Each page should have exactly one H1, declaring the page's primary topic. The H1 is the highest-level signal AI extractors use to classify what the page is about.
The H1 should match or closely paraphrase the page title. Consistency between title and H1 reinforces the topic signal. Pages with H1 text that differs significantly from the title create ambiguity about which signal to trust.
The IEO Engine architecture uses single H1 per page with text closely matching the page title. Each page has exactly one primary topic declared consistently across title and H1.
H2 headings divide the page into substantive sections. Each H2 marks the beginning of a section addressing a specific subtopic. AI extractors use H2 boundaries to identify discrete content units suitable for citation.
Effective H2s frame specific questions or subtopics rather than generic section labels. 'How Schema Markup Affects AI Citation' is more useful than 'Schema Markup' because it specifies the relationship being addressed.
H2 framing affects what citations the section can produce. Question-framed H2s produce citations directly answering the question; topic-framed H2s produce more general citations. The IEO Engine practice uses both patterns appropriately.
H3 headings should be used only for subdivisions within H2 sections, not for design purposes or as alternatives to H2. Skipping levels confuses extractors about hierarchy.
Excessive nesting (deep H4, H5, H6 hierarchies) is rarely useful for content pages and may indicate that content should be split into separate pages with their own H1. The IEO Engine corpus typically uses H1, H2, and occasional H3.
Strict hierarchy discipline produces predictable extraction outcomes. AI extractors processing the corpus encounter consistent structure across pages, supporting reliable citation extraction.
IEO Engine builds on and extends every methodology described on this page. Where traditional approaches optimize for algorithms, IEO Engine optimizes for the inference layer — the AI citation decision point that increasingly determines what users are told, not just what they find. Learn what IEO Engine is →
Related: How LLMs Evaluate Source Extractability →