UX Research Methods: Which One To Use And When

Choosing between UX research methods is less about finding the single best technique and more about matching a method to the question you need answered. A user interview will not tell you how many visitors abandon checkout, and an analytics dashboard will not tell you why they abandon it. Teams that pick a method first and a question second usually end up with data that looks impressive but does not change a single design decision.

This guide explains the main methods used by product, design and marketing teams, how to choose between them, and what each one is actually good for. It covers interviews, surveys, usability testing, card sorting, tree testing, first click tests, five second tests, preference tests, A/B tests, diary studies, analytics and contextual inquiry, along with a practical framework for deciding which to run and when.

Whether you run a SaaS product, an ecommerce store or a marketing site, the same logic applies: define the decision you need to make, then pick the cheapest method that gives you enough confidence to make it.

How To Choose UX Research Methods: A Simple Framework

The most widely cited way to organize user research methods comes from Nielsen Norman Group. In its article When to Use Which User-Experience Research Methods, Christian Rohrer maps 20 methods across three dimensions. You do not need to memorize the chart, but the three questions behind it will help you pick the right method almost every time.

Attitudinal vs Behavioral

Attitudinal research captures what people say: their opinions, beliefs, expectations and stated preferences. Interviews and surveys are the classic examples. Behavioral research captures what people do: where they click, where they get stuck and whether they complete a task. Usability testing, analytics and A/B tests fall on this side.

The gap between the two is where many product mistakes happen. Users often say they want more features and then ignore them, or say a flow is easy after struggling with it for two minutes. When the decision depends on real behavior, favor a behavioral method.

Qualitative vs Quantitative UX Research

Qualitative methods gather data by observing or talking to people directly. They are best at answering why something happens and how to fix it. Quantitative methods gather data indirectly, through an instrument such as a survey or analytics tool, and answer how many and how much. NN/g frames the distinction the same way: qualitative for insight into causes, quantitative for measurement.

Neither is better. Qualitative work usually comes first, to discover problems and form hypotheses, and quantitative work follows to measure how widespread a problem is or whether a fix moved a metric. Many teams combine both in a mixed methods approach.

Stage Of Product Development

The third filter is timing. NN/g groups the product cycle into three broad phases, each with a different research goal:

• Strategize (generative research): find new opportunities and understand the problem space. Typical methods are interviews, field studies, contextual inquiry and diary studies.
• Design (formative research): improve a design while it is still being built. Typical methods are usability testing, card sorting, tree testing and first click tests.
• Launch and assess (summative research): measure how a live product performs against a goal, a previous version or a competitor. Typical methods are A/B testing, analytics, benchmarking and surveys.

If you only remember one rule, make it this: generative methods help you decide what to build, formative methods help you build it well, and summative methods tell you whether it worked.

The Main UX Research Methods Explained

User Interviews

One-on-one conversations, usually 30 to 60 minutes, focused on a participant’s goals, current workflow, frustrations and decision process. Interviews are attitudinal and qualitative, which makes them ideal early in a project when you are still learning who your users are and what problem you are solving. Ask about past behavior (“Tell me about the last time you…”) rather than hypothetical future behavior, since people are poor predictors of what they will do.

Surveys

Surveys collect self-reported answers from a larger group. They are useful for sizing an issue you already understand, segmenting users, or tracking satisfaction over time. They are weak at discovering problems you did not know to ask about. Keep them short, avoid leading questions, and pair closed questions with one or two open text fields.

Usability Testing: Moderated and Unmoderated

Usability testing means watching real people attempt realistic tasks with your product or prototype. It is the workhorse of formative research. The main usability testing methods differ in how the session is run:

• Moderated testing: a facilitator guides the session in person or over video, asks follow-up questions and probes when the participant hesitates. It produces rich insight but takes more time per session.
• Unmoderated testing: participants complete tasks on their own through a testing platform that records their screen and, often, their voice. It is faster and cheaper to run at scale, but you cannot ask follow-up questions in the moment.

Usability testing can be qualitative (observing a handful of people to find problems) or quantitative (measuring success rates and task times across a larger sample). Prototypes from design tools can be tested before any code exists; our comparison of Figma and Sketch for UI design covers the prototyping side of that workflow.

Card Sorting

Card sorting asks participants to group topics or content items into categories that make sense to them. In an open card sort, participants create and name their own groups. In a closed card sort, they sort items into categories you define. It is the go-to method for designing or reorganizing navigation, menus and help centers, because it reveals how users mentally group information.

Tree Testing

Tree testing is the reverse of card sorting. Participants see a text-only version of your site hierarchy, with no visual design, and are asked where they would go to find a specific item. The results show whether people can find things in your structure and where they take wrong turns. A common pattern is to run a card sort to generate a structure, then a tree test to validate it.

First Click Testing

First click testing shows participants a screen or mockup, gives them a task, and records where they click first. The premise is simple: if the first click is on the right path, people are much more likely to complete the task. It is a fast way to check whether labels, buttons and layout point people in the right direction.

Five Second Tests

A five second test shows a page for five seconds, hides it, and then asks what participants remember or what they think the page is about. It is designed to evaluate first impressions and clarity of messaging, which makes it popular for landing pages, hero sections and ad creative. It will not tell you whether a page is usable, only whether its main message lands.

Preference Tests vs A/B Tests

These two are often confused. A preference test shows participants two or more design options and asks which they prefer and why. It is attitudinal: it measures stated preference, often with a small panel, and is useful for early direction on visual style or copy.

An A/B test splits live traffic between two versions and measures which one performs better on a real metric, such as signups or purchases. It is behavioral and quantitative, and it needs enough traffic to reach a reliable result. Use preference tests to narrow options before launch and A/B tests to measure impact after launch.

Diary Studies

In a diary study, participants log their experiences, thoughts and actions over days or weeks, usually through a mobile app or short daily prompts. Diary studies capture behavior that unfolds over time, such as onboarding, habit formation or how a tool fits into a weekly routine, which a single session cannot show.

Analytics and Session Recordings

Product and web analytics show what users do at scale: traffic sources, funnel drop-offs, feature adoption and retention. Session recordings and heatmaps add a visual layer, letting you replay real sessions to see where people scroll, hesitate or rage click. Analytics tell you where a problem is; you usually need a qualitative method to learn why it happens.

Contextual Inquiry

Contextual inquiry combines observation and interviewing in the user’s own environment, such as their office, warehouse or home. The researcher watches people do their real work and asks questions as it happens. It is especially valuable for B2B and workflow software, where the environment, interruptions and workarounds shape how a product is used.

Which Method Answers Which Question

Use this table as a quick reference when planning a study. Start from the question in the second column, not the method name.

How Many Participants Do You Need?

Sample size depends on whether you are looking for problems or measuring something. Here is what Nielsen Norman Group recommends, with the sources linked so you can read the reasoning:

• Qualitative usability testing: about 5 users per round. NN/g’s well-known guidance on testing with 5 users reports that five participants uncover about 85% of usability problems, and recommends running several small rounds with redesigns in between rather than one large study.
• Card sorting: the same NN/g article recommends at least 15 users.
• Quantitative usability studies: NN/g’s summary of quantitative sample sizes recommends 40 participants in most cases, based on a 15% margin of error at 95% confidence.

If you are testing two or more distinct user groups, such as buyers and administrators, recruit a few participants from each group rather than five total. For surveys, A/B tests and analytics, the right number depends on your traffic and the size of the effect you want to detect, so use a sample size calculator rather than a rule of thumb.

Recruiting Participants

The best method in the world produces bad data if you test the wrong people. A few practical rules:

• Recruit for behavior, not demographics. “Bought furniture online in the last six months” is a better screener criterion than “age 25 to 40.”
• Use a short screener survey. Ask neutral questions that do not reveal the right answer, so people cannot guess their way into the study.
• Tap your own users first. Customer email lists, in-app intercepts and support contacts are often the cheapest and most relevant source.
• Use a participant panel when you need strangers. Most research platforms offer access to recruited panels with demographic and professional filters.
• Offer a fair incentive. Gift cards or account credits improve show-up rates, and specialist B2B audiences usually expect more.
• Over-recruit slightly. Some participants will cancel or not show, so book a buffer for moderated sessions.

Research Tools Worth Knowing

You do not need an enterprise stack to start. Many teams run interviews over standard video calls, surveys with general form tools, and analytics with the platform already installed on their site. Dedicated UX research platforms become worthwhile once you run studies regularly, because they bundle recruiting, unmoderated testing, card sorting, tree testing, first click and five second tests, and session recordings in one place. Our UXtweak review is a useful example of what an all-in-one research platform includes and what to look for when comparing options.

When evaluating any tool, check which methods it supports natively, whether it includes a participant panel, how it handles data privacy and consent, and whether its pricing scales with the number of studies or the number of responses.

Common Mistakes To Avoid

• Asking users to design the solution. Users are experts on their problems, not on your interface. Ask about needs and behavior, then let the team design.
• Relying only on what people say. Stated preference and actual behavior often differ. Back up attitudinal findings with a behavioral method before making big bets.
• Running one big study instead of several small ones. Iterative rounds of testing catch more problems because each round validates the previous fix.
• Using a qualitative sample to make quantitative claims. Five participants are enough to find problems, not to say that 60% of users will fail a task.
• Leading questions and tasks. “How easy was that?” assumes it was easy. Use neutral wording and realistic tasks.
• Testing too late. Testing a prototype costs far less than rebuilding shipped code. Research early and often.
• Letting findings sit in a slide deck. Share clips, prioritize issues by severity and assign owners, or the research will not change anything.

Frequently Asked Questions

What is UX research?

UX research is the practice of studying users to understand their needs, behaviors and pain points so a product can be designed around them. It includes both qualitative methods, such as interviews and usability testing, and quantitative methods, such as surveys and analytics. The goal is to reduce guesswork in product and design decisions.

What are the UX research methods and when should you use them?

Use generative methods like interviews, diary studies and contextual inquiry early, when you are deciding what to build. Use formative methods like usability testing, card sorting and tree testing while designing. Use summative methods like A/B tests, analytics and surveys after launch to measure results.

Is usability testing qualitative or quantitative?

It can be either. Qualitative usability testing observes a small group, often around five people per round, to find and understand problems. Quantitative usability testing measures metrics such as success rate and time on task across a larger sample, which NN/g puts at about 40 participants in most cases.

How do you recruit participants for UX research?

Start by defining who you need based on behavior, then write a short screener to filter applicants. Recruit from your own customer base, in-app prompts or a research panel, and offer an incentive that matches the audience. Book a few extra participants to cover no-shows.

When should you do usability testing?

As early as you have something to test, even a paper sketch or clickable prototype, and again at each major design iteration. Testing before development is far cheaper than fixing problems after launch. Many teams also run periodic tests on the live product to catch regressions.

How do you do remote usability testing?

Choose moderated sessions over video if you need to probe and ask follow-up questions, or an unmoderated testing platform if you need speed and scale. Write clear, realistic tasks, run a pilot session to catch confusing instructions, and record sessions so the whole team can review them.

Conclusion

The right UX research methods depend on the question, not on habit. Decide whether you need to understand attitudes or behavior, whether you need insight or measurement, and where you are in the product cycle. That framework points you to interviews and contextual inquiry early, usability testing, card sorting and tree testing during design, and A/B tests and analytics after launch.

Start small: five users in a usability test, a quick tree test on your navigation, or a handful of customer interviews will surface more useful insight than an ambitious study that never gets finished. Run research in short, repeated rounds, share what you learn, and let the findings drive the next design decision.