What you need to know about usability testing

Usability testing is a research method where real users attempt real tasks with a product while a researcher observes where they succeed, where they hesitate, and where they give up. The goal isn’t opinions—it’s behavior: what people actually do when no one is coaching them.

How usability testing works

Running a usability test follows roughly the same five steps regardless of the product, platform, or budget behind it.

1. Recruit participants. Find people who match your actual target users, not whoever’s available in the office. Five participants who represent your real audience surface more useful friction than fifteen who don’t.
2. Define tasks. Write specific, realistic tasks that mirror what someone would actually try to do with the product—“find and cancel an upcoming reservation,” not a vague prompt like “explore the app.”
3. Run sessions. Watch participants attempt each task, thinking aloud, while a moderator stays hands-off. The goal is to capture what happens without help, not to guide anyone to success.
4. Analyze patterns. Look across every session for the same hesitation, wrong turn, or drop-off point repeating, rather than treating any single participant’s experience as the verdict.
5. Report findings. Turn what you observed into specific, actionable recommendations tied to evidence—this screen, this task, this number of participants who stumbled—rather than general impressions.

Types of usability testing

Usability tests vary along two independent choices: whether a moderator is present, and whether the session happens in person or remotely. Most teams mix and match based on budget, timeline, and how naturally they need participants to behave.

Moderated

A researcher runs the session live, guiding participants through tasks, asking follow-up questions in the moment, and probing an unexpected reaction as it happens. Moderated tests surface the richest qualitative detail, but they take longer to schedule and run one participant at a time.

Unmoderated

Participants complete tasks on their own, usually through a testing platform that records their screen and voice with no researcher present. Unmoderated tests scale faster and cost less per participant, at the cost of losing the ability to ask a follow-up question in the moment.

Remote

Participants join from their own environment, over video call or an unmoderated platform. Remote testing widens the pool of people you can recruit and lets someone act more naturally in their own setting, at the cost of the body language and context a researcher would catch in the room.

In-person

Participants and researchers share a physical space, typically a lab or office. In-person testing makes it easier to observe body language and run structured tasks with controlled equipment, but it limits recruiting to whoever can travel to you.

Where usability testing fits in your research stack

Running the test is only half the job. Once you’ve tested more than a handful of participants, the harder problem becomes finding which friction points repeat across sessions instead of treating each recording as a one-off. See the best usability testing tools for platforms built to run the tests themselves, and AI tools for UX research for how teams are speeding up the analysis that comes after.

The terms are often used interchangeably, but usability testing specifically evaluates how easily someone can complete a task with a product. User testing is the broader term—it can also cover reactions to a concept, message, or price that have nothing to do with whether someone can complete a task.

Five participants is the common starting point. Usability research has repeatedly found that a small group surfaces most of a product’s major friction points, and returns diminish quickly after that. Testing in rounds of five and iterating between rounds tends to catch more problems than running one large round of fifteen or twenty participants.

A team preparing to launch a checkout redesign asks five participants to buy a specific item using the new flow, with no guidance beyond the task itself. The researcher watches for where each participant hesitates, misclicks, or asks a question out loud, then compares notes across all five sessions to find the moments where more than one participant stumbled at the same step.