How people read online: F-pattern and beyond — Nielsen Norman Group

Source

Nielsen Norman Group — the world's most-cited UX research lab, founded by Jakob Nielsen and Don Norman. Their eye-tracking and behavioural studies set the baseline most enterprise UX teams design against.

Original research: How People Read Online (eye-tracking studies)

Key findings

1. People do not read web pages — they scan them. The seminal 1997 finding (Nielsen) holds in 2025. F-pattern scanning still dominates left-aligned, content-heavy pages.
2. Eye-tracking shows roughly 79% of users scan; only 16% read word-by-word.
3. Headlines, the first two paragraphs, and bolded keywords get the most fixations. Everything else is glanced at, not read.
4. Card-grid layouts produce a "spotted" pattern — readers fixate on visually distinct elements (logos, numbers, icons) and skip the surrounding text.
5. AI-search overlays change the pattern again. When users see an AI-generated summary at the top of a SERP, the F-pattern collapses into a "block-and-bounce" pattern: read the summary, scan the page once for confirmation, leave.

How this changes landing pages in 2026

• Put the answer in the first sentence of every section.
• Bold the noun the buyer is searching for.
• Use card grids — but make sure the visually-distinct element on each card carries the message (a number, a logo, a one-line claim — not a stock illustration).
• Write H2s in question form so AI summarisers can quote them verbatim.

Practical impact

For a typical SaaS landing page, applying just the first three findings (answer-first paragraphs, bolded keywords, card-grid hierarchy) tends to lift comprehension scores 30–40% in 5-second tests, which is a leading indicator of conversion lift.