Designing pages for scanning, not reading start to finish

Most visitors to a business website scan a page looking for the specific piece of information they need, rather than reading it start to finish the way they’d read an article or a letter. A page written and laid out like an essay — an opening paragraph, a developed argument, a conclusion — fights against that behaviour, however good the writing actually is. It assumes a reading pattern most visitors simply don’t use.

This isn’t a comment on attention spans or a complaint about how people read online. It’s how reading a screen for a practical purpose has always worked, and designing around it rather than against it is one of the highest-leverage, lowest-cost changes available on most small-business sites.

Why scanning, not reading, is the default behaviour: the F-pattern and the layer-cake pattern

A visitor arrives at a page with a specific question — what do you charge, do you service my suburb, can I book online. Their goal is to answer that question as fast as possible, not to appreciate the page as a piece of writing. Eye-tracking research on how people read web pages has repeatedly found scanning patterns rather than linear reading. Visitors’ eyes move in an uneven pattern across headings, bolded text and the first few words of paragraphs, hunting for the answer. They only slow down to read closely once they’ve found the section that looks relevant.

Two named scanning patterns show up repeatedly in that eye-tracking research.

The F-pattern

The F-pattern describes how users scan a page of dense paragraph text. There is a full horizontal sweep across the top line or two, a shorter sweep partway down, and then a vertical scan down the left edge looking for a word or heading worth stopping for — a shape that, plotted on a heat map, genuinely resembles the letter F.

The layer-cake pattern

The layer-cake pattern is more relevant to a well-structured page. It describes users skipping straight to headings and subheadings, reading each one, and only dropping into the paragraph beneath it when the heading itself looks relevant. It treats the page as alternating layers of heading and content, rather than one continuous block. A page built around clear, front-loaded headings is deliberately designing for the layer-cake pattern, instead of leaving a visitor to fall back on the F-pattern’s much less effective sweep of undifferentiated text.

Designing for that means the structure of a page has to carry meaning on its own, skimmed, without requiring every word to be read to be understood.

What actually supports scanning: headings, subheadings and how users scan

Front-loaded headings that say the point, not tease it

A heading like “Fast, reliable service” tells a scanning reader nothing distinguishing. A heading like “Same-day quotes, most jobs booked within a week” tells them the actual fact before they’ve read a single sentence of body copy.

Short paragraphs, one idea each

A paragraph that mixes two or three separate points forces a scanning reader to read the whole thing to extract the one point relevant to them. A paragraph with a single idea can be skipped entirely, correctly, by a reader who doesn’t need it.

Bullet points for genuinely list-shaped information

Prices, inclusions, service areas and steps in a process are naturally list-shaped, and forcing them into paragraph form makes a scanning reader do the work of re-extracting the list themselves.

Bold used sparingly, on the words that matter most

A handful of bolded phrases per page act as a second, faster layer a scanning reader can read on their own, skipping the surrounding sentence structure. But this only works if bold is rare enough to still draw the eye. See typography for readability for why overusing weight defeats this.

What actively works against scanning

Pattern	Why it fights scanning
Long, unbroken paragraphs	No visual entry point for a reader looking for one specific fact
Vague, “clever” headings	A scanning reader can’t tell what’s under a heading without reading the whole section
Important information buried mid-paragraph	Scanners read beginnings and skip middles; a caveat placed there is functionally hidden
Dense navigation with unclear labels	Forces a reader to read every menu item rather than recognise the one they need

Scanning patterns and density are the same underlying problem

A page that’s too dense fights scanning for the same reason it fights reading generally. There’s no visual rest between ideas, so a scanning eye can’t tell where one section ends and the relevant one begins. White space and content density covers the layout side of this argument directly. The two pages describe the same underlying fix from different angles: structure, and the space around it, doing the work that dense unbroken text can’t.

Scanning behaviour and the words themselves

Designing for scanning sets the container. The words still have to be written to fit it. A heading that’s structured to be scanned but written vaguely still fails, and a bullet list of genuinely undifferentiated filler still reads as filler no matter how well it’s formatted. Website copywriting covers how the actual sentences on a business site should be written, including for scanning specifically. The two disciplines need to be solved together. Good structure around weak content, or good content in poor structure, both fail the same visitor for different reasons.

Where this matters most on a business site

The homepage and any page with a clear commercial job — a service page, a pricing page, a contact page — carry the highest cost when scanning fails. A visitor who can’t quickly find the answer to “do you do this” or “what does it cost” simply leaves, rather than reading further to find out. Informational or about-style content can tolerate a slightly more narrative structure, because a reader choosing to be there has already signalled more patience. But even there, front-loading the point, rather than building to it, respects how the page will actually be used.

What to check on your own site for scannability

Pick your busiest page and give someone thirty seconds to find one specific fact on it — the price of your most common service, or whether you cover a particular suburb. If they can’t find it in that time, the fault is very rarely that the fact isn’t on the page. It’s that the page was structured to be read, not scanned, and the fix is a restructure, not a rewrite.