User Testing in eCommerce: How to Find Conversion Killers Before Your Customers Do

User testing in eCommerce: how to find conversion killers before your customers do

User testing in eCommerce is how you watch real people use your shop, before they become customers or leave. It is not about opinions. It is about behaviour. What do they click? Where do they hesitate? Where do they give up?

That is the data analytics does not give you.

Why analytics alone is not enough

Google Analytics tells you that 67% of visitors abandon the checkout. It does not tell you why. Worked example: a furniture shop with 80,000 sessions a month shows exactly that pattern. The analytics data looks normal.

The test videos show that customers do not understand the shipping cost logic, because the information sits behind a tooltip. Replacing the tooltip with a plain table cuts the abandonment rate by 14%, and that is the kind of fix only a user test surfaces.

That is the difference. Analytics tells you that something is happening. User testing tells you what is happening.

The three methods we use

Moderated user testing

A moderator guides the participant through defined tasks and asks follow-up questions. This gives the deepest insights, because you can dig deeper in real time. The downside: it is more expensive and slower. We use it for complex products or B2B shops, where the buying process does not explain itself.

Unmoderated remote testing

The participant completes tasks alone, recorded via screen and camera. Platforms like UserTesting or Maze deliver results in 24–48 hours. This is our standard method for quick validation. Cost per participant: 30–60 euros. With 5–8 participants per round, you get feedback for under 500 euros that saves you weeks.

Think-aloud protocol

The participant says out loud what they are thinking as they navigate. It sounds awkward. But it is the fastest way to understand mental models. If someone says “I’m looking for the size chart” and you know the chart sits right below the product image, you have a design problem, not an information problem.

Myths that cost shops dearly

“User testing is too expensive”

Wrong. An unmoderated test with 5 participants costs less than an hour of consulting. The ROI is measurable. Worked example: a 45-minute test at a fashion shop with 40,000 sessions a month uncovers a filter problem that blocks 12% of mobile users. The fix takes 2 hours of development and is worth around 8,000 euros a month.

“But we already have analytics and heatmaps”

Heatmaps show where people click. They do not show why someone sits on a page for 30 seconds without scrolling. Reading? Searching? Waiting? User testing gives you the answer. We use heatmaps and user testing together: heatmaps identify hotspots, tests explain them.

“Our users are different”

Every shop believes that. The truth: people buy online by the same principles. Trust signals have to be visible. The checkout has to work without friction. Product information has to be instantly clear. The specific content varies. The psychological barriers do not.

What user tests show again and again

The problems are not exotic. They are predictable, once you know what to look for.

Mobile checkout: in a typical round of user tests, 6 or 7 out of 10 find friction in the mobile checkout. Keyboards covering forms. Buttons too close to the edge of the screen. Autofill that does not work. These are not edge cases. They are the norm.

Filters and search: Users expect filters to respond instantly. If a filter click triggers a page reload, you lose users. In a test with an electronics shop, 3 out of 8 participants gave up because filtering was too slow. The solution: AJAX-based filtering. Effort: low.

Price transparency: Hidden shipping costs are the classic conversion killer. But unclear pricing for variants or subscriptions also leads to drop-offs. We always test price perception explicitly. If users cannot tell what they will pay within 3 seconds, you have a problem.

How we build user testing into our CRO process

User testing is not a one-off event. It is a recurring step in the optimisation cycle.

1. 1
   

   Hypothesis from data.

   We identify a problem through analytics, heatmaps or A/B testing data. Example: high abandonment on the product page.

2. 2
   

   Test design.

   We write 3–5 tasks that address the problem. Not: “Find a product.” But: “You are looking for a gift for someone who is into technology. Find something suitable under 50 euros.”

3. 3
   

   Running the test.

   5–8 participants per round. We test our target audience and fringe groups, which often reveal the biggest comprehension problems.

4. 4
   

   Synthesis.

   We look for patterns, not individual cases. If 2 out of 8 users have a problem, we note it. If 5 out of 8 do, we prioritise it.

5. 5
   

   Implementation and validation.

   The results feed straight into the CRO process. Changes are validated with an A/B test, not simply switched live.

Costs and time at a glance

These costs are not an expense. They are insurance against bad decisions. A single avoided mistake pays for a whole year of testing.

Frequently asked questions about user testing in eCommerce

How many participants does a user test need?

Five to eight participants uncover 80–85% of usability problems. That is not gut feeling. It is what Nielsen Norman Group data has shown for 20 years. More participants bring diminishing returns. Two rounds of 5 participants are better than one round of 10.

How do I find the right test participants?

Platforms like UserTesting or TestingTime let you filter by demographics and behaviour. We define age, gender and buying behaviour: “Has shopped online in the last 3 months.” That matters more than an exact match with your target audience.

When should you use user testing?

Before a relaunch, after a relaunch, when your conversion rate stagnates, before big A/B tests. Not only when something is obviously going wrong. The best time is before you invest time and money in a solution nobody understands.

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

User testing explains the why. A/B testing measures the what. User testing produces hypotheses. A/B testing validates them. We use both in sequence: user tests to find problems, A/B tests to prove solutions.

Does user testing work for small shops too?

Yes. It costs less than most shop owners assume. An unmoderated test with 5 participants costs under 300 euros. For shops with 10,000+ sessions a month, it almost always pays off. Below that size, we prioritise other methods.

How long does a typical user test take?

15–30 minutes per participant. Longer is rarely necessary. Concentration fades, and the most important problems show up in the first tasks. We plan 5 tasks per test, not 15.

How do you evaluate user testing results?

By patterns, not individual opinions. We build a table: task → success rate → time → errors → severity. If 60% of users cannot complete a task, that is a red flag. Everything else is prioritisation.

Is user testing worth it before a shop relaunch?

Absolutely. A relaunch is the most expensive moment to make mistakes. We test wireframes, then prototypes, then the beta version. Every phase you test earlier saves money. A problem in a wireframe takes hours to fix. The same problem live costs thousands.