Design Reference
UX Research Methods
UX research methods split by the question they answer: discovery methods (interviews, contextual inquiry) tell you what problem exists, evaluative methods (usability testing, heuristic review) tell you whether your solution works, and quantitative methods (analytics, A/B tests, surveys) tell you how much. Choosing the wrong category is why research often fails to inform a decision.
Browse by Category
All UX Research Methods (24)
Discovery (6)
What problem exists, and for whom.
| Method | Type | What it tells you | Effort · scale |
|---|---|---|---|
User interviews | Qualitative | One-to-one conversations about what someone actually did, not what they would do. The foundation of most discovery. | Medium · 5–8 people |
Contextual inquiry | Qualitative | Watch someone work in their own environment. Surfaces the workarounds nobody thinks to mention in an interview. | High · 4–6 sessions |
Diary study | Longitudinal | Participants log experiences over days or weeks. The only way to see behaviour that is spread over time. | High · 1–4 weeks |
Jobs to be Done interview | Qualitative | Reconstructs the moment someone switched to your product and why. Good at exposing the real competitor. | Medium · 6–10 people |
Competitive analysis | Desk research | Learn the conventions of your category before deciding which to break. Cheap, and usually skipped. | Low · 1–2 days |
Stakeholder interviews | Qualitative | Surfaces constraints, politics and success criteria before you design against the wrong goal. | Low · 3–6 people |
Type
Qualitative
What it tells you
One-to-one conversations about what someone actually did, not what they would do. The foundation of most discovery.
Effort · scale
Medium · 5–8 people
Type
Qualitative
What it tells you
Watch someone work in their own environment. Surfaces the workarounds nobody thinks to mention in an interview.
Effort · scale
High · 4–6 sessions
Type
Longitudinal
What it tells you
Participants log experiences over days or weeks. The only way to see behaviour that is spread over time.
Effort · scale
High · 1–4 weeks
Type
Qualitative
What it tells you
Reconstructs the moment someone switched to your product and why. Good at exposing the real competitor.
Effort · scale
Medium · 6–10 people
Type
Desk research
What it tells you
Learn the conventions of your category before deciding which to break. Cheap, and usually skipped.
Effort · scale
Low · 1–2 days
Type
Qualitative
What it tells you
Surfaces constraints, politics and success criteria before you design against the wrong goal.
Effort · scale
Low · 3–6 people
Structure & Synthesis (5)
How information and journeys should be organised.
| Method | Type | What it tells you | Effort · scale |
|---|---|---|---|
Card sorting | Generative | Participants group content their own way, showing you the mental model your navigation should match. | Low · 15–20 people |
Tree testing | Evaluative | Tests whether people can find things in your structure, with no visual design to rescue it. | Low · 20–30 people |
Journey mapping | Synthesis | Lays out the whole experience over time, including the parts you do not own. Finds the gaps between touchpoints. | Medium · workshop |
Personas | Synthesis | Condenses research into a few archetypes so a team argues about the same user. Worthless if invented rather than derived. | Medium · after research |
Service blueprint | Synthesis | A journey map plus everything backstage — staff, systems, policies — that has to work for it. | High · workshop |
Type
Generative
What it tells you
Participants group content their own way, showing you the mental model your navigation should match.
Effort · scale
Low · 15–20 people
Type
Evaluative
What it tells you
Tests whether people can find things in your structure, with no visual design to rescue it.
Effort · scale
Low · 20–30 people
Type
Synthesis
What it tells you
Lays out the whole experience over time, including the parts you do not own. Finds the gaps between touchpoints.
Effort · scale
Medium · workshop
Type
Synthesis
What it tells you
Condenses research into a few archetypes so a team argues about the same user. Worthless if invented rather than derived.
Effort · scale
Medium · after research
Type
Synthesis
What it tells you
A journey map plus everything backstage — staff, systems, policies — that has to work for it.
Effort · scale
High · workshop
Evaluative (7)
Whether what you designed actually works.
| Method | Type | What it tells you | Effort · scale |
|---|---|---|---|
Moderated usability test | Qualitative | Watch someone attempt real tasks while you observe. Five participants surface most serious problems. | Medium · 5 people |
Unmoderated usability test | Qualitative | Same tasks, recorded without you present. Cheaper and faster; you lose the ability to ask why. | Low · 10–20 people |
Guerrilla testing | Qualitative | Grab passers-by for five minutes with a prototype. Crude sampling, and far better than no testing. | Very low · an afternoon |
Heuristic evaluation | Expert review | An expert audits against usability principles. Finds obvious problems fast, and cannot replace real users. | Low · 2–3 reviewers |
Accessibility audit | Expert review | Checks against WCAG with tooling and manual keyboard and screen-reader passes. Often a legal requirement. | Medium · per release |
First-click test | Quantitative | Where do people click first? Strongly predicts whether they complete the task at all. | Low · 30+ people |
Five-second test | Quantitative | Show a screen briefly, then ask what it was for. Tests whether the message lands before anyone reads. | Very low · 20+ people |
Type
Qualitative
What it tells you
Watch someone attempt real tasks while you observe. Five participants surface most serious problems.
Effort · scale
Medium · 5 people
Type
Qualitative
What it tells you
Same tasks, recorded without you present. Cheaper and faster; you lose the ability to ask why.
Effort · scale
Low · 10–20 people
Type
Qualitative
What it tells you
Grab passers-by for five minutes with a prototype. Crude sampling, and far better than no testing.
Effort · scale
Very low · an afternoon
Type
Expert review
What it tells you
An expert audits against usability principles. Finds obvious problems fast, and cannot replace real users.
Effort · scale
Low · 2–3 reviewers
Type
Expert review
What it tells you
Checks against WCAG with tooling and manual keyboard and screen-reader passes. Often a legal requirement.
Effort · scale
Medium · per release
Type
Quantitative
What it tells you
Where do people click first? Strongly predicts whether they complete the task at all.
Effort · scale
Low · 30+ people
Type
Quantitative
What it tells you
Show a screen briefly, then ask what it was for. Tests whether the message lands before anyone reads.
Effort · scale
Very low · 20+ people
Quantitative (6)
How many, how often, and whether a change caused it.
| Method | Type | What it tells you | Effort · scale |
|---|---|---|---|
Surveys | Quantitative | Sizes a pattern you already found qualitatively. Poor at discovery — you can only ask what you thought of. | Low · 100+ responses |
Analytics review | Behavioural | What people actually did, at scale. Tells you what happened and never why. | Low · existing data |
Funnel analysis | Behavioural | Where people drop out of a multi-step flow. Points at the screen to investigate qualitatively. | Low · existing data |
A/B testing | Experimental | Causal evidence that one version performs better. Needs real traffic and the discipline not to peek. | Medium · weeks |
Session replay | Behavioural | Watch anonymised recordings of real sessions. Powerful, and a privacy obligation to handle carefully. | Low · continuous |
Satisfaction metrics | Attitudinal | NPS, CSAT and SUS track sentiment over time. Useful as a trend, misleading as a single number. | Low · continuous |
Type
Quantitative
What it tells you
Sizes a pattern you already found qualitatively. Poor at discovery — you can only ask what you thought of.
Effort · scale
Low · 100+ responses
Type
Behavioural
What it tells you
What people actually did, at scale. Tells you what happened and never why.
Effort · scale
Low · existing data
Type
Behavioural
What it tells you
Where people drop out of a multi-step flow. Points at the screen to investigate qualitatively.
Effort · scale
Low · existing data
Type
Experimental
What it tells you
Causal evidence that one version performs better. Needs real traffic and the discipline not to peek.
Effort · scale
Medium · weeks
Type
Behavioural
What it tells you
Watch anonymised recordings of real sessions. Powerful, and a privacy obligation to handle carefully.
Effort · scale
Low · continuous
Type
Attitudinal
What it tells you
NPS, CSAT and SUS track sentiment over time. Useful as a trend, misleading as a single number.
Effort · scale
Low · continuous
Frequently Asked Questions
How many people do I need for a usability test?
Five per round for qualitative testing surfaces most serious problems, and running three rounds of five beats one round of fifteen because you fix things in between. Quantitative methods are different — first-click tests and surveys need dozens to hundreds for the numbers to mean anything.
Qualitative or quantitative research?
They answer different questions. Qualitative tells you why and what to change; quantitative tells you how many and whether it worked. Use qualitative to find the problem and quantitative to size it — a survey cannot discover something you did not already think to ask.
Can I skip research if I have analytics?
Analytics tell you exactly what happened and never why. A drop-off at step three is a question, not an answer — you still need to watch someone attempt step three to learn whether it is confusing, slow or simply the wrong step.
What is the cheapest useful method?
A five-second test or guerrilla testing — both cost an afternoon and routinely surface problems the team had stopped being able to see. Perfect sampling matters far less than testing at all.