Most people do not read a web page. They scan it, decide within a few seconds whether it is worth more attention, and then either read a little more or leave. This is not a claim about attention spans. It is a measured behaviour, recorded with eye tracking equipment, repeated across thousands of pages and two decades of studies.
If your website has to do work before a sales conversation starts, it is worth knowing what that scanning behaviour looks like. It changes where you put things.
In this article
The study everyone cites
In 2006, the Nielsen Norman Group recorded how 232 users looked at thousands of web pages. The researchers were not asking people what they read. They were tracking where the eye landed, in what order, and for how long.
The dominant pattern looked like the letter F. Readers moved horizontally across the top of the content area. They dropped down and moved across again, over a shorter distance. Then they scanned down the left edge of the text.
The finding held across different sites and different tasks. It was not a quirk of one design.
What the pattern tells you
The F shape itself is the least useful part of the finding. Nielsen Norman Group has since revisited the research and made the point directly. The F is not a law that readers obey. It is what scanning looks like when a page gives the eye nothing better to do.
The underlying behaviour is simpler than the shape. People read the beginnings of things. First lines get read. First words of those lines get read. Everything after that gets sampled.
That has three consequences for a page.
The first few words of a heading carry most of the weight. “Reduce manufacturing downtime” and “We help you reduce manufacturing downtime” contain the same promise. Only one of them puts it where the eye lands.
Text that starts with the same phrase repeatedly gets skipped. Five service headings that each begin with “Our approach to” give the scanner nothing to distinguish them. The eye reads the first, samples the second, and stops.
Formatting is not decoration. Subheadings, short paragraphs and bold text are what allow a scanner to find a reading entry point. A page of even, unbroken paragraphs is a page a scanner cannot enter.
The fold, and what replaced the argument about it
The other question this research settles is how far down the page anyone looks.
Nielsen Norman Group’s scrolling and attention study measured 130,000 eye fixations across 120 people. Users spent about 57% of their viewing time above the fold, and 74% of it within the first two screenfuls.
Two numbers, two different conclusions, and people usually only take the first one.
The first conclusion is that the top of the page matters more than anything below it. That much is obvious and most sites already act on it.
The second conclusion is the useful one. Attention does not stop at the fold. It declines sharply and then keeps going. A quarter of viewing time happens below the second screenful. People do scroll. They just scroll with less and less attention as they go.
So the practical rule is not “put everything above the fold.” It is that the further down something sits, the more the reader has already decided. Content low on a page is read by people who are already interested. That is the right place for detail, method and proof. It is the wrong place for the sentence that explains what you sell.
Reading resolved into an F-shaped path across 232 users: a pass along the top of the content, a shorter pass lower down, then a scan down the left edge.
Nielsen Norman Group, F-Shaped Pattern For Reading Web Content, 2006
What the research does not say
Two misreadings of this work are common enough to be worth naming.
The first is that the F-pattern is a template. It is not. It is a description of what happens when a page fails to direct attention. A page with strong visual hierarchy produces a different pattern, because the eye follows the hierarchy instead of falling back on scanning. The F is the default behaviour, not the goal. If your page produces a clean F, that is a signal the design is not guiding anyone.
The second is that scanning means people are unwilling to read. The same research shows long reading sessions on pages that earn them. Scanning is how a reader decides whether to invest. It is a filter, not a verdict. The job of the layout is to survive the filter, not to compress everything into the part that gets scanned.
There is also a limit worth stating plainly. Eye tracking tells you where attention went. It does not tell you what the person concluded, whether they believed it, or whether they were going to buy. It is evidence about mechanics, not about persuasion. Treating it as more than that is how teams end up optimising layout while the underlying message stays unclear.
The left edge, and what it means for navigation
The vertical stroke of the F is the part most often ignored. Readers scan down the left edge of the content when they are looking for a place to re-enter the text.
That behaviour explains why certain layout choices cost more than they appear to. Centre-aligned body text removes the consistent left edge the eye is scanning against, which makes re-entry harder. Long paragraphs with no visual break give the left-edge scan nothing to catch on. Bulleted lists work well partly because every item starts at the same left position.
It also explains why the first word of a navigation item matters as much as the first word of a heading. A menu is read as a vertical or horizontal list of first words. “Solutions”, “Services” and “Offerings” are three labels that occupy the same mental slot and distinguish nothing. A buyer scanning for evidence that you handle their situation is reading for nouns they recognise.
Applying this to a B2B page
B2B pages have a specific problem. The buyer is often not one person. Six to ten people may touch the decision, and they arrive with different questions. A technical lead is checking whether the work is competent. A finance stakeholder is checking whether the company is stable. Someone is checking whether you have done this before in their sector.
Scanning behaviour means each of those people is looking for their own entry point, quickly, on the same page.
That leads to a layout principle that is easy to state and hard to hold to. Every section should be identifiable from its first few words.
In practice that means a few specific things.
- Write headings that describe content, not headings that create suspense. “How the audit works” tells a scanner what is underneath it. “The Matsio difference” tells them nothing.
- Front-load every heading and every opening sentence with the words that carry meaning.
- Break the page into sections a reader can skip. A scanner who can skip cleanly reads more, not less, because skipping is cheap.
- Put the proof a specific stakeholder needs in a section that stakeholder can find. Certifications, security details and client references should be locatable without reading the whole page.
There is a test for whether a section is findable. Cover the page and show a stakeholder only the headings. Ask them where they would click to answer their own question. If a finance stakeholder cannot find the section that addresses stability and references, that section is not written for them, however good its content is.
Where this shows up in results
The value of this research is that it removes a category of argument from the design process. Whether a heading should be clever or clear is not a matter of taste once you know that only the first two or three words are reliably read.
Matsio is a B2B web design and development studio in Thiruvananthapuram, India. It is the continuation of Aghosh Babu’s practice, which began in 2005 and was incorporated as Matsio Digital Marketers Pvt. Ltd. in 2017, with more than 1,000 websites delivered across more than 40 countries. Across those projects, the pages that perform are consistently the ones that are easiest to scan, not the ones that are most carefully composed. 85% of the studio’s clients return for further work, measured across every engagement since 2005. That difference comes from treating layout as a question with evidence behind it rather than a question of preference.
What to do with this
You can test your own pages against the research without any equipment.
Open a page. Read only the first three words of every heading, in order, top to bottom. Then read only the first sentence of every paragraph.
What you have just read is close to what most visitors take in. If that reduced version does not explain what you do, who it is for, and what happens next, the page is not doing its job. The rest of the words are not fixing it, because the rest of the words are not being read.
Fix it in this order. Rewrite the headings so the first words carry the meaning. Rewrite the opening sentence of each section so it can stand alone. Then, and only then, look at the body copy.
Two related pieces cover the next part of this. First impression research explains how quickly the judgement about a page is made, and heatmap research covers what you can and cannot learn from attention data on your own site. If you want the version of this that applies to machine readers rather than human ones, how AI assistants read websites covers the same problem from the other side.
The short version
People scan before they read. They read the beginnings of things. Attention drops sharply below the fold but does not stop.
None of that is new. It has been measured, repeatedly, since 2006. The reason it still matters is that most B2B pages are still written to be read from the top left corner to the bottom right, by someone with no other options.
Nobody reads that way. Building the page as though they do is the most common and most expensive layout mistake in B2B.
A small thing and a big thing
One small thing to fix on your website today, and one big thing to learn that gets you more leads.
Your information is
safe with us. No spam.