User research is how product teams find out what people actually need, do and struggle with, instead of guessing. The hard part isn't learning the methods. It's choosing the right one for the question you have. This guide maps common methods to the questions they answer, shows how many participants each typically needs, works through a small study plan, and covers consent under India's data protection rules.
Two questions that pick the method
Nielsen Norman Group's guide to choosing user-research methods, by Christian Rohrer, maps 20 methods across three dimensions. Two of them do most of the work:
- Attitudinal or behavioural? Do you need to know what people say (opinions, preferences, reasons) or what they do (actual actions, errors, time taken)?
- Qualitative or quantitative? Do you need to understand why and how, from a small number of people, or how many and how much, from a large number?
The third dimension is the context of use, for example whether people use the product naturally, in a scripted test, or not at all.
| Your question | Leans | Methods that fit |
|---|---|---|
| What problems do people have, and how do they cope today? | Attitudinal, qualitative | In-depth interviews, contextual inquiry (observing people where they work or live) |
| Can people complete key tasks with our design? | Behavioural, qualitative | Moderated usability testing |
| How common is this problem or preference? | Attitudinal, quantitative | Surveys |
| Which of two designs performs better? | Behavioural, quantitative | A/B testing, analytics |
| How do people group or label our content? | Attitudinal and behavioural | Card sorting, tree testing |
| How does use change over weeks? | Behavioural, qualitative | Diary studies |
Many projects need two methods: a qualitative one to understand the problem, and a quantitative one to measure how widespread it is.
Keep preparing
Continue with Sarkari Resume Templates₹299 — coaching के एक महीने से काफ़ी सस्ता / far cheaper than a month of coachingHow many participants?
- Qualitative usability testing. In "Why You Only Need to Test with 5 Users" (March 2000), Jakob Nielsen modelled the share of problems found with n users as 1 − (1 − L)ⁿ. He used a typical L of 31%, the share of problems a single user reveals. With five users, that gives 1 − 0.69⁵ ≈ 0.84, close to the 85% he cites. His point was to run several small rounds and fix problems between them, not one large test. The formula holds only for comparable users. Nielsen notes that when you have highly distinct user groups, such as children and parents, you need to test people from each group.
- Interviews. Plan for enough sessions to stop hearing new themes. Recruit a few more than you think you need, because some participants will cancel.
- Surveys and A/B tests. These need much larger samples. Decide the smallest difference worth detecting before you start, and check the required sample size. See our A/B testing guide for a worked example.
Worked example: a two-week study plan
The product and details below are illustrative.
A team building a bill-payment feature in a banking app is seeing many users abandon the flow partway through.
| Day | Activity | Output |
|---|---|---|
| 1 | Write the research question: "Where and why do users abandon the bill-payment flow?" Pull analytics for step-by-step drop-off. | Drop-off chart showing the biggest exit point |
| 2–3 | Recruit 5 recent abandoners and 5 regular users, across the languages the app supports | Confirmed schedule and signed consent |
| 4–6 | Run 10 moderated sessions (45 minutes each): a short interview, then tasks on the live flow | Session notes and recordings, stored securely |
| 7–8 | Synthesis: write one observation per sticky note, group them, and count how many participants hit each issue | 5–8 themes, each with evidence |
| 9 | Prioritise the issues by frequency and severity | A ranked list with proposed fixes |
| 10 | Share a one-page readout. Agree which fixes to A/B test | Decisions and next steps |
Notice the order: analytics show where people leave, and sessions show why. Measuring the fixes with an A/B test then shows whether they helped.
An interview guide you can adapt
- Warm-up (5 minutes): "Tell me about the last time you paid a bill on your phone."
- Recent behaviour (15 minutes): "Walk me through what you did, step by step. What happened next? What did you do when that didn't work?"
- Task observation (20 minutes): "Please try paying this sample bill. Think aloud as you go."
- Wrap-up (5 minutes): "Is there anything I should have asked but didn't?"
Ask about specific past behaviour, not hypothetical future use. "Would you use this?" invites polite yeses. "When did you last…?" gets you real stories.
Consent and personal data in India
User research usually collects personal data: names, phone numbers, recordings and behaviour. According to a PIB release of 14 November 2025, the government notified the Digital Personal Data Protection Rules, 2025, operationalising the DPDP Act, 2023. The release says the Rules set an 18-month phased compliance timeline. It also says organisations must give standalone, clear consent notices that explain the specific purpose for which personal data is collected.
For a study, that means you should:
- Tell participants the purpose, what you'll record, who will see it and how long you'll keep it
- Get consent before recording, in a language the participant understands
- Collect only the data the study needs, and store recordings in a controlled location
- Remove names from notes and quotes before sharing
- Follow your organisation's privacy team's guidance on the current DPDP obligations
Common mistakes
- Asking people to predict their behaviour. Observe or ask about the past instead.
- Leading questions. "Don't you find this confusing?" tells participants what to say. Ask "What did you expect to happen?"
- Recruiting only enthusiasts. Include people who stopped using the product, or never started.
- Reporting everything. Prioritise by frequency and severity, and link each finding to a decision.
Building a user research portfolio
For UX research, product or design roles, show one complete study rather than a list of methods. Include the question, why you chose the method, who took part, what you found (with anonymised evidence), and what changed as a result. A small, honest study with a clear decision at the end is more convincing than a large one with no outcome. For framing research inside product decisions, see our guide to product strategy frameworks.