What schema markup is, and which types a business site uses

What structured data actually is

A page's words tell a person what it is about. Structured data tells a machine the same thing in labelled fields. This is a business, this is its name, this is its phone. This is the page's title, this is a question and this is its answer. The vocabulary is shared and published at schema.org.

It goes into the page as a small block of JSON-LD, a text format, inside the head. Visitors never see it. Crawlers read it before they read anything else, and the AI answer engines like ChatGPT and Google's AI answers use it the same way.

The types a business site uses

That is the whole list for most businesses. Service, Product and Review types exist, but they carry conditions and are easy to get wrong. Start with the six above and add others only when you have real data for them.

• Organization: your legal name, logo, phone, address and profiles.
• LocalBusiness: the same, plus hours and the area you serve, for a place people visit or call.
• WebSite: the site's name and address, once, on the homepage.
• WebPage: the page's own title, description and address.
• BreadcrumbList: the trail from the homepage to this page, in order.
• FAQPage: each question on the page and its plain answer.

What it does and does not do

Structured data removes guessing. It does not add words the page lacks and it does not decide a ranking. Google says it may use certain types to show a richer result, such as expanded questions under a listing. It says just as clearly that valid markup does not promise that.

So treat it as hygiene. Correct markup means the engine knows your phone number is a phone number. Wrong or missing markup means it guesses, and guesses are sometimes wrong.

The mistakes that break it

The first two are the ones Google's guidelines call out: markup must match what a person sees. The last one is the one that happens most, and it is silent. The page looks fine and the block does nothing.

• Markup that describes something not on the page, such as reviews you never show.
• A FAQPage with questions that are not actually visible to a visitor.
• Two Organization blocks with different names.
• A phone number in one format in the markup and another on the page.
• JSON with a missing comma, which makes the whole block unreadable.

How the builder emits it on every page

Website Builder Studio writes structured data for every page from what you told the in-app assistant: who you are, what you offer, where. Every page carries it, along with breadcrumbs a visitor can see. The questions on a page are written for a visitor to read, so the markup describes what is really there.

Because the markup and the copy come from the same answers, they agree with each other. Then a check that runs before anything publishes looks at the page, validated against Google Search Essentials and modern web standards. A page that fails does not publish, and the assistant says why.

When you change your phone number or your hours, tell the assistant once. The copy and the markup change together, because they were never two separate things.

How to test with the Rich Results Test

Open Google's Rich Results Test and paste a page address. It fetches the page, finds the structured data and lists what it found: the types, whether each is valid and which items are eligible for a richer result. Errors and warnings are named with the field that caused them.

Test the homepage and one of each kind of inner page: a service page, a location page, a page with questions. If the test reads nothing, the block is missing or broken. If it reads the wrong business name, the source of the data is wrong, not the markup.

Schema.org's own validator checks the same block against the full vocabulary, including types the Rich Results Test does not care about. Use it when you want to know the markup is correct rather than whether Google has a result format for it.