Usability Testing Script

Five Users Will Find 85% of Your Usability Problems. But Only If You Ask the Right Questions.

Jakob Nielsen published this finding in 2000, and it's held up remarkably well. Five participants in a usability test will surface roughly 85% of the issues in an interface. The catch — and it's a big one — is that the quality of what you find depends entirely on the quality of your test script. A bad script with 20 participants will find less than a good script with 5.

I've sat through usability sessions where the moderator asked "Is this easy to use?" and every participant said yes while visibly struggling with the interface. Leading questions don't just waste your time. They actively mislead you.

Where Usability Scripts Go Wrong

The most common mistake is writing tasks that mirror your navigation structure instead of your user's goals. "Click on Settings, then Notifications, then Email Preferences" tests whether someone can follow instructions. It tells you nothing about whether they can figure out how to stop getting emails they don't want.

A 2024 UserTesting benchmark report found that task-based usability tests (where participants pursue a goal) uncover 3.4x more issues than feature-walkthrough tests (where participants follow instructions). Yet most teams still write scripts that read like QA checklists.

The second failure point is the warm-up. Teams skip it or phone it in. But the first five minutes of a usability session set the tone. If participants feel like they're being tested — like they can give wrong answers — they'll perform instead of behaving naturally. Steve Krug's approach in "Don't Make Me Think" is to spend the first few minutes making the participant feel like *the product* is being tested, not them. That small reframe changes everything.

How This Prompt Helps

This prompt generates a complete usability script — warm-up questions, task scenarios, follow-up probes, and a scoring rubric — tailored to your specific product and research goals. The tasks are written as goal-oriented scenarios ("You want to invite a teammate to your project") rather than step-by-step instructions. The follow-up probes use the "think-aloud" protocol to capture what participants were actually experiencing.

When to Reach for This

• You're testing a prototype or beta feature before a wider launch and need a professional-grade test script in minutes
• Your team has never run a usability test and needs a structured starting point
• You're running remote unmoderated tests (UserTesting, Maze) and need precise task wording that works without a moderator
• You've identified a specific flow with high drop-off rates and want to understand why users bail
• You're preparing for a design review and want to show up with user evidence, not opinions

What Good Looks Like

A good usability script feels like a conversation, not an exam. Tasks should be framed as realistic scenarios with enough context that the participant can behave naturally. The scoring rubric should capture both quantitative data (completion rate, time on task, error count) and qualitative observations (confusion points, workarounds, emotional reactions). Aim for 4-6 core tasks — more than that and participants fatigue, giving you worse data.

Sources

• Why You Only Need to Test with 5 Users — Nielsen Norman Group
• Usability Testing Benchmarks 2024 — UserTesting
• Don't Make Me Think — Steve Krug