Website Usability Testing: A Practical UX Guide

You can plan a website, app or online shop in great detail and still discover after launch that it is hard to use. The navigation doesn't make sense, an important button is missing, or the information customers are looking for is buried somewhere nobody ever looks. The result is predictable: people can't do what they came to do, and they close the tab.

To prevent that, product and design teams run UX research: they watch how real people use a service, where they get stuck and why. The good news is that you don't need a corporate budget to do it. Even a small business can test its website with five participants and find the problems that are costing it customers.

In this guide we explain:

• what UX research is and why it saves money;
• the difference between qualitative and quantitative research, and which methods exist;
• how to run a usability test, step by step;
• which tools teams use for testing in 2026;
• how to turn results into design decisions, and which mistakes to avoid.

What UX research is and why you need it

UX (user experience) research looks at how people use a product and which tasks they are trying to complete with it. It helps a team understand what its audience needs, test assumptions about how the product is used and find problems in the existing interface.

The designer who built a website menu may be convinced it is perfectly clear. Yet users get confused by the category names and look for information in the wrong place. A classic example: delivery terms are shown inside the product page, while the customer looks for a separate page in the menu. They can't find it, give up and buy from a competitor.

UX research helps you spot problems like these and understand why they happen. A usability test, for example, shows whether users can complete an action, at which steps they struggle and what stops them from finishing an order. The team then uses those findings to improve the interface, both before launch and after it.

What UX research gives a business

With research you can:

• understand what your audience really needs, not what you assume it needs;
• test user journeys: can people sign up, place an order or send an enquiry without getting stuck;
• catch mistakes before development, when fixing them is cheap;
• measure whether a change you made actually improved anything.

The last point matters most from a financial perspective. Fixing a menu in a Figma prototype takes an hour. Fixing it on a live website, after the ads are already running and leads have been lost, costs far more. That is why research happens at several stages: before building you learn about needs, during planning you test a prototype or wireframe, and after launch you test the finished interface.

What happens when you don't test

A common scenario in redesign projects: an existing app or website that people use reluctantly. Instead of rebuilding everything, a test with a handful of users reveals that the problem is concentrated in one place, such as the catalogue or the sign-up form. Fixing that one mechanism first is much faster and cheaper than a full redesign.

And the opposite scenario: a small change goes live without testing. The "Close" button in a dialog is replaced by a different button in the same spot, and users click it out of habit. One day like that is enough to generate dozens of angry support requests.

Qualitative vs quantitative research

Depending on the goal, UX research falls into two types: qualitative and quantitative.

Qualitative research: why it happens

Qualitative research helps you understand the user's experience. The researcher talks to people, observes what they do and examines specific situations.

Suppose visitors to an online shop hardly ever use the search bar. Interviews may show that they don't know the exact product names and prefer to browse categories. That is a reason to test how easy the catalogue and filters are to use. Or users abandon sign-up halfway through because the form is too long and asks for information without explaining why it is needed.

Quantitative research: how often it happens

Quantitative research compares results in numbers: how often users find the right product, how many reach checkout, at which step most of them leave. For this you use surveys, analytics and A/B tests.

In one sentence: qualitative methods help you understand behaviour, and quantitative methods measure how widespread it is.

In which order to combine them

There is no fixed order. Sometimes you start with the numbers: analytics shows that very few visitors complete a purchase, which is a reason for qualitative research into what their journey looks like and where exactly it breaks. Sometimes it works the other way round: qualitative research uncovers a problem with two possible solutions, you put both into an A/B test and see which one wins.

The main UX research methods

You choose a method based on the task: some methods help you understand behaviour, others measure how common it is. These are the ones teams use most.

In-depth interviews

A one-to-one interview of 30–60 minutes that explores habits, needs and past experience. Interviews are useful not only for an existing product but also before you build anything: if you are planning a booking website for a restaurant, first hear how people book today, what annoys them and what makes them choose one place over another.

Usability testing

In a usability test, the participant completes a task on a website, in an app or in a prototype, while the researcher watches where they make mistakes, what they don't notice and where they hesitate. It is the most direct way to find out whether an interface works.

There are two main formats. In a moderated test, the researcher is with the participant, in person or on a video call, and can ask follow-up questions. In an unmoderated test, the participant completes tasks alone on a dedicated platform that records the screen. It is faster and cheaper, but you can't ask "why did you do that?".

Observation

Observation looks at how people use a service in real conditions. A courier trying to tap small buttons on a phone between deliveries, or a parent booking a clinic appointment while holding a baby, will show you things no test in an office ever could.

Surveys

Surveys help you see which difficulties are most common, but they don't always explain the reasons. A short post-purchase survey or a one-question pop-up on the website can give you a broad picture in a short time.

Analytics, heatmaps and session recordings

Data on page transitions, errors, conversions and drop-offs shows how people use the product and where they stop. Heatmaps show what people click and how far they scroll, and session recordings let you watch real users browsing. Alongside these, it is worth knowing the expert review of an interface against Nielsen's 10 usability heuristics, which catches some problems before you involve users at all.

In practice you combine several methods: analytics points to what to test, interviews provide context, and usability testing checks the solution.

How to run a usability test, step by step

Step 1: Define the question

Decide what the team wants to find out and which decisions the result will inform. Say a Pilates studio's website gets very few sign-ups for a trial class. You need to understand what gets in the way: unclear terms, an awkward way to choose a time slot, or the form itself.

Don't take the first explanation as fact. Research should test hypotheses, not confirm a solution you have already chosen.

Step 2: Choose a method and recruit participants

To learn about habits, use interviews; to test a user journey, use a usability test; to measure how common a problem is, use quantitative methods.

Participants must match the product's audience. Software for accountants should be tested with practising accountants, not with colleagues who are good with interfaces. In fact, it is better not to test with colleagues at all: anyone working on the project knows the context too well and misses what an outsider sees immediately. How to find the right people, how much to pay them and what to ask in screening is covered in our guide to choosing participants for UX research.

A small qualitative study usually starts with five or six participants. That is enough to identify the main difficulties in a single journey, but not to draw conclusions about your whole audience. If the website serves different audiences, such as private and business customers, test each group separately.

Step 3: Prepare tasks and questions

In a usability test, give a concrete task, but don't explain where to click. Instead of "Click 'Contact' and fill in the form", write: "You want to find out whether the studio is open on Saturday morning. Try to find the answer." That way you test the interface, not the participant's ability to follow instructions.

In an interview, ask about past experience, not hypotheticals. Instead of "Would you use an app like this?", ask "How did you solve this problem last time?" or "What was inconvenient about it?". People are good at describing what they did and bad at predicting what they will do.

To test an interface before development, an interactive prototype in Figma is enough, containing only the screens of the journey you are testing.

Step 4: Run the test

At this stage, don't help the participant complete the task. The moment you drop a hint, you no longer know whether the interface is clear. Ask the participant to think aloud, and note mistakes, pauses, repeated actions and moments of hesitation.

Before you record anything, get the participant's explicit consent and explain what you will do with the recording and when you will delete it. If your participants are in the EU, the GDPR applies to recordings and personal data, so collect only the information you genuinely need and store it securely. This is general guidance, not legal advice: for anything beyond a simple study, check your obligations with a specialist.

Step 5: Analyse the results

Work in order: where the difficulty is, what the likely cause is, and what the fix might be. For example, three out of five participants didn't notice the link to the delivery terms. That is the observation. The likely explanation: the link isn't prominent enough. A possible fix: change its design or position.

A single observation doesn't mean you need to rebuild anything. But if it is a serious obstacle, such as a missing navigation step that makes it impossible to complete an order, you have to fix it.

Document your findings in a short report, a list of observations or a customer journey map: what you found, how many participants ran into it and which solutions are worth considering. A journey map also helps you spot opportunities: where it makes sense to offer a complementary product or an upgrade, such as a power supply for someone buying an LED strip.

Step 6: Fix and test again

The team implements fixes in order of priority, and the designer updates the screens. Where needed, run a follow-up test to confirm the experience really is easier now. Sometimes changing a single button label is enough; sometimes you need to rebuild an entire journey. Several small rounds of testing and fixing are better than one big study once a year.

Usability testing tools in 2026

You don't need a lab. Most testing today happens remotely, with familiar tools and a video call.

Prototype and video call

Figma for an interactive prototype, and Zoom or Google Meet for a moderated test with screen sharing and recording. This is the basic set-up, and it costs next to nothing.

Remote testing platforms

Platforms such as Maze, Lyssna and UserTesting let you run unmoderated tests: you upload a prototype or a link to your website, set tasks and get recordings, completion times and success rates. Some of them also offer a paid participant panel, but before relying on it, check that it has enough people who match your audience in the countries and languages you serve.

Recordings and heatmaps on a live website

Microsoft Clarity is free and provides session recordings and heatmaps, including "rage clicks" (repeated clicks on something that doesn't respond). Hotjar offers similar features alongside short on-site surveys. Tools like these suit a website that is already live: you watch dozens of sessions and see where people get stuck. If your visitors are in the EU, make sure these tools only run with the consent your cookie banner collects.

Analytics

Google Analytics 4 shows your funnels: how many people reached the product page, how many added to the basket and how many completed the purchase. When you see a sharp drop at a particular step, you know what to look at in your next usability test.

Usability testing for international websites: what to watch for

If your website serves customers across several countries, a few things are worth building into the test plan from the start.

Mobile first

A large share of traffic to small and mid-sized business websites comes from phones, especially from social media campaigns. So test the most important journeys on mobile first: the contact form, the WhatsApp or chat button, appointment booking and payment.

Multilingual audiences

A website for European customers often runs in two or three languages at once, for example English, Greek and Portuguese. Test each language version with native speakers, not just a translation that looks fine to the team. Longer words in one language can break buttons and menus that looked perfect in English.

Forms, formats and mixed text

International forms create problems you won't see in a single-market website: phone fields that reject country codes, address forms that assume one postcode format, dates written day-first in one market and month-first in another, prices in the wrong currency. If you also serve right-to-left languages such as Arabic, watch for arrows pointing the wrong way and mixed-direction text breaking apart. Pay particular attention to forms, checkout steps and mixed-language content.

Accessibility

Since June 2025, the European Accessibility Act has required many consumer-facing digital services in the EU, including a large share of online shops, to meet accessibility requirements, with the harmonised standard EN 301 549 built on WCAG. Usability testing with people who use a screen reader or magnification reveals problems that automated checkers miss. We cover the requirements in more detail in our guide to website accessibility. As with data protection, treat this as an overview, not legal advice.

How to turn results into design decisions

UX research helps you understand the user experience, but it is the designer who chooses the solution.

Not every request is a specification

A user may ask for an extra button when what they actually want is to repeat their previous order faster. If you simply fulfil the request, the interface becomes more complicated and the original problem remains. Look for the need behind the request.

Link every recommendation to an observation

If users can't find a page, work out what gets in the way: an unclear name, its position in the menu or a different expectation. Every recommendation in the report should point to a real observation, so it is easy to explain to a manager and a developer.

Prioritise by business impact

Not every finding is equally urgent. Rank problems on two axes: how many users ran into them and how much they hurt the business goal. A problem that prevents people from submitting the lead form comes before a problem on the "About" page. Involve the developer early: sometimes a small design change solves what looked like a big project.

Common UX research mistakes

Research designed to confirm a hypothesis

The most common mistake is running research to prove you were right. Questions should be neutral and allow for different answers, even if you already have an idea where the weak spots are. Good research uncovers problems nobody on the team had thought of.

The wrong participants

Fellow designers will find their way around a prototype quickly because they know interfaces well. That doesn't mean your audience will find it just as easy. The same goes for the business owner: "I'm the CEO and I want it this way" is a position that fails more often than not, because the owner is rarely the typical customer.

Designing for taste instead of behaviour

A solution the designer finds less attractive often works better than the polished version. A familiar, predictable interface is clearer to users than an innovative concept. Test what works, not what impresses.

Jumping to conclusions

If one user gets stuck on the sign-up form, that doesn't mean you need to rebuild it. Look at all the results: if most participants hit the same problem, it is systemic. If only one did, it may be human error, or renaming a single field may be enough.

A report that stays in a drawer

Getting results isn't the end. Decide what to fix first and when, even if not every recommendation is implemented straight away. Research that nobody acts on wastes your time and your participants' time.

How RJL Studio helps you test and improve your website

In every UX/UI design project, we test the key journeys with real users before the design goes into development. On existing websites, we start with analytics and session recordings, find where visitors get stuck, run a focused usability test and fix issues in order of priority. We build and maintain the websites themselves in Webflow and on other platforms, so fixes go live quickly.

If your website gets traffic but not enough enquiries, tell us about your project and we will work out together where users drop off.

Frequently asked questions

How long does UX research take?

It depends on the task, the method and the number of participants. A usability test of a single journey can be completed in a few days, including preparation and analysis. Studies with many interviews, several audiences or quantitative analysis can take a few weeks.

How many participants do you need for a usability test?

Small qualitative tests usually start with five or six participants per audience group. That is enough to find the main problems in a journey. Quantitative research needs a much larger sample, and its size depends on the goal and the method.

Can you do UX research without a budget?

Yes. You can interview a few customers yourself, test a prototype with people from your target audience and watch session recordings in a free tool such as Microsoft Clarity. The key is to define the question in advance, avoid hinting at answers and keep observations separate from assumptions.

What is the difference between usability testing and A/B testing?

Usability testing is a qualitative method: you watch a few participants and understand why they get stuck. A/B testing is a quantitative method: you show two large groups of visitors different versions and measure which one converts better. Usually, usability testing finds the problem and A/B testing decides between solutions.

When should you test: before launch or after?

Both. Before launch you test a prototype, when fixes are cheap and quick. After launch you test the live website based on real data and fix whatever hurts conversions. A good website improves in regular rounds, not through a one-off test.