The case for lean usability testing

Usability testing is often associated with large budgets and formal research labs. In practice, some of the most valuable insights come from small, focused sessions with a handful of users. This guide explains how to run them effectively — and how regular testing reduces the risk of building the wrong thing.

Why small tests work

Usability researcher Jakob Nielsen famously argued that testing with around five users uncovers most of a product's usability problems. After that, each new person mostly repeats what you've already seen.

For early-stage teams, this is significant. Testing does not need to be a large study at the end of a project. It can be quick, affordable and frequent — a regular part of the development process.

Recruit the right participants

Five participants are only effective if they represent your intended users. Recruit people who share the context, needs and habits of your audience, rather than whoever is most convenient.

Colleagues and friends are convenient but often too familiar with the product and too inclined to be supportive. Participants who match your audience will provide more objective feedback.

Recruitment can be straightforward: a short message in relevant communities, a small incentive for participants' time and flexible scheduling will usually secure five sessions within a week.

Use realistic tasks and observe

Give each participant a small number of realistic tasks, such as “book a ride for tomorrow morning”, rather than open-ended instructions. Ask them to think aloud as they work.

Avoid explaining, assisting or defending the design during the session. Any moment where you feel the need to intervene is likely to be a finding worth recording.

What to watch for

Participants' behaviour is more reliable than their comments. Pay particular attention to:

• Hesitation — pauses that indicate something is not obvious.
• Navigation errors — where users go when they expect to be somewhere else.
• Workarounds — the creative detours people invent when the obvious path fails.
• Language gaps — words on screen that mean something different to them than to you.

A single participant struggling may be an exception. Several participants struggling at the same point indicates a clear issue.

“Observe what users do, not only what they say. Behaviour is the most reliable evidence you have.”

Turn observations into decisions

The value of a test depends on what happens afterwards. Immediately after the sessions, we review findings as a team and group them into three categories:

• Fix now — problems that stop people completing the core task.
• Fix soon — friction that slows people down or causes doubt.
• Watch — one-off issues we’ll look for again in the next round.

We then prioritise the changes with the greatest impact and schedule them for the next build. The output is a short list of clear decisions rather than a lengthy report.

Fix, then test again

The value comes from repetition. Address the most significant issues, then test again with new participants. Three rounds of five users will generally provide more insight than a single round of fifteen, with each round starting from an improved product.

By launch, the most obvious problems have been resolved, and the team has established a habit of designing with real users rather than assumptions.

Further reading: Why You Only Need to Test with 5 Users — Jakob Nielsen, Nielsen Norman Group