Usability testing sounds like something only large companies can afford: research labs, one way mirrors, and specialist teams. In reality, the most useful version costs very little. If you can run a video call and share a prototype, you can run usability testing on a budget and learn more in an afternoon than in weeks of internal debate.
This guide shows you how to plan, run, and learn from a small usability study, step by step.
What Is Usability Testing?
Usability testing is a research method where real people try to complete realistic tasks with your website, app, or prototype while you observe. The goal is not to ask whether they like the design. It is to see where they succeed, where they hesitate, and where they get stuck.
It is one of the most direct user testing methods available, because it replaces opinions with evidence. Instead of arguing about whether a button is clear, you watch five people try to find it.
Why Usability Testing on a Budget Works
Teams often delay testing because they think it needs a large sample, special equipment, or a dedicated researcher. None of these is required for useful results.
Nielsen Norman Group has long recommended testing with around five users per round. Small groups tend to reveal most of the obvious problems, and running several small rounds as you improve the design is more useful than one large study. Combined with free or low cost video tools, that makes usability testing one of the best value activities in any project.
What You Need to Get Started
- A clickable prototype or a working version of the product
- A video call tool that lets participants share their screen
- Five people who match your real users
- A short usability test script with three to five tasks
- One person to lead the session and, ideally, one to take notes
- A simple thank you for participants, such as a gift card
How Many Users to Test
| Situation | Recommended Participants | Why |
|---|---|---|
| Early prototype check | 5 | Finds the most obvious problems quickly |
| Two distinct user groups | 3 to 5 per group | Different groups often struggle in different places |
| Before a major launch | 5 per round, 2 to 3 rounds | Confirms fixes and catches new issues |
| Comparing two designs | 8 to 10 or more | Enough sessions to see a real difference |
Step 1: Recruit the Right People
Testing with colleagues or friends is better than nothing, but they already know too much. Look for people who match your audience: existing customers, people from your mailing list, or users recruited through a testing service.
A short screener helps you avoid the wrong participants. Ask one or two questions that confirm they fit, such as their role, how often they shop online, or which tools they already use. Schedule a couple of backups, because some people will cancel.
Step 2: Write a Usability Test Script
A usability test script keeps sessions consistent, so you can compare what different people did. It usually has four parts:
- A short introduction explaining that you are testing the design, not the person
- A few warm up questions about their background
- Three to five realistic tasks
- A few closing questions about their overall experience
Good tasks describe a goal, not the steps. Instead of “Click Pricing and choose the middle plan”, say “You run a team of ten. Find out how much this would cost you each month.” That way you see how people actually look for information.
Step 3: Run the Session
Introduce yourself, ask for permission to record, and remind the participant there are no wrong answers. Ask them to think out loud as they work. Then give the first task and stay quiet.
The hardest part of usability testing is not helping. When someone gets stuck, resist the urge to explain. Instead ask neutral questions like “What are you looking for?” or “What would you expect to happen?” Most sessions take 30 to 45 minutes.
Note where people:
- Hesitate before clicking
- Click something unexpected first
- Use words that differ from your labels
- Say they expected to find something that isn’t there
- Give up or ask for help
Remote Usability Testing vs In Person
Remote usability testing has become the default for most teams, and for good reason. It is cheaper, faster to schedule, and lets you reach people anywhere. In person sessions still have a place, especially for physical products, complex workflows, or audiences who are less comfortable with technology.
| Method | Best For | Trade Off |
|---|---|---|
| Remote moderated | Most websites and apps | Depends on participant’s connection and setup |
| Remote unmoderated | Quick checks of simple tasks | No chance to ask follow up questions |
| In person moderated | Complex flows and specialist audiences | More time and cost per session |
Step 4: Turn Notes Into Fixes
After all five sessions, list every problem you observed and how many people ran into it. Group similar issues together, then rank them by frequency and severity. A problem that stopped three people from finishing a task matters more than a cosmetic issue one person mentioned.
Fix the top three issues first, then test again with a new group. Each round makes the product noticeably easier to use, and the improvements are easy to show to stakeholders because you have recordings of the before.
Common Mistakes to Avoid
- Leading participants with questions like “Was that easy?”
- Testing only with internal staff or friends
- Writing tasks that give away the answer in their wording
- Running one big study instead of several small rounds
- Collecting findings but never scheduling the fixes
Simple Usability Metrics to Track
Watching sessions gives you the most insight, but a few simple measures help you compare rounds and show progress to stakeholders. You don’t need specialist software to record them.
| Metric | What It Shows | How to Record It |
|---|---|---|
| Task success rate | Whether people can complete the task at all | Mark each task as completed, completed with help, or failed |
| Time on task | How much effort the task takes | Note start and end times from the recording |
| Errors | Where people go the wrong way | Count wrong clicks, dead ends, and restarts |
| Ease rating | How difficult the task felt | Ask people to rate each task from 1 (very hard) to 7 (very easy) |
With only five participants, treat these numbers as signals rather than statistics. Their real value is in comparison. If task success improves and ease ratings rise after a redesign, you have clear evidence the changes worked.
Sharing Findings With Your Team
Findings only matter if they lead to changes. Keep your report short: the top problems, a short video clip of each one, and a recommended fix. Watching a real person struggle is far more persuasive than any written summary, so invite stakeholders to observe at least one live session whenever possible.
How Usability Testing Supports SEO and AI Search
Pages that are easy to use tend to perform better in search too. When people find what they need quickly, they stay longer and are less likely to bounce back to the results page. Usability testing also reveals the exact words people use to describe their problems, which is valuable for headings, page titles, and FAQ content that search engines and AI answer tools can match to real questions.
Questions and answers
Five people per round is a good starting point. Nielsen Norman Group research found that small groups uncover most of the common problems, and several small rounds teach you more than one large study.
A basic remote round can cost little more than participant incentives and a few hours of team time. Costs rise if you use a recruiting service, test with specialist audiences, or hire an external researcher.
Yes. Testing a clickable prototype is one of the cheapest ways to find problems, because changing a prototype takes minutes while changing built code takes days.
Yes. A small thank you such as a gift card shows respect for their time and makes recruiting much easier. The right amount depends on your audience and region.
In moderated testing, a facilitator runs each session live and can ask follow up questions. In unmoderated testing, participants complete tasks on their own while software records them. Moderated sessions give richer insight; unmoderated ones are faster to run at scale.
Little and often works best. A small round every few weeks during design and development catches problems while they are still cheap to fix.




