Skip to content
NewTemplates & toolkits are on the way — join the waitlist
Hecham Gazhi

UX Design · Lesson 10 / 13

Usability testing

The complete guide to usability testing: types of tests, how many participants, writing tasks, moderating sessions, measuring results, rating severity and reporting findings that lead to action.

6 min read · 5 quiz questions

What is usability testing?

In a usability test, representative users try to complete realistic tasks with a product or prototype while an observer watches. The goal is to find where people struggle, why, and how serious it is.

It is fundamentally different from asking people for opinions. You don't ask "Do you like it?" — you watch what they do. People are often polite in what they say, but their behaviour is honest.

Usability testing finds problems in a design. It doesn't tell you whether people need the product at all — that's the job of discovery research.

Types of usability tests

TypeHow it worksProsCons
Moderated in-personFacilitator and participant in the same roomRich observation, body language, flexible follow-upsSlower, needs a location
Moderated remoteVideo call with screen sharingReach people anywhere, still interactiveTech issues, less context
Unmoderated remoteParticipants complete tasks alone with a testing tool that records themFast, cheap, many participantsNo follow-up questions, less depth
Guerrilla testingQuick tests with people in a café or public placeVery cheap and fastParticipants may not match your users

How many participants?

Jakob Nielsen's widely cited guideline is that testing with about 5 users per round finds most of the common usability problems in a design. The idea is that the first few users reveal the biggest issues, and each additional user mostly repeats what you've already seen.

The key is iteration: test with 5, fix the problems, then test again with 5 new people. Three small rounds usually teach you more than one big round of 15.

  • If you have very different user groups (e.g. patients and doctors), test each group.
  • For quantitative usability studies that report metrics with confidence (e.g. success rates for benchmarking), you need larger samples — often 20 or more per group.

Writing good tasks

Tasks are the heart of a usability test. A good task is a realistic scenario with a clear goal, written in the user's language.

Weak taskBetter task
Click 'Plans' and choose Pro.Your team of 8 designers needs shared files. Find a plan that works and tell me what it would cost per month.
Use the filter to find red shoes.You need running shoes for a race next week, under $100. Find a pair you'd buy.
Go to Settings and change your notification preferences.You're getting too many emails from the app. Make it stop sending you promotions.

Do

  • Give context and a goal, not instructions.
  • Use the user's words, not labels from the interface.
  • Make the end point clear so you know when the task is complete.
  • Order tasks naturally, from simple to complex.

Don't

  • Name the button or menu item they need (it gives away the answer).
  • Write tasks that depend on each other so one failure blocks the rest.
  • Include more tasks than fit in the session — 5 to 8 is typical for 45–60 minutes.
  • Ask "Would you…" hypothetical questions as tasks.

Running a moderated session

How to respond when participants ask for help

  • "What would you do if I weren't here?"
  • "What are you looking for?"
  • "What do you think will happen if you click that?"
  • Repeat their question back: "You're wondering whether it saved?"
  • Only step in if they're completely stuck and the session can't continue — and note it as a task failure.
  1. 1Welcome and consent. Explain that you're testing the design, not them: "If something is confusing, that's our fault, and exactly what we need to find."
  2. 2Explain think-aloud. Ask them to say what they're looking at, thinking and trying to do as they work.
  3. 3Warm-up questions about their background and how they currently do this task.
  4. 4Tasks. Read each task, give them a printed copy, and let them work. Stay quiet and observe.
  5. 5Follow-up after each task. "How did that go?", "What did you expect to happen when you clicked that?"
  6. 6Wrap-up. Overall impressions, anything they'd change, and thanks.

Common mistake: Helping, explaining or defending the design. The moment you explain how something works, you've lost the chance to see whether people can figure it out alone — which is what real users will have to do.

What to measure

MetricWhat it tells you
Task successDid they complete the task? (Success / partial / fail)
Time on taskHow long it took, compared across designs or rounds
ErrorsWrong paths, mis-taps, incorrect inputs
Path / directnessDid they go straight there or wander?
Post-task ratinge.g. Single Ease Question: "How easy was this task?" on a 7-point scale
Post-test questionnairee.g. System Usability Scale (SUS)
Observations and quotesWhy problems happened — usually the most valuable data

Rating severity

Not all problems are equal. Rate each finding so the team fixes the most important ones first. A common scale:

SeverityMeaningExample
CriticalPrevents task completionUsers can't find the Pay button below the fold
MajorCauses significant delay or frustrationUsers repeatedly enter the date in the wrong format
MinorSmall slowdown or annoyanceUsers hesitate over an unclear icon
CosmeticNoticed but doesn't affect the taskInconsistent icon sizes

Consider frequency too: a major problem that affected 4 of 5 participants matters more than one that affected 1 of 5.

Reporting findings

  1. 1Debrief with observers right after each session while memories are fresh.
  2. 2Collect all observations and group them by task or theme.
  3. 3For each problem, write what happened, how many participants it affected, why it happened, and its severity.
  4. 4Add a recommendation for each problem.
  5. 5Include short video clips — seeing a real person struggle convinces stakeholders faster than any slide.
  6. 6Prioritise, agree on fixes with the team, and plan the next test round.

Example · A finding, written well

Problem (Major, 4/5 participants): Participants didn't notice the delivery fee until the final payment screen, and two said they felt "tricked".

Why: The fee appears only in the order summary, in small grey text.

Recommendation: Show the delivery fee on the restaurant card and in the cart, before checkout.

Key takeaways

  • Usability testing means watching real people attempt realistic tasks — not asking for opinions.
  • About 5 participants per round finds most common problems; iterate with several small rounds.
  • Write tasks as realistic goals without using the interface's words.
  • Moderate neutrally: use think-aloud, don't help, ask 'What would you do if I weren't here?'
  • Measure success, time, errors and satisfaction; rate severity and report with clear recommendations.

Exercise

Run a mini usability test

  1. Pick a website or app you didn't design, e.g. a public transport ticket site.
  2. Write three realistic tasks with goals but no instructions.
  3. Test with three people using think-aloud; don't help them.
  4. Record task success and note every problem with a quote.
  5. Write the top three findings with severity, cause and recommendation.

Test yourself

Usability testing quiz

5 questions. Score 70% or more to complete the lesson.

  1. 1.What do you primarily observe in a usability test?
  2. 2.Nielsen's guideline says about how many users per round find most common problems?
  3. 3.What's wrong with the task 'Click Plans and choose Pro'?
  4. 4.A participant asks 'Where do I click?' What's the best reply?
  5. 5.A problem prevents users from completing a task. Its severity is…