User Testing on a Budget: Find Problems With 5 People

Analytics shows what people do on your site. User testing shows why. Watching even a handful of real people attempt real tasks reveals confusion that no report will. It's one of the cheapest, fastest ways to improve a website.

Why small tests work

Most serious usability problems are encountered by many people. Small tests of around five participants often reveal the main recurring issues, particularly early on. Run another small test after fixing them. It's more productive to run several small rounds than one large one.

What you need

• A clear goal, such as "Can visitors find our prices and request a quote?"
• Three to five realistic tasks
• Five or so participants who resemble your customers
• A way to observe, in person or by video call, with screen sharing
• Permission to record, if you do

Write good tasks

Give scenarios, not instructions, and don't use your site's own labels. Instead of "Click on Services," try "You need someone to fix a leaking tap. Find out whether this company can help and how to get a quote."

Good tasks:

• Find specific information, like opening hours
• Compare two options
• Complete a form or purchase
• Find help or contact details
• Use search

Find participants

• Existing customers or subscribers (offer a thank-you gift or small voucher)
• Friends of friends who fit your audience, though avoid people who know your business well
• Social media followers or local groups
• Paid participant panels and testing platforms for remote testing

Avoid testing only colleagues or those who already know the site.

Run the session

1. Explain it's the site being tested, not them. There are no wrong answers.
2. Ask them to think aloud, saying what they're looking at and thinking.
3. Give one task at a time.
4. Stay quiet. Don't help or explain. If they struggle, note it and ask "What are you expecting to happen?" afterwards.
5. Watch behavior, not just words. Hesitation, backtracking, and misclicks are clues.
6. Finish with a few questions: What did you like? What confused you? What would make you more likely to contact us?

Allow 20 to 30 minutes per session.

Unmoderated and quick options

• Remote unmoderated tests where participants record themselves completing tasks
• Five-second tests, showing a page briefly and asking what they remember or what it's for
• First-click tests, asking where they'd click to do something
• Guerrilla testing, where you ask people in a café or at an event to try a task for a few minutes

Record and analyze

Note each problem, how many people hit it, and how serious it was (blocked the task, slowed it, or just annoyed). Look for patterns across participants. Prioritize problems that stop people reaching your main goal.

Fix and retest

Make changes, then test again. Combine findings with other data such as heatmaps, surveys, and funnel reports.

Include accessibility

Where possible, include people who use assistive technology or have different needs. See how to test with a screen reader.

Respect privacy

Get consent, explain how recordings will be used, store them securely, and delete them when no longer needed. See privacy policy basics.

The bottom line

Watch five real people try to use your site and you'll see problems you'd never guess. Cheap, quick, and eye-opening.