Designing for restraint

Every stakeholder wants their feature above the fold. The interface that results belongs to nobody, and it serves nobody either.

There is a specific kind of meeting where a product slowly becomes unusable. Six people, each responsible for one area, each entirely reasonable about their own request. Marketing needs the banner. Support needs the help entry point visible. Sales needs the upgrade prompt. Nobody is wrong, and the sum of them is a screen where the user cannot find the one button they came for.

Hierarchy is subtraction

You cannot create emphasis by adding emphasis. If three things are bold, nothing is. The only way to make one element louder is to make the others quieter — which means every decision to promote something is really a decision to demote something else. Teams rarely frame it that way, and that framing is most of the battle.

Ask what the user is here to do, then ask what everything else on the screen is costing them.

A test that works in the room

When a new element is proposed, we ask one question out loud: what are we willing to remove to make space for this? Not rhetorically — we expect an answer, and we write it down. Roughly half of requests quietly withdraw at that point. The other half are genuinely important, and now we know what they are worth.

The version of this that fails is the compromise: shrink everything slightly so it all fits. That produces a screen where six things are medium-loud and the user's eye has nowhere to land. Compromise is the enemy of hierarchy, not its solution.

What restraint is not

It is not minimalism as a style. Hiding necessary controls behind a hamburger to keep a page looking clean is not restraint — it is the same problem wearing better clothes. Restraint means the important thing is obvious and the secondary thing is one predictable step away. Sometimes that produces a busy-looking screen, and that is fine.

The measure is never how empty the page looks in a screenshot. It is whether someone who has never seen it knows where to go within a couple of seconds. We test exactly that, with five people, before anything ships.