Introduction to Schema.org

An HTML tag says how something should look, not what it means. <h1>Avatar</h1> could be a film or a profile picture. Microdata — a set of attributes introduced with HTML5 — closes that gap, and Schema.org supplies the vocabulary Yandex, Google and Yahoo all read.

The three attributes

• itemscope — this block describes one entity.
• itemtype — what kind of entity, given as a URL such as http://schema.org/Movie. It goes immediately after itemscope.
• itemprop — this element holds a named property of that entity: director, genre, trailer.

A neutral <span> is the usual carrier for itemprop because it changes nothing visually.

Nested entities

A property’s value can itself be an entity. A film’s director is a Person with a name and a birthDate — expressed by adding itemscope after itemprop on the same element. This is what makes the vocabulary expressive rather than a flat list of labels.

The type hierarchy, and why it saves work

The broadest type is Thing, with four properties: name, description, url, image. More specific types inherit from their parents — Place is a kind of Thing, LocalBusiness is a kind of Place and a kind of Organization, inheriting from both. So choosing the most specific type that fits gives you every property of every ancestor for free.

The commonly used set: CreativeWork and its descendants (Book, Movie, MusicRecording, Recipe, TVSeries), the embedded media objects (AudioObject, ImageObject, VideoObject), Event, Organization, Person, Place and its descendants, and the commerce trio Product, Offer, AggregateOffer.

How to start on a real site

Pick the one entity the page is actually about — not every entity mentioned on it. A product page is a Product with an Offer; a service page is usually a Service or an Organization. Mark that properly, validate it, and stop. Markup that describes ten loosely related things describes nothing well, and only the supported types produce anything visible in Yandex anyway.