User testing

You know your website too well to see it with fresh eyes. A user test shows where a stranger pauses, hunts around or gives up. You set a task, such as "find out what it costs to replace your windows", and watch. You explain nothing along the way.

Five people go a long way

Jakob Nielsen wrote in 2000 that a test with five users finds around 85 per cent of the usability problems, and that several small rounds give more than one big one. So you do not need an agency or a lab to get started.

How it differs from A/B testing

An A/B test tells you which version wins, but not why. A user test tells you why, but with too few people to give reliable numbers. The two complement each other. A heatmap sits somewhere in between.

How to do it this week

Find three to five people who resemble your customers, not colleagues who know the site. Give them two tasks: find a price, and get in touch or book. Ask them to say out loud what they are thinking. A video call with screen sharing works fine. Write down where they hesitate. Fix first the problem that more than one person runs into.