Conversion and UX

User Testing

Also called usability testing, moderated testing

Watching real people attempt real tasks on a site, to learn why they hesitate, misread or give up.

Quick facts: User Testing

Category
Conversion and UX
Also called
usability testing, moderated testing
Level
Intermediate
Affects
Conversion rate, navigation, form design, copy clarity
Where to see it
Screen sharing and recording tools, unmoderated testing platforms, prototypes
In this article4
  1. How user testing works
  2. Why user testing matters
  3. Where user testing goes wrong
  4. How to act on it

How user testing works

You give someone a task rather than a tour. “Find out whether this company works with businesses like yours, and get a price” is a task. Then you watch, and you keep quiet. The value is in the pauses, the scrolling back and forth, the moment they read a heading twice, and the question they ask out loud that your page was supposed to answer.

Sessions come in two shapes. Moderated means you are present, on a call or in the room, and can ask why they did something — slower to arrange, much richer. Unmoderated means the participant records themselves working through a task list on their own time, which is cheaper and easier to run at odd hours across time zones. Both need a real task, a participant who resembles a real customer, and someone taking notes on what happened rather than what you hoped would happen.

Why user testing matters

Analytics tell you where people stopped. They cannot tell you why. A drop-off on a form could be a confusing field label, a price the visitor was not expecting, a missing payment option, or a button that looks like a heading. Guessing between those and rebuilding the page is expensive; watching two or three people attempt it usually settles the question in an afternoon.

It also breaks the curse of knowing your own business. You wrote the labels, so they read as obvious to you. Watching a stranger stall on the word you chose is the fastest cure for that blindness I know of, and it makes every later argument about design far shorter, because there is a recording rather than an opinion.

Where user testing goes wrong

Leading is the biggest risk. Asking “was that easy to find?” invites a polite yes; asking someone to talk through what they are looking at gets the truth. Helping is the same mistake in a kinder form — the moment you point at the button, the test is over and you have learnt nothing.

Recruiting the wrong people is next. Colleagues, friends and anyone who already knows the business will skate through a journey a real prospect would fail. So will testing only on a fast laptop when your customers arrive on mid-range Android phones over mobile data, which is the normal case in Nepal.

The third failure is collecting sessions and changing nothing. Findings that sit in a document are an expense, not research.

How to act on it

Start small and specific. Pick the one journey that matters most — enquiry, booking, checkout — write three tasks, and find a few people who could plausibly buy from you. A screen share and a recording is enough equipment; test in the language your customers actually use.

Write findings as observations, not opinions: what the person was trying to do, what they did, where they stopped. Sort them by how many people hit the same wall and how much damage it does, fix the worst two or three, then test again. Pair it with session recordings from live traffic to see whether the same pattern shows at scale, and treat the whole loop as ordinary conversion rate optimisation rather than a one-off project.

Do and do not

Do

  • Give a real task, then stay quiet
  • Recruit people who could genuinely buy from you
  • Test on the devices and networks your customers use

Do not

  • Ask leading questions or point at the answer
  • Test only with colleagues and friends
  • File the findings without fixing anything

Questions people ask about this

How many people do I need to test with?

Fewer than most people expect. The point of a small round is to find serious problems, and serious problems repeat quickly — by the third or fourth person you will usually be watching the same wall being hit again. Run a short round, fix what you found, then run another. Large samples belong in quantitative testing, not in this.

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

User testing tells you why something fails; A/B testing tells you which of two versions performs better. Watching people gives you the explanation and the ideas, but no proof of impact. A split test measures the impact but cannot suggest the idea. On low-traffic sites, watching people is often the only one of the two you can realistically run.

Can I test a design before the site is built?

Yes, and it is the cheapest moment to do it. A clickable prototype, or even printed screens with someone pointing at where they would tap, will expose confusing labels and missing steps before any code exists. The tasks are the same; only the fidelity changes. Fixing a wireframe costs a conversation, fixing a live site costs a project.

Related terms

Found this useful?

Share it, or ask an AI to summarise it

Back to the glossary

Knowing the term is the easy part

Applying it to your own site and budget is the work. Book a call and I will tell you what actually applies to you.