Semantic HTML for AI: Title, Meta Description, lang, and Landmarks
A browser can make any <div> look like a heading or a menu. A machine can’t see the styling — it reads the tags. Semantic HTML means using tags that say what each part is (<title>, <main>, <nav>, <article>), so an AI can tell your headline from your sidebar. It’s low-effort and high-leverage.
The four that matter most
<title>— the page’s name. It shows in browser tabs and search results, and it’s the first thing a machine reads about the page. Make it specific and unique per page.- Meta description — a one-sentence summary in the
<head>. It isn’t a ranking factor, but it’s often the text shown or summarized for your page. langattribute —<html lang="en">(orja,es, …) tells a machine what language the page is in, so it doesn’t have to guess.- Landmarks —
<main>,<nav>,<header>,<footer>,<article>mark the regions of the page. They let a machine focus on the main content and skip the chrome.
A copy-paste skeleton
<!doctype html>
<html lang="en">
<head>
<title>Your specific page title</title>
<meta name="description" content="A one-line summary of this page.">
</head>
<body>
<header><!-- site name, nav --></header>
<main>
<article><!-- your real content --></article>
</main>
<footer><!-- ... --></footer>
</body>
</html>The common mistake
A page built entirely from <div>s. It can look identical to a human and be a structureless wall to a machine — no clear main content, no language, sometimes no real title. You don’t need every semantic tag; you need the few that mark out what matters.
The bottom line
Semantic HTML is the cheapest way to be understood: a unique <title>, a meta description, a lang attribute, and real landmarks around your content. It’s one of the “be understandable” steps in our GEO overview.