Where the number comes from
Jakob Nielsen's analysis showed that the first few participants uncover most usability problems, and each new one repeats more of what you have already seen. Five users find about 85 percent of the problems in his curve.
His recommendation was not to stop at five. It was to run small tests repeatedly. NN/g suggests that a budget for fifteen participants is better spent on three rounds of five, fixing issues between rounds, than on one round of fifteen.
What five users can tell you
- Where people get lost, confused, or stuck.
- Which words and labels they misread.
- Which flows take far longer than expected.
- Which problems are severe enough to fix first.
What they cannot tell you
| Question | Why five is not enough |
|---|---|
| How many users hit this problem? | Small samples cannot estimate frequency |
| Is design A faster than design B? | Needs quantitative testing and larger samples |
| Does it work for every segment? | Test three to four users per distinct group |
| What is our success rate? | Success rates need larger studies |
Run a good session
Write tasks
Realistic goals, not instructions. "Find and book a slot for next week."
Brief and stay quiet
Ask the person to think aloud. Do not help.
Observe
Note where they hesitate, backtrack, or misread.
Debrief
Ask what was hardest and what they expected.
Then fix and retest
Rank issues by severity and how many participants hit them. Fix the top ones, and test the new version with a few fresh people. Short cycles beat big reports that arrive too late to matter.


