Usability testing is watching real people try to do something on your website or app, like book an appointment or find a price, while you note where they hesitate, misclick or give up. It shows you problems you cannot see yourself because you already know how your site works.
How a usability test runs
You pick a few realistic tasks ("You need your gutters cleaned next week. Book it.") and recruit people who resemble your customers. Each participant tries the tasks while thinking aloud, saying what they expect and what puzzles them. The facilitator does not help or explain; silence when someone is stuck is the point. Afterwards you list the problems, how many people hit each one and how badly it blocked them.
Tests can be moderated, with you present in person or on a video call, or unmoderated, where people record themselves using an online platform. Moderated sessions let you ask follow-up questions; unmoderated ones are quicker to run in numbers. You can test a live site, a competitor's site or a clickable prototype before anything is built.
How many people you need
Jakob Nielsen of Nielsen Norman Group argued in a widely cited 2000 article that testing with about five users finds most of the common problems in a design, and that several small rounds beat one large one. That guidance applies to finding problems, not to measuring statistics. Five friendly strangers will tell you that your booking button is hard to find; they cannot tell you your exact conversion rate.
Five testers at a physio clinic
A clinic asks five people to "book a first appointment for a sore back". Three of them open the "Treatments" page, read about back pain, and then cannot find a booking link on that page. Two look for prices and assume the clinic is expensive because none are shown. Two fixes follow: a booking button on every treatment page and a short price list. Neither problem had shown up in the owner's own checks, because the owner always went straight to the "Book" menu item.
The right moments for a usability test
Before a redesign, after launching a new booking or checkout process, and whenever analytics show a page where many people leave. It complements numbers from tools like a heatmap or an A/B test: those show what happens, testing shows why.
Usability testing mistakes
- Leading the participant. "Click the blue Book button" tests nothing.
- Testing only with colleagues or family. They know too much about the business.
- Asking opinions instead of watching behaviour. "Do you like it?" matters less than whether they finished the task.
- Testing on desktop when customers use phones. Match the device your visitors actually use.
Related terms
- A/B testing: compares versions with live traffic rather than observed sessions.
- Heatmap: shows aggregate click and scroll behaviour.
- Prototype: something to test before building.
- UX design: the practice that relies on testing to check its decisions.