User Testing
Usability testing with real users
What is User Testing?
User testing (also called usability testing) is a research method in which real or typical users carry out specific tasks on a website, online shop or app while you observe them. A facilitator gives a task, for example "find the price of installation and send an enquiry", watches where the person hesitates, makes mistakes or gives up, and asks why. The result is not an opinion on the visuals but a list of specific places where the website stops visitors from completing what they came to do.
Why It Matters
Website owners and the people who built the site know it too well to see its problems. User testing reveals them before they cost money: a confusing form, a hidden price or an unclear button. A small test with five people takes a few days and often uncovers issues that reduce enquiries more than weak advertising does. Fixing what you find raises the conversion rate and improves the return on the campaigns that bring traffic to the website.
Real-World Examples
- 1Five people from the target audience are asked to find a price and send an enquiry on a construction company's website; three of them miss the contact form hidden in a drop-down menu.
- 2A remote moderated test over a video call: the participant shares their screen, comments aloud on what they are doing and the facilitator asks follow-up questions.
- 3An unmoderated test in an online tool: the participant completes written tasks alone and the researcher receives a screen recording.
- 4Testing a Figma prototype before development, while changing the structure still takes only a few hours.
- 5Model (fictional) example: a coffee shop finds that people overlook the subscription option, moves it above the add-to-cart button and tracks the share of subscription orders in analytics.
How user testing works
According to Nielsen Norman Group, every test has three elements: the facilitator, the tasks and the participant. The facilitator explains the session, gives a task and then mostly stays quiet and observes. Questions come after the task so that leading questions do not influence the participant's behaviour.
- Tasks are realistic situations ("You need your roof repaired by the end of the month. Find out whether the company can do it in time."), not instructions such as "click Contact".
- Participants belong to your target audience or share its needs. Colleagues, family and people who helped build the website are not suitable.
- The think-aloud method: the participant keeps saying what they are looking for, what they expect and what confuses them. Most findings come from these comments.
The output is a list of problems ranked by severity: what blocks the task completely, what slows people down and what is merely irritating.
Usability testing methods
- In-person moderated: the facilitator and participant sit at one computer or phone. The most context, but the most time.
- Remote moderated: the same approach over a video call with screen sharing. Usually the best value for a small business.
- Remote unmoderated: an online tool delivers the tasks and records the screen, and the participant takes the test alone whenever it suits them. Fast and cheap, but with no chance to ask follow-up questions.
- Qualitative vs quantitative: a qualitative test looks for problems and their causes; a quantitative test measures numbers, most often task success and time on task, typically to compare versions over time.
- The SUS questionnaire after the test: the System Usability Scale (John Brooke, 1986) has 10 questions and a score from 0 to 100; the average across roughly 500 studies analysed by Jeff Sauro is 68. The score shows how people perceive the website but not what to fix.
User testing differs from analytics and A/B tests: Google Analytics 4 or heatmaps show what people do and where they leave, while testing with people explains why. An A/B test then checks which of two versions performs better with real traffic.
How many users you need
For a typical qualitative test of a single user group, Nielsen Norman Group recommends five participants: they uncover most of the common problems. Several small rounds with fixes in between work better than one large test. If you have two distinct customer groups (for example businesses and households), test each one separately. Quantitative measurement needs considerably more people, so small businesses rarely run it. More on the five-user principle is in the UX entry.
How to run user testing on a website step by step
- Set a goal: one or two questions you want answered (why few people send an enquiry, why they abandon the basket). Analytics data helps.
- Write 3–5 tasks based on why people visit the website. Each task needs a clear end point so you can tell whether it was completed.
- Recruit five participants from the target audience, agree a time and a small thank-you for their time (a voucher or discount). Ask for consent to record.
- Run the test: 30–45 minutes per person, screen and audio recording, with a second person taking notes. Do not help and do not defend the website.
- Analyse: list the problems, how many people hit each one and how severe it is. Fix what blocks task completion first.
- Test again after the fixes, ideally with new people.
According to Nielsen Norman Group, the simplest study takes about three days: one day to plan, one day to test five users and one day to analyse the findings and turn them into recommendations.
When to test: from wireframe to live website
Testing early is cheapest. A wireframe or clickable prototype lets you check the structure and menu labels before anything is built. On a live website, testing makes sense before a redesign, after a major change (a new checkout or form) and when the site gets traffic but has a low conversion rate. Between tests, the free Microsoft Clarity tool with session recordings and heatmaps shows which pages to check with people.
Common user testing mistakes
- testing with colleagues and acquaintances who know the business,
- leading questions ("Do you like how clear this is?") and helping participants when they get stuck,
- asking for opinions on the design instead of observing behaviour,
- too many tasks at once, leaving the participant tired after half an hour,
- testing without follow-up changes, so the results end up in a drawer.
When to bring in specialists
You can run a simple test yourself. Help pays off when you are planning a new website or online shop and want to check the structure on a prototype, or when paid advertising brings traffic but no enquiries. In website development and e-commerce development at Weboptim, we start from analytics data and the customer's path to the goal, and add testing with people where it affects the budget. The principles of a clear website are also covered in Web design principles and trends.
Frequently asked questions
What is user testing?
User testing is a method in which real users carry out specific tasks on a website or app while you observe them. The aim is to find where people hesitate, make mistakes or leave, and to understand why. Unlike analytics, it shows the causes of problems, not just the numbers.
How many people do I need for usability testing?
Five participants are enough for a typical qualitative test of one target group, and they uncover most of the common problems. Several small rounds with fixes in between are better than one large test. Run a separate test for each clearly different customer group.
What is the difference between user testing and A/B testing?
User testing works with a few people to find out why a website does not work and what to fix. An A/B test compares two finished versions of a page with real traffic and measures which one brings more conversions. Typically you test with people first and then use an A/B test to confirm the chosen change.
How much does website user testing cost?
You can run a simple test with five people yourself for the cost of participant incentives and about three days of work on planning, testing and analysis. Tests with several user groups, in multiple countries, quantitative studies or eye tracking cost more and are rarely needed by small businesses.
Can user testing be done remotely?
Yes. In a remote moderated test, the participant shares their screen over a video call and the facilitator asks follow-up questions. In an unmoderated test, the participant completes tasks alone in an online tool and you receive a screen recording. Remote tests usually save both time and money.