User research is how you find out, before you build and while you build, what people actually need and where your design confuses them. It is cheaper to learn that from a handful of people early than from one-star reviews after launch. Match the method to the question: interviews and field visits to understand needs, prototypes and usability tests to find problems, and surveys and analytics to measure at scale. Run small rounds repeatedly rather than one large study.
Key takeaways
- Nielsen Norman Group's advice is to test with no more than five users per round and to run as many small tests as you can afford; its model assumes a typical single user reveals about 31 percent of usability problems and five users find about 85 percent.
- The same article says that if you have distinct user groups you need three to four participants per group (two groups) or about three per group (three or more), and that the five-user figure assumes you fix problems and retest.
- Choose methods along three dimensions: attitudinal (what people say) versus behavioural (what they do), qualitative (why) versus quantitative (how many), and the context of use.
- Use generative methods early (to decide what to build), formative methods during design and summative methods after launch.
- Research is not a phase; it continues after release through analytics and feedback.
Why does user research matter for mobile apps?
Mobile apps are used in short, interrupted moments, on small screens, with unreliable networks, by people who will delete an app that gets in their way. Teams who build from assumptions find the problems late, when changes are expensive and store reviews are public. Research replaces opinions with observations: what do people try to do, where do they hesitate, what do they say, what do they skip? It also keeps the first release small, because you build only what you have evidence people need.
How do you choose a method?
Nielsen Norman Group maps UX methods along three dimensions:
| Dimension | Ends | Use it for |
|---|---|---|
| Attitudinal vs behavioural | What people say vs what people do | Surveys and focus groups lean attitudinal; A/B tests and eye tracking lean behavioural; usability and field studies sit between, usually leaning behavioural |
| Qualitative vs quantitative | Why and how to fix vs how many and how much | Qualitative observation answers "why"; analytics and surveys answer "how many" |
| Context of use | Natural use, scripted use, limited prototype use, or no use of a product | Pick the context that matches the question |
Which methods suit which stage?
| Stage | Question | Methods |
|---|---|---|
| Strategise (generative) | What do people need? | Field studies, diary studies, interviews, surveys, concept testing |
| Design (formative) | Does this design work? | Card sorting, tree testing, usability testing of prototypes |
| Launch and assess (summative) | How is it performing? | Benchmarking, unmoderated testing, A/B testing, analytics |
What are five practical methods for a mobile app?
- Interviews. Talk one-to-one with people in your target group about how they do the task today. Listen for problems, workarounds and words they use.
- Field observation. Watch people in the setting where they will use the app, such as in a shop, on the move or at work. Context shows constraints you would not guess.
- Prototype usability tests. Give representative users realistic tasks on a clickable prototype and watch where they succeed or struggle, without helping. Nielsen Norman Group's guidance is about five users per round, with fixes and retests between rounds.
- Surveys. Use closed questions to measure attitudes across more people, after you know what to ask.
- Analytics and feedback after launch. Track where people leave, which features are used and what reviews and support messages repeat. Add event tracking at the start so you can see the key steps.
Test on real phones, small and large, on poor connections, and with accessibility settings such as larger text. Our mobile app development challenges article lists device and store checks to include.
How many users do you need?
For finding usability problems, a few. Nielsen Norman Group's article, written in 2000 and still widely cited, argues for about five users per round: one user reveals roughly a third of the problems and returns diminish quickly, so several small rounds beat one big one. Distinct user groups need their own participants. For measuring things quantitatively, such as comparing task times, you need many more; the article suggests about 20 users, and about 15 for card sorting. Treat these as rules of thumb and note that the article links to a newer analysis.
How do you test AI features?
AI features need extra checks. Test with realistic inputs, including unclear and unusual ones, and look for wrong answers, confusing explanations and over-trust. Show people what the AI did and let them correct it. Note what happens when the feature is unavailable on a device or offline. See Mobile app technologies in 2026 for on-device and cloud choices.
How do you fit research into a project?
- Before design: five to eight interviews or visits to clarify the problem and the user.
- During design: a usability round on each prototype iteration, fix, retest.
- Before launch: a test on the real build on real devices, including store review checks.
- After launch: monitor analytics and feedback weekly, and plan the next round from what you see.
Keep notes short and share clips or quotes with the team so research changes decisions. Our app development team builds this into projects, and UX and UI design trends in 2026 covers current interface expectations.
FAQ
How many people should I test with?
About five per round for finding usability problems, repeated across rounds, according to Nielsen Norman Group. If you have distinct user groups, test each group.
Is user research only for large projects?
No. A few interviews and a prototype test are enough to catch the biggest problems in a small project.
When should research start?
Before design, so you build the right thing, and continue after launch.
What is the difference between usability testing and a survey?
A usability test shows what people do and where they struggle; a survey records what many people say. They answer different questions.
Can analytics replace research?
No. Analytics shows what people do and where they leave; research explains why.
Sources (3 links)
- Nielsen Norman Group, "Why You Only Need to Test with 5 Users": https://www.nngroup.com/articles/why-you-only-need-to-test-with-5-users/
- Nielsen Norman Group, "When to Use Which User-Experience Research Methods": https://www.nngroup.com/articles/which-ux-research-methods/
- Nielsen Norman Group, "Usability 101: Introduction to Usability": https://www.nngroup.com/articles/usability-101-introduction-to-usability/
