User Research Methods
UI/UX Design Principles
Chapter 2 · User Research Methods
Chapter 1 covered the Design Thinking process's first stage, Empathize, only by name. This chapter makes it concrete: the actual methods teams use to learn what real users need, rather than guessing — interviews, surveys, and the artifact both feed into, the persona.
The Real Gap Between What Users Say and What They Do
The correction matters because the quote is usually used to argue the opposite of good UX practice: "don't bother asking users, they don't know what they want." The real, evidence-based lesson researchers actually rely on is narrower and more useful: what people say they want in an interview and what they actually do when using a product often diverge — not because research is worthless, but because self-reported preference is only one kind of data. Good research combines what users say (interviews, surveys) with what they actually do (usability testing, analytics, contextual observation).
User Interviews
A one-on-one interview is the real workhorse of qualitative research — open-ended questions, asked to a real user, about their own actual experiences and behavior rather than hypothetical future features.
- Ask about the past, not the future — "Tell me about the last time you did X" gets real behavior; "Would you use a feature that did Y?" gets a guess.
- Avoid leading questions — "Don't you find this confusing?" primes the answer you're looking for.
- Follow up on specifics — a vague answer ("it was fine") is worth a genuine "can you walk me through exactly what happened?"
How Many Users Do You Actually Need? Nielsen's Real Formula
Jakob Nielsen's own widely-cited claim — that testing with just 5 users finds around 85% of usability problems — comes from a real formula, not a rule of thumb pulled from nowhere:
Plugging in L ≈ 0.31 and n = 5 gives roughly 85% of problems found — the source of the famous number.
- This applies to qualitative usability testing — a quantitative study (measuring metrics like task completion time) needs roughly 20 users instead, for statistical reliability.
- If your product has several genuinely distinct user groups, the 5 users need to be split per group (e.g. 3-4 per group for two groups) — 5 users total across two very different audiences won't cover either one well.
- Nielsen recommends three separate 5-user studies, testing → fixing → retesting each time, over one single 15-user study — a redesign introduces new problems a single big study never gets the chance to catch.
Surveys
A survey trades depth for breadth — closed-ended, structured questions sent to many users at once, well-suited to quantitative questions ("how satisfied are you, on a 1-5 scale") but poorly suited to discovering something you didn't already think to ask about. Interviews find the unknown unknowns; surveys measure how common a known issue actually is.
Personas: A Real, Corrected History
Personas — fictional, composite users built from real research, used to keep a design team focused on actual human needs rather than an abstract "the user" — are most associated with designer Alan Cooper, who began using prototype personas as early as 1983, built from informal interviews with 7-8 real users. The technique reached a wide audience through his 1999 book, The Inmates Are Running the Asylum.
- Goal: Complete grocery ordering in under 5 minutes, from a phone, often one-handed
- Frustration: Re-entering a delivery address every single order
- Quote (from a real interview): "I don't need it to be pretty, I need it to be fast"
Three Research Methods, Compared
| Method | Best for | Real limitation |
|---|---|---|
| Interviews | Discovering the unknown — real motivations, context, unexpected problems | Small sample, time-intensive, relies on self-report |
| Surveys | Measuring how common a known issue is, across many users | Can't discover something you didn't think to ask about |
| Usability Testing | Watching actual behavior with a real interface (say vs. do) | Needs a real or prototype interface to test against |
Hands-On Exercises
A product manager cites the "faster horses" Henry Ford quote to justify skipping user research entirely for a new feature. Explain, in your own words, what's wrong with both the quote itself and the conclusion being drawn from it.
📄 View solutionA team runs one single usability study with 15 participants and, separately, has three genuinely distinct types of users for their product. Explain, in your own words, two real ways this study design deviates from what Nielsen's own research actually recommends.
📄 View solutionA team wants to quickly build a persona in a single meeting, using only what everyone already assumes about their typical user. Explain, in your own words, why this wouldn't count as a genuine persona per this chapter, and what real step would fix it.
📄 View solutionChapter 2 Quick Reference
- The "faster horses" Ford quote is real, widely considered apocryphal — the genuine lesson is that stated preference (say) and real behavior (do) are different, complementary kinds of data, not that research is worthless
- Interviews ask about the past, avoid leading questions, follow up on vague answers
- Nielsen's real formula:
N × (1 − L)ⁿ, giving ~85% of problems from 5 users — but only for qualitative studies, split per distinct user group, ideally as 3 rounds of 5 rather than one round of 15 - Surveys measure how common a known issue is; interviews discover unknown issues
- Personas: popularized by Alan Cooper (1999 book, prototype use since 1983), with a genuine independent 1993-94 precursor from Angus Jenkinson — only real if built from actual research, never from assumption alone