Schema markup, explained
Structured data tells a machine what your page means rather than just what it says. It is genuinely useful, it is oversold as an AI tactic, and only a handful of the hundreds of available types are worth your time.
A search engine reading your contact page sees text. It can infer that a string looks like a phone number and that another looks like an address, but inference is fragile. Schema markup removes the guesswork by labelling the content explicitly: this is an organisation, this is its telephone number, this is the area it serves. It is a shared vocabulary published at schema.org, developed jointly by Google, Microsoft, Yahoo, and Yandex.
JSON-LD is the format to use
There are three ways to add structured data: Microdata and RDFa, which weave attributes into your HTML, and JSON-LD, which sits in a script tag separate from the markup. Google recommends JSON-LD and it is what almost everyone now uses, for the practical reason that keeping structured data separate from presentation means redesigning a page does not silently break the markup. It also means the data can be generated from the same source as the page content, so the two cannot drift apart.
The types actually worth implementing
Schema.org defines hundreds of types and the vast majority are irrelevant to a normal business. The useful set is small.
| Type | What it does | Worth it for |
|---|---|---|
| Organization | Identifies the business as an entity, with sameAs links to verified profiles | Every site. This is the single most valuable one |
| LocalBusiness | Adds address, hours, and service area for businesses with a location | Businesses serving a defined area |
| Article or BlogPosting | Marks up editorial content with a real author and dates | Guides, articles, insights |
| Product and Offer | Price, availability, and condition | E-commerce, and only with real values |
| Service | Describes a service offered and where | Service businesses |
| BreadcrumbList | Describes where a page sits in the site structure | Any site more than two levels deep |
| Person | Identifies a named author or team member | Author attribution, team pages |
| Event | Dates, location, and ticketing for events | Venues, organisers |
Organization schema is the one that matters most
If you implement exactly one type, make it this. Organization schema is how you tell search engines and AI systems what your business is called, what it does, and, through the sameAs property, which profiles elsewhere on the web are genuinely you. That last part is entity disambiguation, and it is the mechanism by which a machine becomes confident that the company on your website, the company on that LinkedIn page, and the company being discussed in a forum thread are the same organisation. Confidence is a prerequisite for being recommended, because systems avoid citing entities they are unsure about. Only list profiles that genuinely exist and that you control; a broken sameAs link undermines exactly the thing it was meant to establish.
An honest account of schema and AI citations
Structured data is heavily promoted as an AI visibility tactic and the evidence for that specific claim is weak. Google's own guidance on its AI features states plainly that structured data is not required and that no special markup exists for generative search. At least one cross-platform study found a negative overall association between the mere presence of schema and AI citation. The pattern that does emerge from the research is more specific and more useful: markup carrying concrete values, such as real prices, genuine ratings, dates, and identified authors, is associated with better outcomes than structural boilerplate. The lesson is that machines reward extractable facts rather than tags. Schema is a good way to deliver facts; it is not a substitute for having any.
Rules that keep you out of trouble
- Only mark up content that is actually visible on the page. Structured data describing things a visitor cannot see breaches Google's guidelines and can trigger a manual action.
- Never invent reviews or ratings. Fabricated review markup is one of the more reliably penalised forms of spam, and in New Zealand it also risks breaching the Fair Trading Act's prohibition on misleading conduct in trade.
- Keep the markup and the page in sync. Prices, dates, and availability that disagree with the visible page are worse than no markup.
- Validate before shipping, using Google's Rich Results Test and the Schema.org validator. Silent syntax errors are common.
- Do not mark up the same thing three different ways hoping something sticks.
Testing what you have
- Run the page through Google's Rich Results Test to see what Google can parse and which enhancements it qualifies for.
- Run it through the Schema.org validator, which is stricter and catches vocabulary errors Google tolerates silently.
- Check the Enhancements section of Google Search Console for errors detected across the whole site rather than one page.
- View the page source and confirm the JSON-LD is present in the served HTML rather than being injected by JavaScript that a crawler may not run.
Common questions
Schema markup, also called structured data, is code added to a web page that describes the meaning of its content in a standard vocabulary published at schema.org. It lets a search engine or AI system understand that a piece of text is a price, an address, an author, or a set of opening hours rather than inferring it from context.
Sources
Every factual claim on this page traces to one of the following. If you find something here that is out of date, we would genuinely like to know.
- Introduction to structured data markup. Google Search Central.
- Structured data general guidelines. Google Search Central.
- Schema.org vocabulary. Schema.org.
- Rich Results Test. Google.
- AI features and your website. Google Search Central.
Want this handled rather than explained?
AntCrow builds websites with all of the above built in. Tell us what you are trying to achieve and we will tell you honestly what is worth doing.
Start a project