Put it in someone's hands

There's a single test that beats every opinion, every guess, every "I think it's pretty clear." It costs nothing and it's slightly painful:

Hand your thing to a real person. Say nothing. Watch.

Not your best friend who wants you to feel good. Not you, pretending to be new. A real person, using it for the first time, while you sit on your hands.

You'll learn more from watching one person get stuck than from a hundred of your own guesses. The stuck moment is the truth.

One idea: watch, don't rescue

The whole test lives or dies on one hard rule: when they get stuck, you don't help. The second you jump in with "oh, you just tap there," you've destroyed the experiment — because the real user, at home, won't have you sitting next to them.

Your job is to watch and take notes:

Watch for…	Because it tells you…
Where they pause	Something's unclear right there.
What they tap by mistake	Your layout is pointing the wrong way.
What they say out loud	"Wait, what?" marks the exact broken spot.
Where they give up	The most important thing to fix, first.

Ask them to "think out loud" as they go. Every confused mutter is a free gift — a bug report in plain English. And notice: this is the claim-check loop from Unit 4, aimed at your own creation. You're testing the claim "this is easy to use" against real evidence, instead of trusting your first answer.

Do the thing

Run one real test. Find one person, hand them your thing, give them a simple goal ("try to make a summary with this"), and then go quiet. Write down:

• First place they hesitated: ______
• Something they did that surprised you: ______
• Where they got stuck or gave up: ______
• The one fix that would've helped most: ______

The urge to explain will be enormous. Beating that urge is the skill.

Quick check. If you found yourself explaining, you're not alone — everyone does it the first time, and it always feels helpful. But a fix you had to say out loud is a fix the product still needs. The goal isn't for the test to go smoothly. The goal is to find the sticky spots while it's still cheap to fix them — with one friendly tester, not a hundred strangers.

Why this matters

Watching a real user is the highest-value thing a builder can do, and almost nobody wants to do it because it stings a little.

• One quiet test beats a week of guessing. Real behavior settles arguments opinions can't.
• Their confusion is data, not judgment. Every stuck moment is a map to a better version.

Last lesson of the whole course: turning what you learned into a shipped thing — launching small, improving with feedback, and doing it responsibly. Let's finish this.