Textbook of Usability

The five-user rule (sometimes "Nielsen's rule of five") comes from Jakob Nielsen and Thomas Landauer's 1993 mathematical model of usability problem discovery. Their formula predicts the proportion of problems found by $n$ participants:

$$P(\text{found}) = 1 - (1 - L)^n$$

where $L$ is the probability that a single participant will encounter a given problem (~31% on average). With $n=5$, the formula predicts ~85% of problems will be discovered.

The rule has been widely cited to justify small-sample formative testing. Nielsen's advice is not "test with 5 and stop" but "test with 5, fix the biggest problems, test again with 5." Iterative testing with small samples is more cost-effective than a single large test because each round fixes the most important problems before they waste later observations.

Caveats are important:

• The 85% figure applies to problems with discovery probability ~31%. Rarer problems (e.g., affecting only 10% of users) require many more participants to detect.
• The rule applies to formative testing, not summative measurement. Quantitative performance metrics require larger samples for statistical power.
• Different user populations (novices vs experts, different contexts) each need their own round of testing. "Five users" means five per population, not five total.
• The rule describes major problem discovery, not comprehensive coverage. Critical safety issues may warrant larger samples.

The five-user rule remains the most practical guidance for formative usability testing and underlies the predominant small-sample, iterative tradition in UX practice.