Definition
User research is the structured study of users — their goals, behaviours, contexts, and obstacles — through interviews, surveys, observed sessions, and data analysis, conducted to inform product decisions with evidence rather than assumption.
User research replaces the team's mental model of the user with the actual one. It exists because the team is not the user — engineers know the system too well, designers care about subtleties no user notices, and executives have years of context the prospect arriving today does not.
Research splits into two broad modes. Generative research happens before a product exists or before a redesign begins, and answers "what should we build?" Evaluative research happens once a design exists and answers "does this work?" Both modes use overlapping methods (interviews, observation, surveys, analytics) but the questions and outputs differ. Generative outputs include personas, journey maps, and problem statements. Evaluative outputs include usability findings, A/B test results, and bug reports.
Origin
Rooted in human factors research from World War II (cockpit ergonomics, equipment design). Brought into software through the work of Jakob Nielsen, Don Norman, and the contextual-inquiry methods of Hugh Beyer and Karen Holtzblatt at Apple in the 1990s.
How it works
- Define the research question — what decision will this research inform?
- Choose the method that fits the question — interviews for motivation, observation for behaviour, surveys for scale.
- Recruit a sample that matches the target audience, not just "available users".
- Run the sessions; record (with permission); take structured notes.
- Synthesise findings into themes — what patterns repeat across users?
- Translate themes into design or product decisions; share findings with the broader team.
When to use it
Use when
- Before designing or redesigning a major flow.
- When the team is divided on what to build — research breaks the tie with evidence.
- When a launched product is underperforming and the team can't agree on why.
- On a recurring cadence (quarterly) for products with a fast-changing user base.
Skip when
- When the decision is already made and research is being used to justify it.
- When you can't act on what you'd learn — research without a budget for change is theatre.
Key metrics
- Number of distinct insight themes per study (signal of saturation).
- Decisions influenced — research is only valuable if it changes what gets built.
- Time from research to design decision (lag = waste).
- Recurrence: same insight appearing across studies signals a structural issue.
Examples
- User research showed that B2B buyers wanted pricing on the site, not gated behind a form.
- We spent three weeks on user research before writing a single line of code.
- Five interviews exposed a category of user we hadn't designed for at all.
In practice at Makreate
Makreate UX projects start with research so the product solves a real, validated problem. A recent edtech client wanted to add an AI tutor feature; two weeks of interviews with 12 students revealed that the actual blocker to outcomes was teacher feedback delay, not lack of AI help. We redirected the build. Six months later, the teacher-feedback redesign drove the highest retention lift in the product's history — the AI tutor would have been months of wasted engineering.
UX Design →Common mistakes
- Asking users what they want. Users describe problems well; designing solutions is the team's job.
- Recruiting from the team's network. Friends-and-family samples are biased toward existing supporters.
- Drawing conclusions from N=2. Patterns need at least 5 sessions per audience segment.
- Treating research as a one-off project. Insights decay; the user base evolves.
- Hoarding findings on the research team. If only researchers know, nothing changes.
Frequently asked
Interviews or surveys?
Interviews for depth, motivation, and the unexpected. Surveys for scale and validating an existing hypothesis. Use interviews to find the questions; surveys to size the answers.
How many participants do I need?
5 per audience segment for qualitative interviews (saturation kicks in around there). 100+ for surveys with statistical claims. 20+ for moderated usability with quantitative measures.
Should I incentivise participants?
Yes — fair compensation broadens the sample beyond people who'll show up free (who skew toward enthusiasts). $50–$150 for a 30–60 minute session is standard.