Assume nobody reads

Most people won’t read your screen. They glance at it, spot something that looks like the next step, and tap it. If that tap is right, the design worked. If it only works for people who read the instructions, it didn’t.

So I design every flow around one rule. The happy path has to work for someone who reads nothing.

People skim

In 2008, Harald Weinreich and three colleagues logged every page 25 people visited over several months (Weinreich et al., ACM Transactions on the Web). Jakob Nielsen used their 45,237 page views to estimate how much people read. On an average page, people have time to read 28% of the words at most. About 20% is more likely (Nielsen Norman Group).

Eye tracking points the same way. When nothing on a page guides the eye, people read the first line or two and then scan down the left edge in an F shape (Nielsen Norman Group).

People have done this for decades. In 1987, IBM researchers John Carroll and Mary Beth Rosson named it the paradox of the active user. People skip the manual and start clicking, even when reading it first would save them time (Nielsen Norman Group).

The rule in a real flow

I recently designed a sign-up for a physician-led health practice. Patients fill it out on a phone in the waiting room, and it’s built to take about two minutes. These are the choices the rule led to.

Answers that make sense on their own. If someone skips the question, the answers should still tell them what to pick. “More energy” and “Better sleep” work without a heading. “Option A” doesn’t. I test this by covering the question and reading only the answers.

A way out of every question. An “I’m not sure” answer keeps people from stopping to look something up, or quitting because they don’t know. Survey researchers have debated this option for decades. A 2020 study of 1,282 people found that most who chose “don’t know” weren’t hiding an opinion, and it recommended keeping the option (Wetzelhütter, Survey Practice).

Rules before mistakes. Show the password rules or date format before people type, so they don’t learn them from an error. In 2014, researchers at Google and the University of Basel improved three real sign-up forms with guidelines like this and tested them with 65 people. On one form, the time to finish dropped by a third. The number of people who got it right on the first try went from 9 of 32 to 24 of 33 (Seckler et al., CHI 2014; Google Research summary).

Defaults that help people. People who don’t read keep whatever is already picked. In an online experiment by Eric Johnson and Daniel Goldstein, 82% of people agreed to be organ donors when that was the default. Only 42% agreed when it wasn’t (AMA Journal of Ethics summary). Pre-select what most people want. Never pre-select something that helps you more than them.

One obvious next step. Each screen has one main button. Anything else, like “Learn more first,” looks clearly secondary. A quick glance finds only one thing to tap.

Reading is a bonus

Some people do read every word, and they should find real detail when they do. The flow just can’t depend on it. If the happy path works for the person who skims, it works for the person who reads too.

I test my own flows by blurring my eyes. I click whatever feels right and see where that lands me. If I end up somewhere I didn’t mean to go, the screen only works for people who read it.

Then I hand the flow to someone else and ask them to go as fast as they can. I watch where they pause. Each pause marks a spot where the design asked them to read.