Usability Testing: How to Run Your First Test

You are too close to your own product to see its problems. You know where everything is, what each label means and which button to press. Real users do not. Usability testing closes that gap by watching people try to use your product and noting where they struggle. It is one of the highest-value activities in design, and you can start with a handful of participants and almost no budget.

What usability testing is (and isn’t)

Usability testing observes real users attempting realistic tasks with your product or prototype. It answers: Can people use this? Where do they get stuck? Why?

It is different from:

• Interviews, which explore needs and motivations. See how to run user interviews.
• Surveys, which measure opinions at scale
• A/B tests, which compare variants using live traffic

Use usability testing to find and fix problems before and after launch.

How many users do you need?

Research from Nielsen Norman Group suggests that around five users per round typically reveal a large share of the major usability problems, and that it is better to test small, fix, and test again than to run one large study. If you have distinct user groups with different needs, test about five of each.

Step 1: Set clear goals

Decide what you want to learn. Examples:

• Can new users complete sign-up without help?
• Do people find the settings they need?
• Is the checkout flow understandable?

Pick a few critical journeys instead of testing the entire product.

Step 2: Choose what to test

You can test:

• Paper sketches or wireframes for early direction
• Clickable prototypes before development. See the UX design process.
• The live product, or a competitor’s, to find opportunities

Test as early as possible. Fixing a prototype takes minutes, while fixing shipped code takes days.

Step 3: Write realistic tasks

Tasks should describe a goal in the participant’s own terms without revealing how to do it.

Weak task	Better task
“Click the Settings icon and turn on notifications.”	“You want to stop getting emails from the app. Show me how you would do that.”
“Use the search to find a blue jacket.”	“You are looking for a waterproof jacket for hiking under a certain price. Find one you would consider.”

Provide a short scenario with context and a clear end point. Prepare three to six tasks for a session of about 30 to 45 minutes.

Step 4: Recruit representative participants

Aim for people who match your real users in relevant ways: experience level, device and goals. Avoid colleagues and friends. They know too much and are too polite. Sources include your customers, email lists, communities, panels or recruitment services. Offer a fair thank-you, be clear about time and obtain consent for recording.

Step 5: Prepare the session

• Write a script with a welcome, a reassurance that you are testing the product, not them, and instructions to think aloud
• Set up the prototype or product with test accounts and realistic data
• Choose a method: moderated (you guide in person or by video call) or unmoderated (participants complete tasks alone while recording)
• Test your setup, recording and links before the first participant arrives
• Assign a note-taker if possible

Step 6: Run the session

1. Welcome and set expectations. Explain that there are no wrong answers and that problems found are useful.
2. Ask a few warm-up questions about their background.
3. Give tasks one at a time. Ask them to think aloud as they go.
4. Stay quiet and observe. Resist the urge to help. If they struggle, that is data.
5. Use neutral prompts such as “What are you thinking?” or “What would you expect to happen?” Avoid leading questions.
6. Note behaviour, not just comments: hesitations, wrong paths, errors, expressions.
7. Debrief at the end: overall impressions, what was confusing, what they liked.
8. Thank them.

If a participant is stuck for too long, offer a small nudge so the session can continue, and record that they needed help.

Step 7: Capture and analyse

Right after each session, record the key observations. Then, across sessions:

• Group issues by theme (navigation, wording, layout, errors)
• Count how many participants hit each issue
• Rate each by severity:
  • Critical: prevents task completion
  • Serious: causes significant difficulty or errors
  • Minor: slows or annoys but does not block
• Keep quotes and short clips to illustrate problems for stakeholders

Useful measures include task success rate, time on task, error counts and a standard satisfaction questionnaire such as the System Usability Scale (SUS) if you want a number to track.

Step 8: Report and fix

Write a short, visual summary: top problems, evidence and recommended changes. Prioritise by severity and frequency, and by effort. Assign owners, make the fixes and test again to confirm they worked.

Remote vs in-person

Remote sessions are cheap and let you reach people anywhere. In-person sessions allow richer observation of body language and the environment. For mobile apps, observe how people hold and use their device, and test on real phones. Include assistive technology users to uncover accessibility barriers that automated tools miss.

Common mistakes

• Testing too late
• Leading participants or explaining the interface
• Testing with colleagues instead of real users
• Vague tasks that give away the answer
• Collecting data and never acting on it
• Ignoring severity when prioritising

Checklist

• Goals and key journeys defined
• Prototype or product ready with realistic data
• Tasks written as scenarios, not instructions
• Five representative participants per user group
• Script, consent and recording prepared
• Observed quietly, with think-aloud and neutral prompts
• Issues grouped, counted and rated by severity
• Fixes prioritised, made and retested

Final thoughts

You do not need a lab or a large budget. A handful of sessions with the right people will show you problems that no amount of internal review will find. Test early, test often and act on what you see. For help planning and running research and design improvements, see our UI/UX design service or contact us.