★ FREE TOOL · RUNS IN YOUR BROWSER

Schema Generator.
Valid JSON-LD,
without the empty properties.

Five types that cover most of what a business site needs: Organization, LocalBusiness, Service, Product and FAQPage. Blank fields are dropped rather than emitted as nulls, because a property with no value is worse than one that is simply absent. Nothing is sent anywhere.

Organization — Sitewide identity. One per site, usually on the homepage.

Still needed: Organisation name, Homepage URL.

Paste this into your page
<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Organization"
}
</script>

What schema actually does, and what it does not

Structured data tells a machine what a page describes rather than leaving it to infer that from the prose. It is the difference between a page that mentions a price somewhere in a paragraph and a page that states, in a form a parser can read without ambiguity, that this product costs this amount in this currency and is currently in stock.

What it does is make you eligible for result types that need that certainty, and help search engines and assistants resolve which entity you are. What it does not do is improve a page that is otherwise weak. Markup is a clarity mechanism, not a ranking lever, and adding it to thin content produces thin content that is easier to parse.

It is also worth being precise about AI search, because this is where most of the current selling happens. Google states that there are no additional requirements to appear in AI Overviews or AI Mode and no special optimisations necessary. There is no AI schema type. A proposal that includes one as a line item is charging you for a file.

The two rules that cause most of the problems

It must match the visible page. Structured data is supposed to represent content a visitor can actually see. FAQ markup for questions that are not on the page, a review score that appears nowhere, or a price in markup that differs from the displayed price are all violations rather than clever optimisations. This is the single most common problem we find, and it is usually accidental.

It must stay true. Markup written once at launch is correct for about as long as nothing changes. Prices move, stock runs out, opening hours change, staff leave. On a catalogue of any size, the only sustainable approach is generating structured data from the same source as the displayed content, so the two cannot disagree.

A third rule follows from both: delete markup you cannot maintain. Absent structured data costs you an eligibility you might not have used. Wrong structured data actively contradicts your own page.

Which type to use where

Organization is your sitewide identity and belongs once, usually on the homepage. Its real value is the sameAs list: the profiles that let a machine connect your site to everywhere else you appear. That connection is most of what entity resolution means in practice.

LocalBusiness replaces Organization where there is a physical location or a defined service area. The address and area served are the properties that matter; the rest is decoration. Keep them identical to what your Google Business Profile and your directory listings say, because disagreement between sources is what makes a business hard to resolve.

Service goes on the page describing that service, with a provider that references your Organization. One per page, matching the page, rather than a list of everything you offer repeated on every page.

Product is where accuracy matters most, because price and availability are checked and used. Generate it from your feed. If you cannot, consider omitting price and availability rather than publishing figures that will be wrong by Thursday.

FAQPage is worth including for machine readability even though Google has substantially reduced FAQ rich results in search. Every question must be visible on the page. If you want the version of this argument that applies to being quoted by assistants, it is in answer engine optimisation.

After you paste it in

Three checks, answering three different questions. Google’s Rich Results Test tells you whether Google can use the markup for a result type. The Schema.org validator tells you whether it is well-formed and whether you have used properties correctly. Search Console tells you whether anything broke on live pages after deployment, which is the one most teams forget to look at.

Then set a reminder to check again in three months. Structured data decays silently: no error appears when a price becomes wrong, and nothing visibly breaks when a staff member named in your markup leaves. It simply stops being true, which is the state that costs you the eligibility it was added to earn.

FAQ

Schema markup, asked and answered.

Want your whole schema graph checked?

Send the domain. The founder reviews it personally, including whether your structured data still agrees with your pages, and replies within 48 hours.

Apply for review ↗

Schema markup is structured data added to a page so machines can read what it describes rather than inferring it from prose. It can make a page eligible for rich results and it helps search engines and assistants resolve entities. It is not a ranking factor in itself, and adding it to a weak page will not make the page rank.

Not directly. It makes pages eligible for certain result types and it helps machines understand what a page is about, both of which can affect how your page appears and how often it is chosen. Treat it as a clarity and eligibility mechanism rather than a lever you pull for position.

No. Google states plainly that there are no additional requirements to appear in AI Overviews or AI Mode, and no special optimisations necessary. Valid schema helps for the ordinary reason that it makes content machine-readable, but there is no AI-specific schema type, and anyone charging you to implement one is charging for a file.

In a script tag with type application/ld+json, anywhere in the head or body. JSON-LD is Google’s recommended format precisely because it sits apart from your markup and does not need to be woven through the HTML. Keep one block per logical entity rather than merging unrelated types into a single object.

Far less often than it used to. Google substantially reduced FAQ rich results in search, so adding FAQPage markup in the expectation of visible stars and dropdowns will usually disappoint. It remains useful as machine-readable question-and-answer content, which is a different and more durable reason to have it.

Markup that does not match the visible page. Google requires structured data to represent content that is actually on the page, so FAQ markup for questions a visitor cannot see, or a price in markup that differs from the price displayed, is a violation rather than an optimisation. The second most common is markup that was accurate at launch and has since gone stale.

On any catalogue of real size, yes. Price and availability change constantly, and markup written once by hand is wrong within days. Generating it from the same source as the displayed data is the only way to keep the two in agreement, and disagreement between them is exactly what breaks rich result eligibility.

Use Google’s Rich Results Test for eligibility and the Schema.org validator for correctness, then check Search Console for errors reported on live pages. The three answer different questions: whether Google can use it, whether it is well-formed, and whether anything broke after deployment.

Yes, and often you should. A service page might carry Service, BreadcrumbList and FAQPage. Keep them as separate top-level objects or as an array rather than nesting unrelated types inside each other, and make sure each describes something genuinely present on that page.