In short
- Usability testing shows what people do, which is more reliable than what they say they would do.
- A handful of participants per round finds the most common problems. Test, fix, and test again.
- Good tasks describe a goal in the user's words and never name the button to press.
- Findings only matter if they change the design before the next round.
Also called: User testing, Usability study, Usability test.
What usability testing is
In a usability test, a participant who matches your intended user gets a realistic task, such as booking an appointment or exporting a report, and tries to complete it with your design. You watch. You note where they hesitate, misread a label, take a wrong turn, or give up.
It works on almost anything that can be used. A paper sketch, a clickable prototype, a coded build, or a product that has been live for years can all be tested the same way. Earlier tests are cheaper to act on, because less has been built. Choie's prototypes are coded rather than drawn, so the people trying them react to real behavior instead of tapping through pictures.
Testing is different from asking for opinions. Surveys and interviews tell you what people think and remember. A test shows you what they do, and the two often disagree. That makes it a core habit of UX design, and a useful check during product discovery.
Moderated and unmoderated testing
There are two broad ways to run a session, and many teams use both at different stages.
Moderated vs unmoderated
Moderated
- A facilitator runs each session live
- Can ask follow-up questions
- Best for early or complex designs
- Slower to schedule and run
Unmoderated
- Participants work alone on a recording
- Fixed script, no follow-up
- Best for short, focused tasks
- Faster and easier to repeat
Both can run in person or remotely over a video call with screen sharing. Remote sessions make it far easier to reach people outside your city, and participants use their own devices, which is closer to real life.
How many participants a round needs
Fewer than most people expect. A handful of participants per round is usually enough, because the biggest problems trip up nearly everyone. By the time the fourth or fifth person hits the same wall, new sessions mostly repeat what you already know.
Several small rounds teach more than one large round. Test with a few people, fix what you found, and test again. Each round also catches the problems that the previous fixes uncovered.
When to recruit more
If you have distinct groups of users, such as buyers and admins, run a small round for each group. They use the product differently and stumble in different places.
How to write good tasks
Tasks are where most first tests go wrong. A good task gives a goal and a reason, in language the participant would use, and leaves the path to them.
| Weak task | Better task | Why it works |
|---|---|---|
| Click Settings and change your plan | Your team has grown and you need more seats. Make that change. | Gives the goal and hides the path |
| Try the search feature | Find a vegetarian dinner you could cook tonight in under 30 minutes | Clear enough to know when it is done |
| Explore the dashboard | Find out which client owes you the most money | Has an end point and a right answer |
| Sign up for an account | A friend told you about this app. Get started with it. | Mirrors how people really arrive |
- Avoid words that appear on the screen, or you give away the answer.
- Ask participants to think aloud, and stay quiet when they struggle. Helping too early hides the problem.
- Run one pilot session with a colleague to catch confusing tasks before real participants see them.
What to do with the findings
Write findings up the same day, while they are fresh. For each problem, note what happened, how many participants hit it, and how badly it blocked them. Then sort the list by severity.
- Blockers. People could not finish the task. Fix these first.
- Slowdowns. People finished but hesitated, backtracked, or needed a hint.
- Annoyances. Small friction or confusing wording that did not stop anyone.
Change the design and test again. Resist the urge to argue with what you saw. If three people missed the same button, the button is the problem, and a short ranked list that leads to changes is worth more than a long report nobody reads.
Testing during a build with Choie
Choie's builds come with a live link to use every Friday, so there is something real to put in front of a few users each week instead of waiting for launch. A one-week coded prototype for a fixed $4,900 gives you one working flow to test before a full build. How a project runs has the weekly rhythm.
Common questions
How many people do I need for a usability test?
A handful per round is usually enough to find the most common problems. Run several small rounds with fixes in between instead of one large one.
What is the difference between usability testing and user research?
Usability testing is one kind of user research. It checks whether people can use a specific design, while methods such as interviews explore which problem to solve in the first place.
Should I pay participants?
Yes, in most cases. A fair payment or gift card gets you more reliable participants and respects their time, though existing customers are often happy to help for free.
When should usability testing happen?
As soon as you have something people can try, and again before each major release. Finding a problem in a rough prototype costs far less than finding it after launch, when it may already be part of an MVP people rely on.
Does Choie run usability tests for clients?
Recruiting participants and formal usability testing are not part of the prototype week. The coded prototype is yours after delivery, so you can put it in front of your own users with tasks like the ones above. In the last week of a build, Choie watches real people use the launched product and fixes what gets in their way.