Semantic HTML

Design Systems & Engineering • UX & Product Architecture

Semantic HTML means choosing HTML elements that describe the purpose of content and controls. Elements such as headings, links, buttons, and form labels give browsers information about a page’s structure and how people can interact with it.

How does semantic HTML support accessibility?

A browser uses HTML semantics to expose information such as roles, names, and states through its accessibility tree. A screen reader can use that information to identify a button or navigate by headings. An is a link to a destination. A native

Example: A quote request

<form action="/quote" method="post">
  <label for="quote-email">Email address</label>
  <input id="quote-email" name="email" type="email" required>
  <button type="submit">Request a quote</button>
</form>

The label names the field, and the button text describes what happens next. In a real flow, test whether users can recover from validation errors and perceive the submission confirmation.

How does this help AI browser agents?

Some browser agents use the accessibility tree to inspect a page. The WebArena benchmark, for example, records accessibility-tree observations alongside the agent’s actions. An agent using that view can look for an email field and a “Request a quote” button. Other agents may inspect the DOM or work from screenshots, so test the full task with the agent you care about.

Search discovery is a separate issue. For Google Search’s AI features, a supporting page must be indexed and eligible to appear with a snippet. Google requires no special markup for these features, and it selects supporting pages separately for each response.

Sources