User Research
User research is the practice of generating evidence about user needs, behaviors, and pain points through systematic methods, then synthesizing that evidence into actionable insights. Methods include interviews, observation, surveys, usability testing, diary studies, and analytics, and the output informs product, design, and marketing decisions. It is the input layer underneath product discovery, UX work, value proposition development, and most credible go-to-market positioning.
The discipline splits along two axes that matter. The qualitative versus quantitative axis: qualitative research (interviews, observation, usability tests) tells you why users do what they do and is best run in small N with rich detail; quantitative research (surveys, analytics, A/B tests) tells you what users do and how often, at scale, but doesn't explain motivation. The generative versus evaluative axis: generative research (interviews, contextual inquiry) discovers problems and opportunities you didn't know existed; evaluative research (usability tests, concept tests, A/B tests) tests solutions you've already designed. The four most-leveraged methods for startup teams: customer interviews (30 to 60 minutes, story-based questions about past behavior, ~5 to 8 interviews per segment is enough to surface the major patterns per Jakob Nielsen's research), usability testing (watch 5 users complete tasks on a prototype or live product; Nielsen's classic finding is that 5 users surface ~85 percent of usability issues), surveys (for quantifying preference or sizing problems across a larger sample, but biased toward people who answer surveys), and session recordings / heatmaps (FullStory, Hotjar, LogRocket, for seeing real user behavior at scale without scheduled sessions). The single biggest mistake in startup user research: asking customers what they want instead of asking them what they did. People are unreliable predictors of their own future behavior and reliable narrators of their own past behavior. Anchor questions on the past.
Ryan's Take
User research is the cheapest competitive advantage in startup history and the one most founders refuse to invest in. They'll spend $50,000 on Facebook ads targeting a customer they have not talked to in six months, and decline to spend $0 and an afternoon a week running five customer interviews. The math on that is insane. Five interviews a week for a quarter is sixty conversations with real customers, which is more direct evidence than 90 percent of your competitors will collect this year. The startups that beat you to product market fit are not the ones with better-funded marketing. They are the ones who talked to more customers and listened more carefully.
What founders get wrong: Asking customers what features they want. Customers are good at describing their problems and bad at designing solutions. "Would you use a feature that X" reliably produces inflated yes-answers that don't translate to behavior. Ask about the last time they had the problem, what they tried, what didn't work, and what they would have paid to make it go away. That gets you usable signal.
Related: User Experience · Product Discovery · Usability Testing · Continuous Discovery · Buyer Persona
FAQ
What is user research?
The practice of generating evidence about user needs, behaviors, motivations, and pain points through systematic methods (interviews, observation, surveys, usability testing, diary studies, analytics), then synthesizing that evidence into insights that inform product, design, and marketing decisions.
How is qualitative research different from quantitative?
Qualitative (interviews, observation, usability tests) tells you why users do what they do, best run in small N with rich detail. Quantitative (surveys, analytics, A/B tests) tells you what users do and how often, at scale, without explaining motivation. Most product decisions need both.
How many user interviews do you need?
For usability testing, 5 users surface ~85% of issues (Jakob Nielsen's classic finding). For exploratory interviews, ~5-8 per segment surfaces the major patterns; beyond that you hit diminishing returns. Quantitative sizing needs much larger N (typically 100+ for survey statistical confidence).
Find this article helpful?
This is just a small sample! Register to unlock our in-depth courses, hundreds of video courses, and a library of playbooks and articles to grow your startup fast. Let us Let us show you!
Submission confirms agreement to our Terms of Service and Privacy Policy.