node for its own @context will report perfectly valid markup as broken.

We hit this building our own checker — it flagged a major payments company's markup as invalid until we handled inheritance properly. If a tool tells you your @graph markup is invalid, verify before you rewrite it.

json

{
  "@context": "https://schema.org",
  "@graph": [
    {
      "@type": "Organization",
      "@id": "https://acme.com/#org",
      "name": "Acme Analytics",
      "url": "https://acme.com",
      "sameAs": ["https://linkedin.com/company/acme", "https://github.com/acme"]
    },
    {
      "@type": "WebSite",
      "@id": "https://acme.com/#site",
      "url": "https://acme.com",
      "publisher": { "@id": "https://acme.com/#org" }
    }
  ]
}

sameAs is underused

sameAs links your entity to its profiles elsewhere — LinkedIn, GitHub, Crunchbase, Wikipedia. It is how an engine corroborates that the Acme on your site is the Acme it has seen elsewhere.

For a company nobody has heard of yet, this is one of the cheapest credibility signals available.

Does this apply to your site?

The free checker runs this check and eighteen others, then generates the files to fix whatever fails.

Run the checker