Figma for Product Teams · 16 min · 110 XP
Test it with five people
Why five people is usually enough, and how to ask without telling them the answer.
The reason small tests work is that usability problems are not evenly distributed. The first few participants hit the obvious ones, and each extra person finds fewer that are new. Nielsen Norman Group's long-standing rule of thumb is that around five users surface the large majority of problems in a given design, and that running two small rounds beats one big one, because you get to fix things in between. It is a planning heuristic, not a law: a product with genuinely different user groups needs a handful from each.
Give tasks, not opinions. “What do you think of this screen?” produces politeness. “You've been charged twice this month. Sort it out.” produces behaviour. Write the task as the situation the person is in, never as the steps they should take. The moment you say “go to Settings” you have tested nothing.
Leading: "Tap the Billing tab and cancel your plan."
Vague: "Have a look around the settings."
Good: "You've decided to stop paying for this.
Show me what you'd do."Then say almost nothing. Silence is the instrument. When someone pauses, count to five before you speak; most people resume and narrate the exact confusion you needed to hear. When they ask “should I click this?”, hand it back: “What would you do if I weren't here?” Answering the question ends the test.
Record where people hesitate, what they say just before they get stuck, and the words they use for things. That vocabulary is free interface copy. What someone predicts they'd do later, or how much they liked it, is the least reliable thing in the session.
Do it for real: write three task descriptions for your own prototype, check each one names a situation and never a control, and run them past one person today. Count how many times you were tempted to help.
Loading your workspace…