Run a small usability test without inventing statistics

A small usability test can reveal confusing interactions and missing information. It does not reliably estimate how often every customer will encounter a problem. Use realistic tasks, observe what people do, and keep findings tied to the participants and conditions you actually tested.

Write a task without giving the answer

“Find a time next Tuesday and confirm which timezone it uses” is more useful than “Click the timezone label beside the date.” The second instruction tells the participant where to look and hides the problem you want to observe.

Keep a simple observation log

For a fictional five-person test, “Two participants missed the timezone label” is a bounded observation. It should not become “40% of customers book the wrong time.” The sample and setting do not support that broader estimate.

Ask neutral follow-ups

Ask what the participant expected to happen or what they were looking for. Avoid “Wasn’t that button confusing?” because it suggests the answer. Record the wording when it affects your interpretation.

Organize these usability notes.
Separate observed behavior, participant statements, interpretations, and proposed fixes.
Preserve the task, sample size, device, and assistance given.
Do not estimate population rates or conversion changes from this small test.
Notes: [redacted observations]

Retest the important change

Fix the most consequential confusion and repeat the task with an appropriate participant or fresh starting state. A participant who already learned the interface may not reveal the same first-use problem again.

Keep the test lightweight but honest. A useful result can be a specific wording change or a new question to investigate. Do not turn a handful of observations into an impressive-looking research report with invented confidence intervals or claims about revenue.