Your website has a second version written for search engines.

We recently added JSON-LD to the list of things we check on customer websites. It can reveal a clean, accurate description of a business. It can also reveal old addresses, duplicate identities, missing details, and claims the visible page never supports.

JSON-LD is code inside the page

JSON-LD stands for JavaScript Object Notation for Linked Data. Despite the name, it is usually not a file a customer downloads or opens. It normally sits inside the page's HTML in a script markedapplication/ld+json.

The visible page might say “Known Space,” while the JSON-LD labels that name as an Organization. A service page might show a phone number and hours, while the code identifies them as telephone andopeningHours. The words stay the same. The labels make their meaning explicit.

A website builder may add it. That does not mean the job is finished.

The idea that every template includes JSON-LD is wrong. So is the idea that no template does. Wix, Squarespace, and Shopify generate structured data automatically for supported content. The coverage is different on each platform. Wix's automatic business markup covers the primary location only, while Shopify's strongest default is product information.

WordPress output depends on the theme and plugins in use. GoDaddy and Square vary by builder and site configuration. Figma Sites and Framer provide places to add custom code. Webflow can generate schema with AI or accept it manually, but the owner still has to take that step and publish it.

This is why we inspect the rendered page. The platform name cannot tell us whether the right markup is present, whether it describes every location, or whether its facts still match the business.

Why we check it

Search engines say structured data gives them explicit clues about a page. Correct markup can also make a page eligible for specific search features. That is useful, but only when the code describes what is actually on the page.

Google requires structured data to represent the visible content truthfully. In one web-scale study of Dataset markup, 61% of hosts using the label did not describe datasets. Google researchers had to compare the markup with the page content instead of accepting the label as fact.

Good JSON-LD gives search systems a cleaner description of the page. It cannot rescue a vague or inaccurate page.

What reliable JSON-LD looks like

1. It describes the page people can see.The business name, services, location, hours, prices, and claims agree with the visible content.
2. It uses the most specific honest type.A plumber can use Plumber. A general business should not borrow a narrower type for attention.
3. It includes the fields required for the feature being pursued.There is no universal completeness score. Google documents requirements feature by feature.
4. It connects repeated entities with @id.A stable identifier can show that the publisher, organization, and referenced business are the same entity.
5. It appears in the rendered page.Google can process JSON-LD added with JavaScript, but the markup must be present when the page is rendered.

Where it helps recommendations and browser agents

JSON-LD is not a shortcut to the top of a search result. Its value is that it replaces guesswork with explicit facts. A recommendation system can read the business type, services, location, hours, reviews, offers, and relationships between pages without having to infer every detail from the layout and prose.

That matters beyond traditional search. Ora includes JSON-LD and Schema.org markup when it evaluates whether an agent can access and understand a website. Browser agents benefit from the same clear entity information. It gives them a cleaner answer to basic questions such as what the company does, where it operates, and which page represents which service or location.

JSON-LD and WebMCP handle different parts of the same job. JSON-LD describes the business and its content. WebMCP can expose structured actions that an agent can take inside the browser. One helps an agent understand and recommend the business. The other helps it act. A website prepared for both is easier for an agent to read and use than one that leaves everything buried in visual presentation.

None of this makes the markup a guarantee. One matched study of 1,885 pages found no measurable citation lift from adding JSON-LD alone. The useful conclusion is not that JSON-LD does nothing. It is that clean structured data works as part of a larger evidence system: accurate pages, consistent business details, credible proof, crawlable content, and agent-ready actions.

A check cannot promise that Google will show a rich result or that an AI assistant will cite the page. It can show whether the website gives those systems a clear, consistent set of facts to work with.

{
  "@context": "https://schema.org",
  "@type": "Article",
  "headline": "Your website has a second version written for search engines.",
  "description": "How JSON-LD helps search systems and browser agents understand a business, and why Known Space checks it on customer websites.",
  "datePublished": "2026-09-03",
  "dateModified": "2026-09-04",
  "author": {
    "@id": "https://www.knownspace.space/#organization"
  },
  "publisher": {
    "@type": "Organization",
    "@id": "https://www.knownspace.space/#organization",
    "name": "Known Space",
    "url": "https://www.knownspace.space"
  },
  "mainEntityOfPage": "https://www.knownspace.space/insights/what-json-ld-tells-search-engines"
}