Surveys tell you what people think they'd do. Usability testing shows you what they actually do — and that gap is where most product failures hide.
1Moderated vs. Unmoderated Testing
In moderated testing, a facilitator sits with the participant (in person or on a call), asking follow-up questions and probing hesitation in real time. Unmoderated testing runs asynchronously through a tool that records the screen and voice while the user works alone — faster and cheaper, but you lose the ability to ask 'why did you just do that?' on the spot.
2Why Neutral Facilitation Matters
The way you phrase a task shapes the result. 'Try to find the export button' tells the user the button exists and where to look. A neutral prompt like 'You need to get this report out of the app — what would you do?' preserves the confusion (or clarity) your real users would actually experience.
3Step-by-Step Breakdown
Introduction. Usability testing means putting your product in front of real users and watching them attempt real tasks — not asking whether they 'like' it, but observing exactly where they hesitate, misclick, or give up.
The Think-Aloud Protocol. In a moderated session, you ask participants to narrate their thoughts as they work: 'What are you looking for right now? What do you expect this button to do?' That running commentary exposes the mental model users bring — and where your interface breaks it.
Sample Size and the SUS Score. Nielsen's research found that 5 users typically surface about 85% of usability problems, because issues overlap heavily across participants — running more rounds with small groups beats one giant study. The System Usability Scale (SUS) then gives you a comparable 0-100 score across rounds.
Knowledge Check. Why does Nielsen's research recommend testing with around 5 users per round instead of 20?
- →Usability problems overlap heavily between users, so returns diminish fast after the first few participants
- →Because recruiting more than 5 participants is against research ethics guidelines
Summary. Usability testing turns assumptions into evidence. Run small, frequent rounds with representative users, keep questions neutral, and fix what you observe — not what people say they'd prefer.
