Design principles lists tend to be adjectives — be clear, be consistent, be simple. Nobody sets out to be unclear, so the advice changes nothing.
What follows is six principles written as decisions you can actually make on a Tuesday afternoon, with the specific mistake each one prevents.
1. One page, one job
Before designing anything, write down the single action you want the visitor to take. If you cannot pick one, the page will present three options of equal weight and the visitor will pick none.
This does not mean one link on the page. It means one action that is visually louder than everything else. Secondary actions exist; they just look secondary.
Prevents: the homepage that tries to serve new visitors, existing customers, job applicants and investors simultaneously, and serves none of them.
2. Hierarchy is a decision, not a style
Every element on a page is competing for attention. If you do not decide the order, the largest and most colourful thing wins by accident — which is why so many pages shout their least important content.
The test: squint at your page until it blurs. What stands out? Is that what should stand out? It is a two-second check and it catches an enormous amount.
| Tool | Strength | Use for |
|---|---|---|
| Size | Strongest | The one thing that matters most |
| Weight | Strong | Headings, key numbers |
| Colour | Strong, but overused | The primary action only |
| Whitespace | Underrated | Isolating something important |
| Position | Reliable | Reading order |
3. Consistency beats cleverness
If a button looks a certain way on one page, it looks that way everywhere. If a term means something in one place, it means the same thing everywhere — pick “projects” or “workspaces” and never use both.
Users build a model of your interface within about a minute. Every inconsistency forces them to rebuild it, and rebuilding feels like effort even when they cannot name why.
Prevents: the site where the same action is called Save, Update and Apply on three different screens.
4. Show the state, always
Every interactive element should make three things obvious: that it can be interacted with, that it has been interacted with, and what happened as a result.
- Hover and focus states so people know something is clickable before they click.
- Loading states so a slow response does not look like a dead button. Anything over a second needs one.
- Success and error states that say what actually happened, in words.
- Disabled states that explain why — a greyed-out button with no explanation is a dead end.
Prevents: the double-submitted order, which is almost always a missing loading state.
5. Write the interface, do not decorate it
Most usability problems are writing problems. “Submit” tells the user nothing; “Create account” tells them exactly what will happen. “An error occurred” is a dead end; “That email is already registered — sign in instead?” is a solution.
Two rules that fix most interface copy:
- Buttons say what happens. Verb plus object, from the user’s point of view.
- Errors say what to do next. Not what the system experienced.
6. Design the empty and broken states first
Everyone designs the screen with twelve perfect items in it. Real users arrive at zero items, or at the state where something failed. Those are the moments that decide whether they continue.
An empty state should explain what goes here, why it is useful, and offer the action that fills it. A broken state should say what happened and what to try. Both are more important than the populated view, and both are usually done last, badly, by whoever is left.
This connects directly to the recovery paradox — the moment something goes wrong is where satisfaction is actually decided. We went into that in the five frictions.
Common questions
Do these apply to internal tools?
More than to public sites, because internal users cannot leave — they absorb the cost every day instead. Hierarchy and state feedback matter most; visual polish matters least.
How do I argue against a bad design decision?
Not with taste. Name the principle and the consequence: “there are three primary buttons here, so the eye has nowhere to land.” Specific and checkable beats “I don’t like it.”
Should we use a design system?
If more than a couple of people build interfaces, yes — it is how consistency survives turnover. For one person, a documented set of colours, spacing and button styles is enough.
What about following trends?
Trends are worth knowing and rarely worth chasing. The principles above have not changed in twenty years; the visual fashion around them has changed five times.
Leave a Reply