Module 2 · Lesson 4 of 8 · 2 min read

Usability testing: Watch five people try, and fix what trips them up

Research lesson 4: plan and run a small usability test. Write realistic tasks, use think-aloud, avoid helping, and turn observations into a prioritized list of problems.

Why test with people?

Analytics shows where people leave; surveys say why some of them think they left. Usability testing shows you, directly, where a real person gets stuck. It’s the fastest way to find problems your team can’t see because they know the site too well.

How many people?

For finding problems, small is fine. Testing with about five users typically uncovers most of the usability problems in a design, and several small rounds with fixes in between beat one big study (Nielsen Norman Group).

This applies to finding problems. It doesn’t tell you how common each problem is; analytics and surveys do that.

Write realistic tasks

A good task gives a goal and a reason, not instructions.

Weak task Better task
“Click on Shop, then Jackets.” “You’re going hiking next weekend and need a rain jacket under €150. Find one you’d buy.”
“Use the size guide.” “You’re between sizes. Decide which size to order.”
“Go to checkout.” “Buy the jacket and have it delivered to your work address.”

Avoid words that appear in your navigation labels; they hand participants the answer.

Run the session

Ask participants to think aloud: say what they’re looking for, thinking and expecting as they go. It’s one of the most valuable usability methods because it reveals misunderstandings as they happen (Nielsen Norman Group).

  • Don’t help. If they ask, reply “What would you do if I weren’t here?”
  • Don’t defend the design. You’re there to learn, not to explain.
  • Note behaviour, not opinions. “Scrolled past the size guide twice” beats “said it looked nice”.
  • Test on the device your customers use, usually a phone.

Turn observations into action

  1. List every problem you saw and how many participants hit it.
  2. Rate severity: blocker (couldn’t finish), major (finished with real difficulty), minor (annoyance).
  3. Fix blockers directly; they don’t need an A/B test.
  4. Turn major problems into hypotheses and check analytics for how many visitors they likely affect.

Recap

Five people, realistic tasks, think-aloud, no helping, severity ratings. Next lesson: heatmaps and recordings, to see how often the problems you found happen at scale.

Practical exercise

Write three tasks for your most important journey using the format in this lesson. Then ask one person who has never used your site to try the first task on their phone while thinking aloud. Write down every moment they hesitate.

Check your understanding

1. A participant is stuck and asks you where the button is. What should you do?
Show answer

Ask 'Where would you expect to find it?'. Helping hides the problem you're there to find. Turning the question back reveals their expectation, which is exactly the insight you need.

Sources (2)

  1. Why You Only Need to Test with 5 Users. Nielsen Norman Group. Accessed 14 November 2026.
  2. Thinking Aloud: The #1 Usability Tool. Nielsen Norman Group. Accessed 14 November 2026.

Lesson progress

Saved in this browser only. Accounts with synced progress and certificates are coming soon.

Next: Heatmaps and recordings →

Search lessons, guides, case studies and templates. Press Esc to close.