Week 7 Guide: Testing Your Design
The Real Problem This Solves
You are too close to your own design to judge it. What is obvious to you may confuse a first-time user. Testing replaces your opinion with real evidence of how the design behaves in someone else's hands. The most valuable result is watching a user struggle, because that shows you exactly what to fix.
Key Ideas You Need First
1. Test the design, not the user
If a user struggles, the design failed, not the person. Never rescue them by explaining how it works. Their confusion is your most useful data.
2. Give tasks, not a tour
Do not demonstrate your prototype and ask "what do you think?" Instead, give the user a real task and watch them attempt it. "Show me how you would pack this away" beats "do you like it?"
3. Observe behaviour, then ask
Watch first, in silence. Note where they pause, fumble, or look unsure. Ask your questions only after they have tried.
4. Beware your own bias
You want your design to work, so you will be tempted to hint, defend, or hear praise louder than problems. Resist all three.
Consent And Ethics
Tools And Materials Needed
- Your Week 6 prototype.
- Two or three people who fit your user (not the same people every week is better).
- The Feedback Form Template to structure your notes.
- A phone for optional photos or video, with consent.
Step-By-Step Walkthrough
- Write a short test script. Two or three real tasks the user should attempt, in plain language, plus your consent lines.
- Set the scene. Explain you are testing the design, not them. Ask for consent.
- Give the first task and go quiet. Watch. Do not help, hint, or explain. Count silently to ten before you say anything.
- Note behaviour. Write where they hesitated, what they tried, what surprised them, and their exact words.
- Ask open follow-ups. "What were you expecting there?" "What made you pause?" Avoid "was that easy?"
- Repeat with two or three people. Patterns across users matter more than one opinion.
- Write up honestly. List what worked, what failed, and the top problems to fix. Include the failures, especially the ones that sting.
Demonstrated Example
Prototype: the roll-up desk mat.
Task given: "You have finished working. Show me how you would pack your tools away and take them with you."
Observed: both testers rolled the mat easily, but both let the end tools slide out because of the curling edge. One struggled with the ribbon tie and used two hands and about twenty seconds. Neither commented until asked.
Follow-up answer: "I expected it to just clip, the tie felt slow."
Top problems to fix: curling edge lets tools slide; ribbon tie is too fiddly. Both confirm the questions raised in Week 6.
Why this is strong: the learner set a real task, stayed quiet, recorded behaviour and exact words, and listed the failures plainly.
Your Task This Week
- Write a short test script with two or three real tasks and consent lines.
- Test with two or three people, watching before asking.
- Record behaviour and exact words on the feedback form.
- Write up what worked, what failed, and the top problems to fix.
Submission Checklist
| Upload item | Pass standard |
|---|---|
| Test script | Real tasks, not a demo, plus consent lines. |
| Observation notes | Behaviour and exact words from two or more testers. |
| Top problems list | Honest, includes failures, ranked by importance. |
Assessment And Pass Criteria
- The learner gave tasks and observed, rather than demonstrating and asking for praise.
- Consent was taken and testing was respectful.
- Notes record real behaviour, not just opinions.
- The write-up includes genuine problems, not only positives.
Common Mistakes
- Helping the user. If you explain it, you test your explanation, not the design.
- Asking "do you like it?" People are polite. Give tasks and watch instead.
- Only recording praise. The problems are the point.
- Testing one person. Two or three reveal patterns and reduce flukes.
- Leading questions. "That was easy, right?" plants the answer.
Extension Challenge
Test with one person who is unlike your typical user. Edge-case testers often expose problems the design has been hiding.