Five users, tested on real tasks, will expose the majority of the usability problems in a flow. This has been the working rule in UX research for decades. And yet the median number of usability tests run by the teams we meet is not five. It is zero.
The excuses, ranked
The polite excuse is logistics: no lab, no recruiting pipeline, no time. The honest excuse is fear — watching a real person fail at your interface is painful, and it's easier to ship and hide behind dashboards. The most dangerous excuse is 'we'll see it in the data'. Analytics tell you where users drop off. They never tell you why, and teams fill that gap with whatever story flatters the roadmap.
What five users actually requires
A prototype or a live build, five people roughly like your users, three or four concrete tasks, and the discipline to watch silently instead of demoing. No lab, no eye tracker, no six-week study. An afternoon. The bottleneck was never budget; it was the willingness to be wrong in front of a stranger.
Testing is cadence, not ceremony
We treat small-batch testing as cadence, not ceremony: five users per round, one round per meaningful design change, findings written up the same day while the wincing is still fresh. Fast, cheap, and brutal beats slow, expensive, and comfortable.
Testing with five users isn't a compromise on rigor. Shipping with zero is.
