Stockertim Dictionary · Artificial intelligence
User testing
Also called usability testing, user tests, usability test
In short
Watching real people try to complete tasks with a design, in order to find where it confuses, slows or fails them.
- Beginner
- Process
- Checked
In detail
User testing means giving a design to people who resemble its real audience, asking them to complete realistic tasks and observing what happens. The facilitator watches where people hesitate, misread labels, take wrong turns or give up, and asks them to think aloud. The aim is to find problems with the design, not to test the people.
Tests can be run with paper sketches, wireframes, clickable prototypes or live products. A small round of about five participants often reveals the most common problems, and several small rounds during a project tend to find more than one large test at the end. Results are recorded as task success, time, errors and the reasons behind them.
User testing matters more when AI helps design. Generated layouts, copy and flows can look plausible while missing what real users need, and AI tools that simulate users are not a substitute for real people, especially for accessibility, local language and unusual situations. Testing with people who use assistive technology catches problems that automated checks miss.
At work, teams plan tasks from the brief, recruit suitable participants with their consent, record findings and fix the most serious problems before launch.
Worked example
Northwind Bikes tests a new repair-booking screen with five customers.
- Task: book a repair for next Tuesday.
- Four of the five testers miss the date picker: 4 / 5 = 80%.
- The team moves the date picker above the service list and renames the button "Choose a date".
- A second round of five testers sees all five finish: 5 / 5 = 100%.
One small change found by watching real people raised task success from 20% to 100% in the second round.
In practice
Design and product teams run short tests at each stage, from wireframe to launch. They write tasks from the brief, include people with different needs, record evidence and change the design based on what users did, not what they said they liked.
Watch out
- User testing observes behaviour on tasks; a survey of opinions is a different method.
- Simulated users from an AI tool can help prepare a test but cannot replace real participants.
Learn it properly
Official sources
- WCAG standards and guidelines · W3C Web Accessibility Initiative
- Writing for GOV.UK · Government Digital Service
Facts last checked on 3 November 2026.
Education, not advice: the Stockertim Dictionary explains what terms mean. It never recommends buying or selling anything.