Write for Scanners. Readers Take Care of Themselves.

There is a version of your writing that nobody has ever read: the version word by word, in order, the way you experienced it while typing. Your visitors met a different artifact. They met the headings, the first two words of each line, the links, and whatever their four working memory slots could hold while scrolling with a thumb.

The gap between those two versions is where most websites quietly lose their audience, and almost every fix is structural.

TLDR Mode

This is the checklist piece of the How Humans Read Websites series. The science lives in the overview and its working memory and F-pattern deep-dives; this page is the part you can apply tonight.

The skim is the product

Eye-tracking says 79 percent of visitors scan new pages, and an average visit reads at most 28 percent of the words. Sit with what that means for a second: the skim is not damage done to your carefully written page. The skim is what your page actually is, for most of the people who meet it.

Which reframes the writing job entirely. You are not writing an article that some visitors will skim. You are writing a page that must survive being read out of order, in fragments, by someone half-distracted, and still deliver its argument.

That is a design constraint, and like most constraints it produces better work once you accept it.

The checklist

Seven rules, in the order I would apply them to a page that is fighting its readers:

1. Front-load the first two words. Scanners’ fixations cluster on line beginnings, so the opening of every heading, list item and paragraph carries the most weight. “Pricing: what a five-page site costs” survives a skim. “What you should know about pricing” dies at the first two words, which were “What you”.
2. Make every heading summarize standing alone. Headings get read out of context, in sequence, as a summary of the whole page. Write them so that reading only the headings tells the story; if the heading sequence has a gap, no paragraph below it can repair what the skim already missed.
3. One idea per paragraph, three or four lines. Working memory holds roughly four chunks, and a paragraph that opens three ideas at once spends the whole budget before the reader has agreed to. Short paragraphs are not a stylistic preference. They are the correct response to the size of the workspace.
4. Emphasize almost never. Bold and color work by directing attention, which means every extra bold trains the eye to trust the next one less. Use emphasis the way a road uses a brake light: rarely, and only where the cost of missing it is real. This post, for the record, contains none.
5. Write links that name their destination. “Click here” and “learn more” carry no information scent, and scentless links get scanned past by exactly the people who wanted what was behind them. Describe where the link goes in the visitor’s own words; information scent has its own piece, because it decides more clicks than any other words on the page.
6. Keep lines between 45 and 75 characters. Long lines lose the eye at the return sweep; short lines fragment the reading rhythm, and the research-backed range sits around 50 to 75 characters for comfortable reading.

That last constraint is holding this entire site together on one line of CSS, which pleases me deeply:

.prose {
  max-inline-size: 66ch;
  line-height: 1.65;
}

Two declarations. No framework required.

Does scannable mean shallow?

No, and the confusion comes from mixing up depth with density. Depth is how much a page genuinely covers; density is how much work it takes to extract any of it. Scannable writing removes the density and keeps the depth: summary first, details a click away, long-form sections that open with their conclusion.

That way the reader who already committed can go as deep as they like, and the one who hasn’t still leaves with the argument.

Mortimer Adler called the shallow version inspectional reading and considered it a skill worth a whole chapter: systematically pre-reading a book’s structure before deciding where to invest a full read. The web made inspectional reading the default for everyone.

Writing for scanners is just respecting that the skill your readers already have is the skill your page should be built for.

There is also the finding I enjoy most, because it removes the last excuse: plain, scannable writing helps experts too. Nielsen Norman’s studies found specialists complete tasks faster when the reading load drops. Experts don’t want your dense paragraphs. They want your dense paragraphs with the filler removed.

For the “my content is too complex” corner

Complexity lives in the ideas, not in the fog around them. A genuinely complex topic rewards structure more than a simple one does, because the reader navigating difficulty needs the handrails more, not less. If your argument only survives in unbroken paragraphs, that is not evidence of sophistication.

It is evidence the argument has never been asked to survive a skim, and your visitors are running that test on it daily.

That closes the series: the overview post holds the science, and the four deep-dives cover the F-pattern, working memory, banner blindness and information scent.

If reading this one produced a specific page in your head, the one you already suspect would fail its own skim, that instinct is worth acting on. Bring it to me. Structural edits are my favorite kind of job.

Sources & further reading

1. Morkes, J., & Nielsen, J. (1997). Concise, SCANNABLE, and Objective: How to Write for the Web. CHI 97 conference proceedings.
2. How Users Read on the Web — Nielsen, 1997
3. How Little Do Users Read? — Nielsen Norman Group, 2008
4. Cowan, N. (2001). The Magical Number 4 in Short-Term Memory. Behavioral and Brain Sciences, 24(1).
5. Readability and Line Length — Baymard Institute (summarizing the screen-reading research)
6. Plain Language Is for Everyone, Even Experts — Nielsen Norman Group