Size, weight, contrast, position and whitespace decide what a person sees first — whether you planned it or not.
Every screen already has a hierarchy. The only question is whether it’s the one you chose. When everything is the same weight, nothing reads as important, and the eye stalls. Good hierarchy makes one thing win first — ideally the one action you want taken — and it can do that quietly, with whitespace and restraint, instead of shouting.
Below is a plan card. The numbers show where the eye lands — 1 first, 2 next, 3 third — from a simple attention model. Right now the price wins and the Start free trial button is nearly last. Move the sliders to make the button win first, and try to turn the goal meter green without cranking everything to the top.
pricing card · primary action = “start free trial”
The primary action isn’t winning the eye yet.
The button is ranked 5th of 5 — equal weight with everything else, so it disappears.
Attention isn’t magic; it’s roughly additive. A rough model weighs five levers and ranks the elements by the total. Whitespace (isolation) is the quiet one — it lifts attention without adding noise.
attention ≈ 0.25·size + 0.17·weight + 0.20·contrast + 0.26·space + 0.12·position
start — primary button, everything low:
0.25(.30) + 0.17(.35) + 0.20(.30) + 0.26(.30) + 0.12(.35)
= .075 + .060 + .060 + .078 + .042 = 0.315 → 5th of 5
price tag (fixed, big and bold):
0.25(.90) + 0.17(.80) + 0.20(.70) + 0.26(.60) + 0.12(.60) = 0.729 → 1st
give the button whitespace + moderate size/weight/contrast:
0.25(.76) + 0.17(.70) + 0.20(.73) + 0.26(1.0) + 0.12(.35)
= .190 + .119 + .146 + .260 + .042 = 0.757 → 1st, loudness only 0.73
The lesson in the last line: you can win the eye and still be calm. Whitespace did most of the lifting, so size, weight and contrast never had to max out.
| Any screen with one job | Decide the single primary action, then make it win first. One primary action per screen. |
| A quick critique | Squint until detail blurs; whatever still stands out is your real hierarchy. If nothing does, that’s the problem. |
| The limit | This is a model of attention, not an eye-tracker. Use it to reason and catch obvious faults — then confirm with real users. |
In an interview you’re asked to critique a signup screen: two same-size green buttons, Start free and Book a demo, side by side. Don’t say “I don’t love it.” Say what the model says: with two primary actions at equal weight, neither wins, so the eye stalls and users hesitate. Your fix — make Start free the single primary through size, weight and surrounding whitespace, and demote Book a demo to a quiet text link. One clear first action. That reasoning, not taste, is what the interviewer is listening for.
A landing page has a bold headline, a bold price, and a bold button — all the same weight. The problem?
Cheapest way to make one button win attention without shouting?