Usability Testing: Methods, Types, Tools and Examples

Usability testing is a research method where you watch real people try to complete realistic tasks with your product, so you can see where they get confused, slow down, or give up. You are not asking what they think of the design. You are watching what they do with it.

That difference is the whole point. Opinions are cheap and often wrong, while behavior is hard to argue with. A stakeholder can say a button label is obvious, and then five participants in a row click the wrong thing.

Most product teams already know they should test, and many still skip it because a release date is close. What slows them down is the fear that it needs a lab, a budget, and a researcher. It does not. A prototype, five participants, and a quiet hour each is enough to start. Below you will find the main types of tests, the methods and tools that make them practical, and a few examples from the kind of problems PMs run into every month.

‍

Types of Usability Testing and When Each One Fits

Ask ten UX people for the types of usability testing and you will get a few different lists. The most common framing follows Rubin and Chisnell: exploratory, assessment, validation, and comparison. The labels matter less than the question behind each one.

• Exploratory tests happen early, on sketches or rough prototypes. You want to know whether the concept makes sense at all.
• Assessment tests happen mid-build. You check whether people can complete the main tasks and where they stumble.
• Validation tests happen close to launch. You confirm that earlier fixes worked and no new problems appeared.
• Comparison tests put two designs side by side, such as two checkout flows, to see which one performs better.

There is also a split that cuts across all four. Moderated tests have a facilitator in the session who asks follow-up questions. Unmoderated tests let participants work alone, usually through a remote platform. Moderated sessions give you depth and the chance to ask "why did you click there?" Unmoderated ones give you speed and volume.

The Nielsen Norman Group describes the three core elements of any usability test: a facilitator, the tasks, and a participant. It is a useful checklist because most failed tests break one of the three. The facilitator leads the participant, the tasks are too vague to measure, or the participant is a colleague who already knows the product.

A mistake I see often is running a validation test when the team really needs an exploratory one. They show a polished prototype, participants politely say it looks great, and nobody learns anything. Polish invites politeness. Rougher material invites honest reactions, because nobody feels bad criticizing a sketch.

Match the type to the decision you need to make. If you are choosing between two directions, run a comparison test. If you are about to hand designs to engineering, an assessment test on a clickable prototype will save you the most rework. Our guide to building a prototype strategy covers how to choose the right fidelity for each stage.

‍

Usability Testing Methods and Tools Worth Using

The method is how you run the session. The common choices are in-person moderated testing, remote moderated testing over video, remote unmoderated testing, guerrilla testing, and first-click or tree testing for specific parts of the interface.

Guerrilla testing deserves more credit than it gets. You take a prototype to a coffee shop or a coworking space, ask strangers to try one task, and thank them with a coffee. It is rough, and you cannot control who you recruit, but it will surface obvious problems in an afternoon. For a B2B product, it works better to borrow customers who already have a call booked with your team and add a short test to it.

For remote work, the tool choice depends on the method:

1. Video call plus screen share (Zoom, Google Meet, Teams) is enough for moderated tests. Record with consent and take notes live.
2. Prototype tools such as Figma let you test before anything is built. Share a link and have the participant think aloud.
3. Unmoderated platforms such as Maze, UserTesting, or Lookback handle recruiting, task prompts, and recordings, and are best when you need numbers from a bigger sample.
4. Survey add-ons like the System Usability Scale give you a quick score you can track across releases.

Recruiting is the part that quietly decides whether the test is worth running. Participants should resemble the people who will actually use the product. If you are building an invoicing tool for accountants, five designers from the office will teach you very little. Write a short screener with two or three questions about role and experience, and turn away anyone who does not fit.

A practical rule: start moderated, because you will learn what questions to ask. Move to unmoderated once you know which tasks matter and want to measure them at scale.

Sample size is where teams overthink. Nielsen's well-known analysis found that five users uncover about 85% of usability problems in a single design, and that it is better to run several small rounds than one large one. That holds for finding problems. If you want to measure how often something happens, such as task success rates with a margin of error, you need a larger sample, and unmoderated tools are the sensible route.

One habit saves a lot of pain: write the tasks as goals, not instructions. "Find out how much it would cost to add a teammate" works. "Click Pricing, then Team plans" tests nothing, because you have already told them the answer.

‍

Usability Testing Examples From Real Product Work

Abstract advice only goes so far, so here are three situations where a test changes the outcome.

The first is the signup form. A team believes the form is short and clean. In a moderated session, three of five participants stop at the "Company size" dropdown because they are unsure which option applies to a freelancer. Nobody on the team had thought about freelancers. The fix is a single added option, and it took one afternoon of testing to find it.

The second is navigation labeling. A SaaS product calls its reporting area "Insights." Participants asked to find last month's usage spend a full minute scanning the menu and then give up on it, because to them "Insights" sounds like marketing content. A label test with ten participants through an unmoderated tool confirms "Reports" wins clearly.

The third is a comparison test for pricing pages. Version A lists features in a table, version B leads with three customer stories. Participants rarely say much about the table, but they linger on the stories and can describe the plans afterward. That is a finding you would never get from analytics alone.

Notice what the team did not need in any of these cases: a research department or a long report. A one-page summary with the problem, a clip of the moment it happened, and a proposed fix was enough to move each decision forward. Short video clips of participants struggling persuade stakeholders far better than a slide of quotes ever will.

What these examples share is that the problem was invisible to the team and obvious to the participants within minutes. The standard behind all of this is simple. ISO 9241-11 defines usability as the extent to which specified users can achieve specified goals with effectiveness, efficiency, and satisfaction in a specific context of use. Each example above fails on one of those three words.

If you are unsure where testing fits in the bigger picture, our overview of user experience basics and strategy shows how research feeds the decisions that come before and after it.

‍

FAQ