Getting better at product judgment

I can usually tell when an interface feels awkward before I can explain why. That first reaction is useful, but it isn't enough to base a review on.

“The spacing feels wrong” gives someone very little to work with. “The gap above this heading is smaller than the gap below it, so it looks attached to the previous section” points to something they can inspect.

That's the part of taste I'm interested in developing: noticing a problem, finding its cause, and deciding whether it matters.

Compare the same task

Browsing attractive screenshots is useful for collecting visual references. It tells me less about how a product behaves.

A better exercise is to complete the same task in two products. Create an event, invite a colleague, change a payment method. Notice what each asks you to remember and where you need to stop and interpret a control.

I pay particular attention to what happens after the obvious part. Can I undo the action? If a field is wrong, does the error preserve what I typed? When I come back tomorrow, can I find the result?

A product with plain styling can be very good at these details. A beautiful screen can make them difficult.

Give the reaction a reason

When something feels better, I try changing one thing at a time.

If a page seems crowded, I can hide a secondary label before changing every spacing value. If a transition feels slow, I can shorten it without also changing the distance and easing. That makes it easier to understand which decision improved the result.

I also try the uncomfortable content: a long title, no image, an error message, several lines of translated text. A layout that only works with carefully chosen copy hasn't answered the whole design problem.

This process won't reduce every judgment to a measurement. It does make feedback more useful than a preference presented as a rule.

Other people notice different problems

A colleague may see a broken hierarchy I stopped noticing. A customer may struggle with terminology that feels obvious to everyone on the team.

When reviewing together, I want to know what they looked at first and what they expected to happen. Watching where they hesitate often helps more than asking whether they like the design.

There's room for disagreement. Someone who uses a tool once a month may need visible labels; someone using it all day may value density and shortcuts. Good judgment includes recognizing whose task the decision serves.

Know when refinement is avoiding the work

I've spent too long adjusting screens that were already usable. There's always another spacing value to tune, especially when the next task is less comfortable.

Before another pass, I ask what problem the change would fix. Poor contrast, a misleading label, and a control that's hard to tap deserve attention. A barely different shadow may be worth keeping, but it doesn't necessarily deserve to delay the release.

I don't use a percentage like “ship at 85%.” That hides the question of which part is unfinished. An optional view can wait. A data-loss bug can't.

Faster output makes review more important

AI can generate layout options and suggest improvements. It can also reproduce familiar patterns that don't fit the task. A polished result still needs checking against the actual content, users, and constraints.

I use references and generated alternatives as material to inspect. Why does this arrangement work? What disappears at a narrower width? Is the clever interaction still comfortable on its twentieth use?

The useful skill is being able to answer those questions, explain the trade-off, and revise the decision when using the product proves it wrong.