The first draft of a UI should embarrass you
The fastest path to a good interface runs straight through a bad one you're willing to throw away.
Most people show their first UI to nobody. They sit with a blank canvas, summon their taste, and try to render the finished thing in one pass. It never works, and worse, it's slow. The good interface is hiding behind a bad one, and the only way to get there is to build the bad one first and look at it honestly.
So the first draft should embarrass you. If it doesn't, you spent too long on it — you polished a guess instead of testing one. The point of draft one is not to be right. It's to become wrong in a way you can see.
A bad draft answers questions a blank page can't
When the screen is empty, every decision is theoretical. How long is the typical title? What happens with eleven items in that list, or zero? Does the primary action read as primary when it's sitting next to three other buttons someone insisted on? You can argue about these for an hour. A draft answers them in thirty seconds, because now they're sitting in front of you, ugly and specific.
I once spent a morning agonizing over the layout of a settings page in my head. The moment I threw the real fields onto a screen — unstyled, misaligned, default gray — the actual problem announced itself: there were too many settings, and no two of them belonged in the same group. No amount of pre-planning would have surfaced that. The bad draft did, in one glance.
That's the trade. A draft converts taste, which is slow and arguable, into evidence, which is fast and obvious. You stop defending opinions and start reacting to a thing.
The fear is attachment, not ugliness
The reason people resist the embarrassing draft isn't that they can't make one. It's that they're afraid of getting attached to it. And they're right to be — we defend what we've sunk time into, even when it's wrong. So the move is to make the draft cheap enough that throwing it away costs nothing.
- →Use the bluntest tool that shows the shape: a whiteboard, gray boxes, real text in a plain document.
- →Put real content in it immediately — fake data hides every problem that matters.
- →Give yourself permission, out loud, to delete it. Name it draft-one and mean it.
You can't iterate toward something good if you're too proud to have made something bad.
When the draft is cheap, the feedback gets honest. Nobody softens their critique of a gray box. The polish you'd have added would only have bought you more politeness and less truth.
Polish is a reward for being right, not a tool for getting there
There's a real cost to skipping the ugly stage: you spend your best craft on the wrong structure. I've watched teams ship a beautifully animated, pixel-perfect flow that nobody used, because the underlying model was wrong and the polish made it too expensive to question. The interface looked finished, so it became finished.
Refinement should come last, after the structure has survived contact with real content and real people. Color, spacing, motion, copy — these are how you make a right answer feel inevitable. Spend them on a wrong answer and you've gold-plated a mistake, then defended it because it was shiny.
Get the embarrassing version in front of someone fast. Watch where they hesitate. Throw it away. The good interface was never going to arrive in the first draft — it arrives in the third, after the first two taught you what the thing actually is.
