Definition
H2 coverage measures whether the page's heading hierarchy decomposes the buyer query into 4+ H2 subtopics. AI engines retrieve CHUNKS, not whole pages, and H2 boundaries are the dominant chunk delimiter their crawlers use. A page with 1 H1 + 6 H2 sections yields 6 retrievable chunks, vs a page with no H2s yielding one giant chunk that's less useful for sub-query retrieval.
Page on 'CRM for agencies' with H2s: 'What is a CRM', 'Features for agencies', 'Pricing tiers compared', 'Integration ecosystem', 'How to migrate'. Each H2 = one chunk an engine can independently cite.
How to optimize
Break each landing page into 4+ H2 sections aligned to query subtopics. Make H2 text question-form when possible (engines preferentially cite Q-form H2s for AI Overview snippets). VectorCite's h2-coverage signal scores this.