Principles
What people do before they think.
Psychologists have spent more than a century testing how people see, choose, remember and act. These are the 42 findings we lean on most when we design, in plain words, with how well each one holds up and what we do about it.
They're tendencies, not physics. Each one predicts what most people will do most of the time. When real visitors do something else, the visitors win.
42 principles in 5 groups.
The evidence
How sure we are, and how we keep it current.
Some of these are among the most repeated results in psychology. Some are rules of thumb that were never tested as experiments. Four are disputed: three famous findings that didn't hold up when other researchers went back to them, and an engineering rule its own field has turned against. Every principle shows its grade, the studies behind it, and the date we last checked.
We cite peer-reviewed research first, much of it from human-computer interaction and usability labs. When an idea started in an industry article, we cite that dated original, never a version that has been rewritten since, and we stay wary of any source that may have been written by AI.
Holds up
Tested many times, by different teams, with the same result.
16 principles
Holds, with limits
Real, but smaller than the popular version, or true only under certain conditions.
16 principles
Disputed
Attempts to repeat the original result have failed or come out mixed.
4 principles
Rule of thumb
A sound observation or design maxim, never set up as an experiment you could fail.
6 principles
When the research moves, so do we.
We go back through the research behind every principle at least twice a year, and sooner when a major review or replication comes out. When the evidence changes, the page changes, its review date moves, and the change goes in this log.
Change log
Checked every source against the publisher's record and every number against the paper itself. Fixed the details that didn't match (among them the jam study's figures, the first name on the peak-end meta-analysis, the Pareto land story, and the 400 milliseconds that the Doherty paper never names), replaced the last industry article with a peer-reviewed study, and added research so every principle but Tesler's Law rests on at least two peer-reviewed sources.
Tightened the sources. Peer-reviewed UX research now backs nine principles that leaned on one study or on industry articles, including work on web forms, progress bars, manuals, menus and banner blindness. Industry articles that keep being rewritten were replaced with peer-reviewed studies or with the dated original.
Added twelve principles, including a new group, Researching, on how we find out what people really do. One of them, the Hawthorne Effect, is disputed: the famous factory data turned out not to show it.
Added the evidence for every principle, graded on four levels, with sources and good and bad examples. Nine pages now say where our view differs from the popular version. The biggest changes: Choice Overload and the Zeigarnik Effect didn't hold up as usually told, Miller's seven is closer to four, and Postel's Law is now disputed in its own field.
First edition: thirty principles in four groups, each with what we do about it.
Tradeoffs
Where they pull against each other.
- Jakob's Law and the Von Restorff Effect
- Be familiar where people act: navigation, forms, checkout. Be distinctive where people feel: the opening screen, the product's own world, the image that gets shared.
- Hick's Law and Tesler's Law
- Cutting options is good until the work just moves onto the user. If removing a setting means people have to work it out themselves, nothing got simpler.
- Miller's Law and what people need
- A long list that's well grouped beats a short one that hides what people came for. Chunk it before you cut it.
- The Aesthetic-Usability Effect and testing
- Polish raises satisfaction scores and hides failures. Count who finished, and treat compliments as a nice extra.
- The Goal-Gradient and Zeigarnik effects and calm
- Progress cues motivate. A screen full of unfinished-business reminders turns into a guilt machine. One honest progress signal at a time.
Guardrails
Lines we don't cross.
Several of these describe levers that work just as well against the visitor as for them. We only pull them in the visitor's favor.
- 01
No fake progress
No made-up percentages, no head starts that didn't happen, no countdown timers, no invented scarcity.
- 02
No padded waits
If something takes time, it's because the work takes time, and the screen says what's happening.
- 03
No bias used against people
No decoy prices, no guilt-trip buttons, no add-ons ticked in advance, no made-up reviews or user counts.
- 04
Access is part of the rule
Never color alone. Motion that switches off when people ask. Targets big enough for every hand. Input in whatever form people type it.
A second look
Is your product breaking one of these?
Most teams can't tell from the inside. Send us the page or the flow that's not working, and we'll tell you which of these we'd look at first.
Email hello@synergyprod.com