Structured data.
Structured data is an additional layer that labels a page’s content with a standard vocabulary machines can understand.
When a person reads a page they infer from context what each thing is: this is an author’s name, that is a date, this is a price. For a machine the same inference is guesswork. Structured data removes the guess and labels the information explicitly.
The vocabulary used is schema.org — a shared catalogue of types and fields developed jointly by the search engines. It defines hundreds of types, but in practice the ones that make a difference are few and vary by kind of site.
The layer’s most distinctive property is that it is invisible: it changes not a single pixel of the page. That is both an advantage and a risk — when it breaks, it gives no visual sign either.
The return runs two ways. On the search side it makes rich result presentations possible; on the answer engine side it declares what the page is about at the level of an entity. The second is becoming steadily more decisive.
One golden rule: everything written in the schema must also be on screen. Information that exists only in the schema is both a violation of the guidelines and something that puts the page at odds with itself — and when a model notices the contradiction, trust does not rise, it falls.
Which types to use varies by kind of site, and trying to use them all is a mistake. On a consultancy site the organisation, person, service, article and question-and-answer types provide nearly all the benefit; the rest is noise.
The criterion for choosing is this: does this type describe something that really exists on the page? A product type needs a real product, a price field a real price. Choosing a type in order to declare something that does not exist is taking the wrong road from the start.
- Syntax — can the block be parsed? Automated tools catch this.
- Required fields — are the fields the type demands filled in? Automated.
- Format — are date and number fields in the expected form? Automated.
- Identity links — are the identities referred to actually defined? Partly automated.
- Consistency with the screen — is what the schema says also on the page? HUMAN judgement.
- Meaning — does the graph describe the site correctly? HUMAN judgement.
The last two lines are covered by no general tool. Information that is in the schema but not on screen shows up green in a validator — because the tool reads only the block, not the page.