How People Read

The assumption that people will read what you write, from top to bottom, carefully and completely, is the most dangerous assumption in content design. Eye-tracking research consistently shows that web readers scan rather than read. They skip large portions of text, anchor on specific visual cues, and make their decision about whether to engage further within a few seconds. Writing that ignores this behaviour is writing for a reader who does not exist.

The F-pattern

Jakob Nielsen’s eye-tracking studies identified what became known as the F-pattern: readers make a strong horizontal sweep across the top of a page, a shorter sweep lower down, and then scan vertically down the left edge. The result is an F-shape of high-attention areas, with the vast majority of the page receiving little or no reading.

The F-pattern is not a rule — it is a tendency that varies with content type, layout, and user motivation. But it reveals something important: readers do not give equal attention to all content on a page. They give disproportionate attention to the first line, the first words of subsequent lines, and the left edge of the content area. Content that is not in those high-attention zones is likely to be missed.

What readers actually fixate on

Readers do not scan randomly. They anchor on specific signals that suggest whether a piece of text is worth reading. These signals include:

Headings — a heading is a promise that the following content is about a specific topic. Readers scan headings to navigate, not to read. A heading that is vague, clever, or abstract fails to function as a navigation aid.

Bold text — words in bold within a paragraph attract disproportionate attention. Readers who are scanning will read bold words and decide whether to read the surrounding paragraph.

The first and last sentences of paragraphs — readers often read the opening sentence of a paragraph to assess whether the paragraph is worth reading in full, and skim to the last sentence to confirm. Burying the key point in the middle of a paragraph is a reliable way to have it missed.

Lists — bulleted and numbered lists draw the eye because of their visual structure. They signal discrete, enumerable content, which is easier to scan than continuous prose.

Writing for scanners, not readers

Writing for how people actually read means front-loading information. The most important content goes first: in the headline, in the first sentence of each paragraph, in the first item of each list. The structure of the content mirrors the structure of the reader’s attention — highest priority at the top, supporting detail below.

This is the inverted pyramid structure used in journalism: lead with the conclusion, then provide the evidence and context. For web content, it is the appropriate default. Write as if the reader might stop at any point, and ensure that what they have read so far is coherent and complete.

The exception: sustained reading

Readers who have committed to reading do not scan — they read. Long-form articles, tutorials, and documentation attract readers who have already decided the content is worth their time. For these contexts, the F-pattern assumption is less relevant; good prose structure, not just scan-friendly formatting, is what matters.

The distinction between scanning contexts and reading contexts should shape how you write. A product page, a knowledge base article title, an onboarding flow — these are scanning contexts. A tutorial, a case study, a detailed technical reference — these are reading contexts. The writing style, paragraph length, and formatting choices should differ accordingly.

Cognitive accessibility and reading behaviour

Scanning behaviour is not uniform. Users with dyslexia often read more linearly and rely more heavily on visual structure — short paragraphs, clear headings, consistent formatting — to maintain their place in the text. Users with ADHD may scan more rapidly and abandon content that does not signal relevance quickly. Users with lower literacy read more slowly and benefit from shorter sentences, plain language, and fewer abstract terms.

Writing for how people actually read is, in large part, writing for cognitive accessibility. The same structural choices that improve scanning — short paragraphs, front-loaded information, descriptive headings, minimal jargon — reduce the cognitive load for every reader, not just those with diagnosed conditions. WCAG 3.1.5 (Reading Level) at AAA requires that when content requires more advanced reading ability, a simpler alternative is available. At AA, WCAG 3.1.1 (Language of Page) requires that the human language is identified programmatically so assistive technologies use the correct pronunciation rules.

See plain language for the practical techniques, and the Accessibility hub’s cognitive accessibility section for the full context.

The next section examines writing craft itself: starting with plain language and why simplicity is a design value, not a compromise.