User Research Methods: How to Choose the Right One

User research methods are the structured ways a team learns who its users are, what they need and where a product gets in their way. The right method depends on the question you are asking and the stage your product is in. Interviews and field studies explain why people behave as they do, analytics and surveys measure how many do it, and usability tests show whether a design actually works. Below we group the main methods, match them to product stages, and walk through remote usability testing, including tasks, test types and metrics.

TL;DR

• Start with the question, not the method. Qualitative methods explain why and how; quantitative methods measure what, how many and how much.
• Combine what users say (attitudinal) with what they do (behavioral). Either one alone misleads.
• Match the method to the stage: generative research to find opportunities, formative testing to fix a design, summative measurement to track performance after launch.
• Remote moderated testing gives the richest insight; unmoderated is faster and cheaper but shallower.
• Write concrete tasks with no clues in the wording, and record success, time and errors for each one.

Table of Contents

• What a user research method actually is
• Three ways to sort the methods
• The main user research techniques compared
• Which method at which stage
• Usability testing: types and methods
• How remote usability testing works
• Usability testing examples: tasks worth copying
• Usability testing metrics
• Building a UX research plan
• Our perspective: research is cheapest before code
• Turning research into a product
• FAQ
• Sources

What a user research method actually is

A user research method is a repeatable way of collecting evidence about people's needs, behavior and attitudes toward a product. Yale's UX guidance puts the goal well: connect what users say, do and feel, because the mix of methods you pick should follow your project goals, timeline and research questions.

That last point is why "best method" lists often disappoint. A 30-minute interview and a 3,000-response survey are both legitimate, yet they answer different questions. Treat the methods as a toolbox, and pick based on what you need to decide next.

Research also pays off in places teams do not expect. How people move through a site affects engagement and, in turn, search visibility, a link explored in how user experience shapes organic growth.

Three ways to sort the methods

Nielsen Norman Group's widely used guide maps UX research along attitudinal versus behavioral, qualitative versus quantitative, and the context of product use, plus the stage of development. These dimensions answer most "which method?" questions.

Qualitative vs. quantitative

Qualitative methods explore why and how: interviews, observation, diary studies, moderated usability tests. Quantitative methods measure what, how many and how much: surveys with large samples, analytics, A/B tests, benchmark studies. UserTesting's guidebook describes the split as in-depth contextual understanding versus trends across larger samples, and recommends both for a complete picture.

A simple pairing: analytics shows that 60% of users abandon a checkout step. Interviews or a moderated test explain why.

Attitudinal vs. behavioral

Attitudinal methods capture what people say, believe or intend. Behavioral methods capture what they actually do. The gap between the two is wide: people forget, rationalize and politely agree. MeasuringU's taxonomy notes that observing behavior is what separates much UX research from market research, which leans on attitudes and self-reported behavior.

Generative, formative and summative

This dimension is about purpose. Generative research discovers needs and opportunities. Formative research improves a design while it is still changeable. Summative research measures how a finished product performs.

The main user research techniques compared

The table puts the common techniques side by side. It is a starting point, not a rulebook.

Interviews, surveys and analytics

Interviews are the fastest way to understand a domain you do not know. Surveys let you test what you heard on a larger group. NN/g describes surveys as collecting stated attitudes and behaviors, producing ratings, proportions and open-ended responses, while analytics is behavioral evidence from a live product: navigation, clicks, feature use, acquisition sources and abandonment.

The usual mistake is asking surveys to explain behavior. They are good at "how many" and weak at "why".

Contextual inquiry and diary studies

Both put research into the user's real life, but they differ. NN/g explains that contextual inquiry combines observation and interviewing while users perform tasks in their normal environment, whereas diary studies collect participant-recorded experiences remotely over days or weeks. Use contextual inquiry for a workflow with physical or organizational context, such as a dispatcher's desk. Use a diary study for something that happens sporadically, such as how a household manages a recurring task.

Card sorting

Card sorting reveals how users group and name things. NN/g notes it is used to determine labels and structures for information architecture by exposing users' mental models and terminology. Run it before you finalize a menu, not after.

Which method at which stage

NN/g's framework splits product development into three stages: Strategize, Design, and Launch & Assess. It assigns generative methods to finding new directions, formative methods to improving usability, and summative methods to measuring performance against the product's own past or against competitors.

UserTesting's lifecycle framework lines up with this: interviews, focus groups and competitive analysis before development, then card sorting, tree testing and prototype tests during it.

Do not wait for the perfect moment

NN/g's guidance is that research should happen at whatever stage the team is in, ideally at all stages, with most done early because early findings have the greatest impact. The same page advises treating the method list as a menu, not a rigid sequence: time, system maturity, product type and current concerns decide the mix.

GitLab's handbook makes the same point for teams that have already shipped: research can be inserted at any point, and after release you should pair qualitative feedback with quantitative dashboards measuring performance and engagement.

If you are earlier than any of this, a structured product discovery workshop is a compact way to turn assumptions into testable hypotheses before you invest in a build.

Usability testing: types and methods

Usability testing is the method most teams run most often: real users attempt realistic tasks while researchers observe. The differences between types come down to who guides the session and where it happens.

Moderated vs. unmoderated

In a moderated test, a facilitator watches, adjusts tasks and asks follow-ups live. In an unmoderated test, participants work alone from written tasks, with recorded actions and predetermined follow-up questions. NN/g summarizes the tradeoff: moderated studies offer richer probing, while unmoderated ones are generally quicker, cheaper and easier to scale.

Other variations

Digital.gov's usability guide lists co-discovery, comparative testing, remote moderated and remote unmoderated testing as common variations. Co-discovery has two participants work together and talk aloud to each other. Comparative testing puts two or more designs, or your design and an alternative, in front of users.

Usability testing vs. user testing

The terms are used interchangeably in everyday conversation, but there is a useful distinction. Usability testing evaluates how easily people complete tasks with a specific design. "User testing" is looser: it can mean usability testing, but also concept feedback or preference checks. When you plan a study, name the exact method and the decision it feeds.

How remote usability testing works

Remote moderated testing uses video conferencing or screen sharing so the facilitator can observe, adapt tasks and ask follow-ups in real time. NN/g recommends it when you need deep, rich insight, have geographically distributed or busy participants, or want to adapt tasks and clarify. Unmoderated remote testing is faster and cheaper, but the same guide notes it typically produces less detailed data because nobody can clarify instructions or probe a comment.

Before a session

1. Recruit users who match the audience you are designing for.
2. Set up meeting software so participants, observers and the facilitator can all connect.
3. Prepare the prototype or build and test the link yourself.
4. Run a pilot with a colleague to catch confusing tasks.

During a session

Digital.gov describes a typical session as an introduction, scenarios and a debrief, with participants generally asked to think aloud while finding and interpreting information. Let silence work. When a participant struggles, resist helping; the struggle is your data.

Setting up an unmoderated study

NN/g gives six steps for unmoderated testing: define study goals, select software, write tasks, pilot the test, recruit participants and analyze results. The pilot step is the one teams skip and regret, because with nobody present, an ambiguous task silently ruins every session.

Usability testing examples: tasks worth copying

Good tasks are concrete, realistic and free of clues. NN/g's checklist warns that careless wording can prime participants or reduce the test to matching words on the screen. It also separates two kinds of tasks: exploratory tasks are open-ended and show how people discover information, while specific tasks have a focused endpoint and can support qualitative or quantitative testing.

Examples for different products:

• Ecommerce (specific): "You need a gift for a friend under $50 that arrives by Friday. Find one and get to the point where you would pay." Avoid saying "use the filter".
• SaaS dashboard (specific): "A colleague asked how many new accounts were created last month. Find that number."
• Mobile app (exploratory): "Spend a few minutes looking around this app as if you just downloaded it. Tell me what you expect it to do."
• Onboarding (specific): "Set up your account so your team of three can start using it." Pair this with the checklist in our guide to user onboarding best practices.

What to observe

Watch where people hesitate, backtrack, misread labels or abandon. Note what they say against what they do. Tag each problem by severity: blocks the task, slows it, or merely annoys.

Usability testing metrics

When you need numbers alongside observations, a small set of metrics covers most studies:

• Task success rate: the share of participants who complete a task. Define "success" before the first session.
• Time on task: how long completion takes, useful for comparing designs or versions.
• Error rate: wrong clicks, wrong entries or dead ends per task.
• Satisfaction: a standardized post-test questionnaire such as the System Usability Scale, or a single-item ease rating after each task.

Small moderated tests are best read qualitatively. Report percentages only when your sample is large enough to support them, and say how many people took part.

Building a UX research plan

A workable plan fits on one page:

1. Decision: what will change depending on what you learn?
2. Research questions: two to five, phrased as questions, not feature requests.
3. Method and why: tie it to the stage and the qualitative or quantitative need.
4. Participants: who, how many, how you will recruit them, and screening criteria.
5. Tasks or discussion guide: drafted and piloted.
6. Timeline and roles: facilitator, note-taker, observers.
7. Output: a short readout with prioritized findings and recommended fixes.

The prioritization step is where research becomes design work. Our UI/UX audit checklist shows one way to rank fixes, and our overview of business analysis techniques covers methods that sit alongside research when requirements are unclear.

Our perspective: research is cheapest before code

We build web, mobile and AI products for clients across industries, and the pattern repeats: the expensive surprises are rarely technical. They are a workflow nobody observed, a label nobody tested, a feature nobody asked users about. That is why we treat discovery, business analysis, UI/UX design and QA as one connected effort rather than separate line items.

Our practical advice is modest. Run five or six honest sessions on a clickable prototype before engineering starts. Pair every analytics finding with at least one conversation. And when you add AI features, test them with users as early as you would test a form, because people react to model behavior in ways no spec predicts. We would rather change a prototype in an afternoon than a release in a quarter.

— Dmitry

Turning research into a product

Research only matters if findings reach the build. If you want a partner who can run discovery, shape the design and carry it through development and testing, see our UI/UX design services or start with the discovery phase, where assumptions are validated before you commit to a full engagement. Once a design is built, software testing and quality assurance confirms it behaves as the research said it should.

FAQ

What are the main types of user research methods?

The common groupings are qualitative versus quantitative, attitudinal versus behavioral, and generative, formative and summative. Specific methods include interviews, contextual inquiry, surveys, diary studies, analytics, card sorting and usability testing. Most projects combine two or three from different groups.

What is the difference between qualitative and quantitative UX research?

Qualitative research explores why and how people behave, usually with small samples and rich observation. Quantitative research measures what, how many and how much across larger samples. Use quantitative data to find where a problem is and qualitative data to understand its cause.

When should I choose remote unmoderated testing?

Choose it when you need quick, low-cost answers about simple, well-defined tasks, or feedback from many participants. Expect less detail, since no one can probe a comment or clarify a task. For complex flows or early prototypes, moderated sessions are usually the safer choice.

How many users do I need for a usability test?

There is no single number. Small qualitative rounds with a handful of participants are common for finding design problems, and you repeat them as the design changes. Quantitative metrics need larger samples, so match the group size to whether you want insight or measurement.

Sources

• When to Use Which User-Experience Research Methods, NN/g
• UX Research Cheat Sheet, NN/g
• Quantitative User-Research Methodologies: An Overview, NN/g
• When to Use Context Methods: Field and Diary Studies, NN/g
• Checklist for Planning Usability Studies, NN/g
• Remote Moderated Usability Tests: Why to Do Them, NN/g
• Unmoderated User Tests: How and Why to Do Them, NN/g
• Get to Know Your User, Yale Usability & Digital Accessibility
• A Taxonomy of Common UX Research Methods, MeasuringU
• The UX research methodology guidebook, UserTesting
• When to conduct UX Research, The GitLab Handbook
• Usability testing, Digital.gov
• User Experience in SEO: Driving Organic Growth Success

Recommended

• Business Analysis Techniques: 13 Tools and When to Use Them
• Requirements Gathering Techniques: 9 Methods and When to Use Each
• Get Users to Value in Under 5 Minutes: Onboarding Checklist, PMs & UX