Testing with five people finds most usability problems. More people mostly repeat them.
5 users uncover ~85% of usability issues. More users find the same problems repeatedly.
Five users surface about 85% of usability issues. Run several small rounds rather than one big study.
When you need 'why', talk to 5 people. When you need 'how many', use quantitative.
Watch five people try to book a room on a new flow and you will see the same handful of walls: they miss the date picker, they can't tell which room is selected, they hesitate at the fee line. A sixth and seventh tester mostly re-confirm what the first five already showed. That is Nielsen's point: usability problems cluster, so a small round finds the big ones quickly and cheaply. Run several small rounds as you iterate, not one giant study at the end.
When the question is “how many” rather than “why” (conversion rates, an A/B winner, sizing a market) five users tell you nothing reliable, and you need proper sample sizes. And if a product serves very distinct groups, say buyers and sellers, five per group beats five in total, because their problems barely overlap.
A team wants to know *why* users abandon onboarding step 3.