The biggest myth in product design is that research is expensive. It is not. Bad research is expensive — because it wastes time validating the wrong things. Good research is cheap, fast, and focused.
Start with what you already know
Before you talk to a single user, mine the data you already have. Support tickets, NPS responses, session recordings, analytics drop-off points — these contain more signal than most teams realise. An afternoon of systematic review often surfaces three to five clear research questions worth pursuing.
The goal at this stage is not to find answers. It is to identify the questions where your assumptions are weakest and the risk of being wrong is highest.
Five users is enough — if you recruit the right five
Jakob Nielsen's research showing that five users reveal 85 percent of usability problems still holds. But the five users have to be genuinely representative. One hour of careful recruiting is worth more than five hours of talking to the wrong people.
Define your target user criteria precisely before you recruit. Age, role, and technical familiarity are starting points. The more specific you are, the more useful the sessions will be.
Unmoderated testing scales what moderated cannot
Moderated research sessions are rich but slow. Unmoderated testing tools let you gather behavioural data from dozens of users in the time it takes to schedule one moderated session. Use them for navigation testing, first-click studies, and prototype validation.
The limitation of unmoderated testing is that you cannot ask follow-up questions. The advantage is volume and speed. Use both in combination — unmoderated to find what, moderated to find why.
Continuous discovery over project-based research
Teams that run research as a project — intensive bursts before a launch, then nothing — develop blind spots between cycles. Teams that build continuous discovery into their process — weekly user interviews, live data monitoring, ongoing feedback loops — make consistently better decisions because they are never far from their users.
Even one 30-minute user interview per week, every week, compounds into an enormous competitive advantage over a year.
The biggest myth in product design is that research is expensive. It isn’t. Bad research is expensive because it validates the wrong things and burns time. Good research is cheap, fast, and sharply focused on the riskiest assumptions.
Start with what you already know
Before you talk to a single user, mine the data you already have:
- Support tickets
- NPS responses
- Session recordings
- Analytics drop-off points
An afternoon of systematic review often surfaces 3–5 clear research questions worth pursuing.
At this stage, the goal is not to find answers. It’s to identify where your assumptions are weakest and the risk of being wrong is highest. Research that only confirms what you already believe is not research; it’s confirmation bias with a process attached.
Five users is enough — if they’re the right five
Jakob Nielsen’s finding that five users reveal ~85% of usability problems still holds, but only if those users are truly representative of your target audience.
One hour of careful recruiting is worth more than five hours of talking to the wrong people.
Define your target user criteria precisely before you recruit. At minimum:
- Role
- Technical familiarity
- Product experience
The more specific you are — for example, “mid-level marketing manager at a B2B SaaS company who currently tracks campaigns in spreadsheets” — the more directly applicable your findings will be.
Moderated vs. unmoderated testing
Moderated sessions are rich but slow. A skilled facilitator can:
- Follow interesting threads
- Ask follow-up questions
- Observe the gap between what users say and what they do
Unmoderated testing lets you gather behavioural data from many users in the time it takes to schedule a single moderated session.
Use unmoderated testing for:
- Navigation studies
- First-click tests
- Prototype validation
These work when the question is specific enough that you don’t need follow-ups.
Use moderated sessions when:
- You need to understand why a behaviour occurs
- You’re exploring a problem space rather than testing a specific solution
The strongest approach combines both:
- Unmoderated to find what happens
- Moderated to understand why it happens
The research debt problem
Teams that treat research as a project — intense bursts before launches, then long gaps — accumulate research debt. Between cycles:
- The product changes
- The user base evolves
- Findings from six months ago lose relevance
The answer is not “more big research projects.” It’s continuous discovery: small, regular touchpoints with real users built into the team’s normal rhythm.
Examples:
- A weekly user interview slot
The biggest myth in product design is that research is expensive. It isn’t. Bad research is expensive because it validates the wrong things and burns time. Good research is cheap, fast, and sharply focused on the riskiest assumptions.
Start with what you already know
Before you talk to a single user, mine the data you already have:
- Support tickets
- NPS responses
- Session recordings
- Analytics drop-off points
An afternoon of systematic review often surfaces 3–5 clear research questions worth pursuing.
At this stage, the goal is not to find answers. It’s to identify where your assumptions are weakest and the risk of being wrong is highest. Research that only confirms what you already believe is not research; it’s confirmation bias with a process attached.
Five users is enough — if they’re the right five
Jakob Nielsen’s finding that five users reveal ~85% of usability problems still holds, but only if those users are truly representative of your target audience.
One hour of careful recruiting is worth more than five hours of talking to the wrong people.
Define your target user criteria precisely before you recruit. At minimum:
