R-04 Heading structure must reflect the real topic hierarchy

Why it matters

Headings are how readers scan a page and how crawlers and answer engines work out which part answers which question. If headings are stuffed with keywords or chosen for their size, people cannot find what they need and machines cannot separate one topic from the next.

How to implement

1. Give each page one H1 that names its single main topic.
2. List the real questions users have and turn them into H2s.
3. Add H3s only where an H2 has a genuine sub-topic.
4. Choose heading levels for meaning and use CSS for visual size.
5. Check that the text under each heading answers what the heading says.
6. Remove exact-match keywords that were added to headings only for search.

Checklist

• The H1 states a single main topic.
• H2s are split by real sub-questions.
• Heading order is chosen by semantic structure, not by visual size.
• Derive H2s from users' sub-questions.
• Use H3s only when there is a real sub-topic under an H2.
• The section under each heading must deliver what the heading promises.
• Do not add the exact-match keyword to every heading.

How to audit

• Read only the heading outline: is the logic of the page clear?
• Read only the heading outline: can the page's logic be understood?

Related controls

• R-05 Critical entity references must be explicit P1
• R-06 Tables must be used only for genuinely structured data P1
• R-07 URLs must be readable, stable and have a single canonical identity P2
• R-08 Internal linking must support important pages in context P1

Myth guards that reference this control

• “AI only likes 40 to 60 word chunks”

Sources

• [G2] Creating helpful, reliable, people-first content
• [G3] Optimizing for generative AI features on Google Search
• [B1] Bing Webmaster Tools: AI Performance