There's a well-known finding in UX research: testing with just 5 users typically surfaces the large majority of usability problems in an interface. Here's how to run that test well.
Write Tasks, Not Questions
Ask users to complete a real task ("find and book an appointment") rather than asking their opinion ("do you like this page?") — behavior reveals more than stated preference.
Recruit Representative Users, Not Convenient Ones
Testing with coworkers or friends who already understand the product misses the confusion a genuine first-time user would hit — recruit people who match your actual audience.
Say as Little as Possible During the Test
The instinct to help when someone struggles has to be resisted — that struggle is exactly the data you're there to collect, not a moment to intervene.
Watch for Where They Hesitate, Not Just Where They Fail
A user who eventually succeeds but hesitates or backtracks first is still showing you a friction point worth fixing, even though it wouldn't show up as a failed task.
Turn Findings Into Prioritized Fixes
Not every observed issue needs fixing immediately — group findings by how many users hit them and how severely, then tackle the highest-impact ones first.
