How do itemscope and itemtype identify an HTML microdata item?

The Citead team, updated

In short. itemscope creates the item, while itemtype supplies the vocabulary type URL that gives its properties context.

Definition: itemscope creates an HTML microdata item from an element and the name-value pairs associated with it. itemtype gives that item one or more vocabulary type URLs, which provide the context for its properties.

Key facts

How does itemscope create the item?

The element carrying itemscope is the boundary for a new item. The WHATWG specification defines the attribute as boolean, so its presence is what matters. The resulting item is a group of name-value pairs. Descendant elements can provide properties through itemprop.

The boundary is independent of the element's visual role. A div, section, heading or other HTML element can carry the attribute because the standard says every HTML element may have itemscope. Markup without the microdata attributes has no effect on the microdata model.

What does itemtype identify?

itemtype qualifies the item created by itemscope. The attribute belongs on the same element, and its value is parsed as unique tokens separated by ASCII whitespace. Each token must be a valid absolute URL, and the value must contain at least one token. The URLs must be defined to use the same vocabulary.

The type URL supplies the context for property names. A property such as name receives its meaning from the vocabulary selected by the item type. The standard describes type URLs as opaque identifiers. Unknown type URLs are not dereferenced or deconstructed by user agents to decide how to process the item. itemtype also requires itemscope on the same element.

Schema.org's guide shows the usual pattern with a shared vocabulary:

<div itemscope itemtype="https://schema.org/Movie">
  <h1 itemprop="name">Avatar</h1>
  <span>Director: <span itemprop="director">James Cameron</span></span>
</div>

Here, itemscope creates the item, itemtype identifies it as a Schema.org Movie, and itemprop names properties within that vocabulary. The type URL is an identifier used by the vocabulary; the WHATWG specification says type URLs should not be automatically dereferenced unless the vocabulary specification says otherwise.

Where can the markup appear in a website?

Google's structured data documentation, last updated 10 December 2025, lists Microdata among the formats that Google Search supports when the markup is valid and properly implemented for the documented feature. It describes Microdata as an HTML specification that nests structured data within HTML content and says it is typically used in the body element, while allowing use in the head element.

The same documentation says structured data is in-page markup on the page to which the information applies. It also says not to create an empty page only to hold structured data or add markup about information that is not visible to users. The Schema.org guide supplies the vocabulary examples; Google's feature documentation determines Google's requirements for a particular Search feature.

FAQ

Is itemscope enough to identify a vocabulary type?

No. It creates the item, but itemtype supplies the type URL and vocabulary context. Schema.org's guide presents the two attributes together for a Movie item.

Can itemtype contain several URLs?

Yes, when the types are unique absolute URLs and all use the same vocabulary. The WHATWG specification permits a space-separated list and requires at least one token.

Does a type URL have to be fetched?

No automatic fetch is required by the HTML microdata rules. The WHATWG specification treats item types as opaque identifiers and says unknown types are not dereferenced by user agents.

Sources

  1. WHATWG, "HTML: The Living Standard, Microdata", 19 October 2026
  2. Schema.org, "Getting Started - schema.org", 16 September 2026
  3. Google Search Central, "Intro to How Structured Data Markup Works", 10 December 2025

Check how a page reads to answer engines. Open the GEO score analyzer

Test a page and see what to fix first

The analysis usually takes less than 10 seconds and needs no account.

400,000+ pages scored in the public benchmark