UX is not about looking good
There are apps with admired design that nobody can use. Only one thing measures the difference: did the user finish what they came to do?
"Make it look good" is a request we hear constantly. It is a reasonable thing to want, but it is not the question UX asks.
UX asks: did the user do what they came here to do, how long did it take, and how many times did they get it wrong?
A screen can be beautiful and the answer can still be no.
A concrete example: an address form
The pretty version: three columns, hairline rules, generous whitespace, animated transitions.
The usable version does these things:
- Picking a province fills the district list automatically
- The phone field accepts digits only and inserts the spacing itself
- The error appears under the offending field, not at the top of the form
- Pressing Save shows which field is missing without scrolling up
- Closing and reopening the app does not lose what was typed
None of these are visual choices. They are all answers to "did the user finish". And none of them show up in a screenshot — you only notice them by using the thing.
Where bad UX surfaces
It does not arrive as a complaint. It arrives like this:
- Support answering the same question over and over
- People abandoning the order flow at the same step every time
- Users saying "I didn't get the app" and going back to the website
- The field team still using WhatsApp instead of the app
The last one is the most common. If the team is not using the app, it does not mean the app is bad — it means the app is more effort than the habit it was meant to replace.
Design decisions vs visual decisions
These get conflated. They are easy to separate:
| Visual decision | Design decision |
|---|---|
| What colour is the button | How many buttons belong on this screen |
| Which typeface | Should the user be able to skip this step |
| Rounded corners | What happens when something fails |
| Which icon | Do we actually need to ask for this |
If the right column is wrong, a perfect left column changes nothing. If the left column is wrong, the product looks plain and still works.
Ideally both are right. The order matters: right column first.
Every question you don't ask is a win
The best improvement to a form is usually removing a field, not adding one.
For each field, ask: can the task be completed without this? If the answer is "we'll need it eventually", move it to later. Every extra question at signup is a user who might not sign up.
Is the date of birth genuinely required, or is it there so the data exists?
You cannot discuss UX without measuring
"I think this is better" gets you only so far. After that you need to look at which screens people linger on, where they drop out, and which errors appear most often.
Instrumenting this in the first version is far cheaper than guessing, six months later, why nobody uses the thing.
How we approach it
We work out the flow before drawing any screens. Designing an interface before it is clear what job the user is trying to finish gets expensive later — because at that point the thing that changes is not a colour, it is the order of the screens.