Section density

What this check does

Counts H2 and H3 headings on the page, divides total body word count by that number, and flags pages where sections drift outside the 150–250 words-per-section sweet spot. Both extremes get flagged:

• Under-segmented (e.g. 1,800 words under two H2s) reads as a wall of text.
• Over-segmented (e.g. an H2 every 50 words) fragments ideas and breaks scanning rhythm.

The check only runs on long-form pages (body > ~600 words). Short pages don’t need a TOC.

Why it matters

Nielsen Norman’s eye-tracking studies have shown for two decades that web readers scan in an F-pattern: they read the first line of each section, skim the rest, and stop when nothing catches their eye. Headings are the anchors that decide whether they keep going.

Google’s Helpful Content guidance asks the same question from the other side: can a reader find what they came for without reading every paragraph? Structured long-form ranks better than unstructured long-form because:

• Featured snippets extract from sections with clear headings.
• AI Overviews and answer engines cite passages that sit under a descriptive H2.
• Dwell time improves when readers find their answer fast and stay to read more.

Over-segmentation hurts the opposite way: when every paragraph has its own heading, no heading means anything. The TOC becomes noise and the page reads as outline notes someone forgot to finish.

How to fix it

1. Identify paragraphs that deserve their own H2.

Walk the page top-to-bottom and ask of each paragraph: “If a reader skipped here from the TOC, would this stand alone?” If yes, the paragraph wants a heading. If no, it belongs under the heading above.

A practical test: write a one-line summary of each paragraph. Group consecutive paragraphs whose summaries answer the same question. Each group becomes one section.

2. Convert long paragraphs into bulleted sub-sections.

Wall-of-text paragraphs often hide three or four points doing different jobs. Pull them apart:

• Lead sentence → H2 or H3.
• Each supporting point → its own bullet or short paragraph.
• Caveats and edge cases → final short paragraph under the section.

This is the same pattern the paragraph length check enforces at the paragraph level. Section density is the same idea one zoom level out.

3. Use H3s for sub-points inside long sections.

If an H2 section runs past ~300 words, look for natural sub-divisions. Each one becomes an H3. Keep the heading hierarchy clean — H1 → H2 → H3, no skipping to H4.

4. Don’t over-segment to game the metric.

An H2 every 80 words doesn’t help anyone. Headings exist to mark genuine topic shifts. If two adjacent sections answer the same question, merge them — even if it pushes one section past 300 words. Real structure beats artificial structure.

5. Pair with content depth, not against it.

The fix for thin sections isn’t always more headings. Sometimes the content itself is too thin. See content depth — if a section reads as filler, deepen the content first, then re-check density.

Frequently asked questions

What’s the right words-per-section number for my niche?

Roughly 150–250 words per H2 for general long-form content. Technical reference docs and step-by-step tutorials can run shorter (80–150) because each step is its own atomic unit. Narrative essays, opinion pieces, and case studies can run longer (250–400) because the argument needs room to develop. The metric is a guardrail, not a rule.

My listicle has 20 short items, each an H2. Is that wrong?

No. Listicles are a recognized format where each item earns its own heading by design. The check exempts pages where the heading pattern is clearly enumerative (e.g. “10 ways to…”, numbered H2s). If you’re being flagged on a genuine listicle, the issue is usually that intro and outro sections need their own H2s too.

Does Google actually look at section density directly?

Not as a numeric ranking factor anyone outside Google has confirmed. What Google reliably uses: heading text for relevance, headings to extract featured snippets, and overall page structure as a signal of editorial care. Pages with sensible density tend to win those extraction battles. The metric is a proxy for skimmability, which is the thing that actually matters.