Usability testing is the part of UX education that textbooks describe in clean numbered steps and that reality delivers as a long afternoon of watching someone click the wrong button twelve times while you cannot say a word. Bohdan scheduled five participants for a student project testing a library catalog interface. By participant three, a clear pattern had emerged that no analytics tool would have caught.
The Observation Phase: What Gets Recorded
Each session runs approximately 45 minutes. Bohdan uses a think-aloud protocol, meaning participants narrate their actions as they complete assigned tasks. The data captured includes: task completion rate, time-on-task per scenario, error frequency, and qualitative observations coded into categories.
- Task 1 completion rate: 80% across all participants
- Task 2 completion rate: 40% — participants could not locate the advanced filter
- Average time on Task 2 before abandonment: 3 minutes 12 seconds
- Recurring verbalization: users expected filter options above the results list, not below
That placement assumption was not a preference. It was consistent across four of five participants with no prompting. A single layout decision invalidated a core feature for most users.
What Forty-Three Observations Actually Mean
By the end of the day, Bohdan has a spreadsheet with 43 discrete observations tagged by severity: critical, moderate, minor. According to Jakob Nielsen, testing with 5 participants uncovers approximately 85% of usability problems in a given interface. The remaining 15% surfaces with additional testing rounds, not guesswork.
The afternoon ends with affinity mapping: grouping observations by theme, not by the order they appeared. Three critical issues cluster around navigation labeling. Two around feedback absence. The interface is not broken. It is just confidently wrong about where users look first.