Website user testing on a small budget

The short answer

Website user testing is watching real people try to complete tasks on your site while they think aloud. For a qualitative test, a common rule of thumb is about five people per round, repeated after you fix the problems. Give them realistic tasks such as finding information or sending an enquiry, and note where they hesitate or fail.

What website user testing is

Website user testing, also called usability testing, is a session in which you give a person a realistic goal and watch how they try to reach it, without helping. You learn where the site confuses people, which labels they misread and which steps they skip. It needs no special software for a first round: a screen-sharing call, a short script and a notes document are enough.

It answers a different question from the other measuring methods a site owner has, which is why it is worth doing even if you already have analytics.

How many people do I need to test with?

For a qualitative test, where the aim is to find problems rather than to measure them, the long-standing rule of thumb from usability research is about five people from the same type of audience per round. The reasoning is diminishing returns. The first few people reveal most of the big problems, and each extra person mostly repeats what you have already seen. The practical lesson is to run a small round, fix what you found, then run another small round, instead of one large study.

Some limits apply. If you have clearly different audiences, such as buyers and suppliers, or customers in different countries, test a few people from each group. Methods that count rather than observe, such as card sorting, need more participants. And a test of this size tells you what goes wrong, not how often, so do not quote the result as a percentage.

Who to recruit

The best testers resemble your customers and have not seen the site before. Colleagues and relatives are quick to find, but they already know your business and your jargon, so use them to check the script and nothing more. Better sources are existing customers, people who recently enquired, an email list or social following, and communities where your audience already gathers. Offer a thank-you in proportion to their time. A note of thanks or a small gift is usual for a short session.

Remote sessions by video call with screen sharing let you reach people in other countries and time zones, and they are the practical choice for a business with an international audience. In-person sessions let you see body language and are easy to run with one laptop and one phone on a table. Either works, as long as you see the screen.

What should I ask testers to do?

Give tasks, not questions. A task is a realistic goal written as a short scenario, with no hint about where to click. "Find out whether the shop delivers to your address" is a task. "Click the Delivery tab" is an instruction, and it teaches you nothing.

Choose three to five tasks that match the main reasons people visit. Put the most important task first, so tiredness does not distort it. Start the session with a first-impression question: show the home page for a few seconds, hide it, then ask what the business does and who it is for. If the answer is wrong, you have already found your first fix.

How to run a session

1. Write down what you want to learn as three or four questions, such as "Can people find a way to contact us?".
2. Write a short script. Welcome the person, explain that you are testing the site and not them, ask permission to record, and ask them to think aloud.
3. Make sure the site is in a usable state, with nothing half-finished. Rehearse the script yourself, then with one friendly person.
4. Give one task at a time, on paper or in the chat, and then stay quiet. Silence feels awkward, but it produces the most useful information.
5. If they ask whether they should click something, reply with "What would you do if I were not here?". Step in only when they have been stuck for a long time, and note that you had to.
6. At the end, ask what was confusing, what they expected to find, and what they would tell a colleague about the site.

Keep each session short enough that the tester does not tire. If two people are available, one asks and one takes notes. If you are alone, record the screen and voice, with permission, so you can review it later.

Test on the devices people really use

Many visitors arrive on a phone, so at least some sessions should use one. Ask testers to use their own device where possible, since screen sizes, browsers and habits differ from yours. Watch for tap targets that are too small, forms that are hard to complete with a thumb and banners that cover content. Why layouts change between screen sizes at all is explained in responsive web design explained.

What to look for while you watch

• Hesitation before a click, and a pointer or finger moving back and forth between options.
• Going back, or reaching for the search box or the footer when the menu should have been enough.
• Reading a label aloud and then choosing something else, which means the label does not match the person's idea of the task.
• Skipping text you expected them to read, or reading it and still getting the answer wrong.
• Form errors, abandoned fields and confusion about what happens after sending. The guide to getting more form submissions goes deeper into these.
• Remarks about speed. Slow pages sometimes come from hosting rather than design, and types of web hosting explains the options.

What people say and what they do often differ. Give more weight to behaviour than to opinions about colour or style.

Mistakes that spoil a test

Questions that steer the answer are the most common. "Was the menu easy to use?" invites a polite yes. Ask the person to find something and watch instead. Other mistakes are testing only after launch, when changes are most expensive, and testing only with staff. Rough sketches and clickable mock-ups can be tested well before any code exists. Treating one person's taste as a finding is another trap: look for the same problem appearing with more than one tester before you act on it.

Do not fix the site while the round is running. If you change a page between sessions, the early and late sessions are no longer comparable.

Turn what you saw into a fix list

Right after the last session, while memory is fresh, list every problem you saw and who had it. Group similar ones. A problem that stopped one person from finishing a task is more serious than one that annoyed three. Rate each as blocking (the person could not finish), serious (they finished with difficulty) or minor (they noticed it and moved on), and fix the blocking ones first.

Make the smallest change that removes the cause, then test again with new people. Before editing a live site, take a copy; what to back up lists what that includes. Where you cannot choose between two fixes, an A/B test can compare them, provided the site has enough visitors. If the findings point to work beyond a quick edit, bring the list to whoever builds the changes. The website development service page describes the build work that Liftup Technologies offers.

Related guides

• Website development
• Types of web hosting explained
• Website backups what to back up
• Website forms how to get more submissions
• Responsive Web Design Explained in Plain English
• Website security checklist for small business

Questions and answers

What is user testing?

User testing, also called usability testing, means watching real people try to complete realistic tasks on your website while they think aloud. You do not help them. It shows where they hesitate, misread labels or give up, which analytics alone cannot explain.

How many people do I need to test with?

For a qualitative test, a long-standing rule of thumb is about five people from the same kind of audience per round. Fix the problems you find, then test again with new people. Use a few people from each group if you have very different audiences, and do not treat the result as a statistic.

What should I ask testers to do?

Give them realistic tasks written as short scenarios, such as finding delivery information, booking, sending an enquiry or buying a product. Avoid telling them where to click. Choose three to five tasks that match the main reasons people visit, and start with a first-impression question about the home page.

Do I need special software to test a website with users?

Not for a first round. A video call with screen sharing, a recording of the session with permission, and a notes document are enough. Specialist tools can help with recruiting or recording later, but the value comes from the tasks and from watching quietly.

Can I test a website before it is built?

Yes. Sketches, wireframes and clickable mock-ups can be tested with the same tasks. Problems found at this stage are cheaper to fix than problems found after launch, because nothing has been coded yet.

Is user testing the same as an A/B test?

No. User testing is small and observational: it shows why people struggle. An A/B test shows two versions to different visitors and counts which performs better, and it needs enough traffic. Many teams use user testing to find the problem and an A/B test to compare fixes.