Heading hierarchy

The hierarchy rules

• One H1 per page — the page’s primary topic. Should match (or closely echo) the title tag and natural language phrasing of the target query.
• H2 for major sections — the page’s primary subdivisions. AI engines often extract these as candidate questions in FAQ-style synthesis.
• H3 for sub-sections within an H2
• H4-H6 rarely needed except in very long documents

Why it matters in 2026

• AI engines parse heading structure to identify candidate “answer to X” sections
• Featured Snippets preferentially trigger when headings are formatted as natural questions
• AI Overviews extract heading-bounded sections more cleanly than free-flowing prose
• Accessibility (WCAG) requires logical heading order; screen readers navigate by headings
• User scannability — readers skim headings before deciding to read

Common mistakes

• Multiple H1s on a page (acceptable in HTML5 semantically but still penalized by SEO standards)
• Skipping levels — H1 → H3 confuses screen readers and crawlers
• Using headings for styling — making text big with H2 just because it’s styled prominently
• Title tag and H1 identical — wastes signal; H1 should be slightly different framing
• Sales-pitch H2s instead of descriptive ones — “Get started!” vs “How the program works”

Heading patterns that work for AI search

Heading style	Effect
Natural-language question (“What is X?”)	Triggers FAQ + Featured Snippet candidacy
Comparison (“X vs Y”)	High AI Overview citation rate
Numbered list signaling steps (“3 ways to…”)	HowTo schema eligibility
Definitional (“X explained”)	Generic but reasonable
Sales copy (“Get started today”)	Wastes the slot — Google doesn’t surface these