Most websites are signed off in a meeting room by people who already know where everything is. That is the core problem usability testing solves: it replaces internal opinion with recorded evidence of real people struggling (or succeeding) on your pages.
This guide is written from the perspective of a small agency budget. No research lab, no five-figure research platform, no full-time UX researcher. Just a moderator, a screen recorder, five participants and a set of well written tasks. By the end you will have everything you need to run your first usability testing web design session this week.
What Is Usability Testing in Web Design?
Usability testing is a research method where real users attempt real tasks on your website while you observe, record and note where they hesitate, misread or fail. You are not asking people whether they like the design. You are watching whether they can use it.
The distinction matters more than anything else in this article:
- Opinion research asks: “What do you think of this homepage?” Answers are polite and mostly useless.
- Usability testing asks: “You need a quote for a 40 square metre office renovation. Show me how you would get one.” Answers are behaviour you can act on.
A usability test can be run on a live site, a staging environment, a clickable prototype, or even a competitor’s website for benchmarking. The method stays the same.
Qualitative vs Quantitative Usability Testing
| Aspect | Qualitative (what you should start with) | Quantitative |
|---|---|---|
| Goal | Find and explain problems | Measure performance (success rate, time on task) |
| Participants | 5 to 8 per audience group | 20 or more |
| Output | Clips, quotes, prioritised issue list | Statistics and benchmarks |
| Typical cost | Very low, often under 300 EUR | High, needs tooling and recruitment volume |
| Best for | Redesigns, new features, broken funnels | Proving improvement after a fix |
For 95% of agency projects, qualitative testing with a handful of users is the right call. The Complete Guide to Website Usability Testing is a useful companion to this.
Why Five Users Is Usually Enough
The classic finding from Jakob Nielsen’s research is that around five participants expose roughly 85% of the usability problems in an interface. The curve flattens fast because the same obstacles trip up person after person.
| Participants | Approximate share of problems found | What it feels like in practice |
|---|---|---|
| 1 | ~30% | Shocking, but you cannot tell luck from pattern |
| 3 | ~65% | The big blockers start repeating |
| 5 | ~85% | Clear patterns, enough evidence to convince a client |
| 10 | ~95% | Diminishing returns, mostly minor issues |
Important nuance: five users covers one audience group. If your client’s site serves two clearly different audiences (for example B2B buyers and job applicants), test three to four people per group instead of five for one group only.
Two rounds of five users beat one round of ten. Fix the obvious problems after round one, then test again to see whether the fix worked and what it exposed underneath.
The Five Main Types of Usability Testing
| Type | How it works | Budget fit |
|---|---|---|
| Moderated remote | Video call, participant shares screen, you ask follow-ups | Best value, recommended starting point |
| Unmoderated remote | Participant records themselves following written tasks | Cheap and fast, but no follow-up questions |
| In-person moderated | Same room, you watch body language and hesitation | Great for physical retail or local clients |
| Guerrilla testing | Approach people in a cafe or coworking space with a laptop | Nearly free, good for simple consumer sites |
| Comparative testing | Same tasks on your site and a competitor’s | Powerful in pitches and redesign proposals |
How to Run Your First Usability Test in 8 Steps
Step 1: Decide What You Actually Need to Learn
“Test the website” is not a research goal. Pick two or three questions tied to business outcomes:
- Can a first-time visitor understand what this company sells within 10 seconds?
- Can a prospect find pricing and request a quote without using search?
- Can a returning customer complete checkout on a phone with one hand?
Ask the client where they lose money. Analytics drop-offs, support tickets and sales objections are the best shortlist generators.
Step 2: Write Task Scenarios That Do Not Give Away the Answer
This is where most first tests fail. A task scenario should give context and a goal, never a route or interface vocabulary. The piece What is Usability Testing makes a good next read.
| Weak task (leads the user) | Strong task (realistic scenario) |
|---|---|
| “Click Solutions, then Enterprise, and find the pricing table.” | “Your company has 60 employees. Find out how much this product would cost you per month.” |
| “Use the filter to find a red jacket in size M.” | “You need a waterproof jacket for a hiking trip next month, under 150 EUR. Find one you would consider buying.” |
| “Is the contact form easy to use?” | “You want someone to call you back tomorrow morning about a broken boiler. Do whatever you would normally do.” |
Rules for writing good tasks
- Keep each task to one or two sentences read aloud.
- Never use words that appear as button or menu labels on the site.
- Give the user a motivation, not an instruction.
- Make the success criteria measurable for yourself, not for them.
- Plan 4 to 6 tasks for a 45 minute session. Less is more.
- Start with an easy warm-up task so the participant relaxes.
Write down, before the session, what “success” means for each task. Example: Participant reaches the quote form and submits it in under 3 minutes without using site search.
Step 3: Recruit Five Participants Without a Recruitment Agency
Professional recruitment panels are excellent and expensive. On a small budget, these sources work well:
- The client’s own customer list. Ask for 20 recent customers or leads, send a short invite with a clear time commitment and incentive.
- A banner or exit intent prompt on the live site offering a voucher for a 30 minute call. Best source of genuinely representative users.
- LinkedIn or niche communities for B2B audiences, Facebook local groups for consumer services.
- Friends of friends who match the profile, never your own colleagues or anyone who has seen the design.
- Guerrilla recruiting in a coworking space or cafe for broad consumer sites.
A three-question screener is usually enough
- Have you bought or researched [category] in the last 12 months?
- Do you work in web design, marketing or UX? (If yes, exclude.)
- Which device would you normally use for this kind of task?
Incentives: a 25 to 50 EUR voucher for consumers, 75 to 100 EUR or a charity donation for professionals. Always recruit six people for five slots. One no-show is normal.
Step 4: Assemble a Zero-Cost Testing Toolkit
| Need | Low or no-cost option |
|---|---|
| Video call with screen share and recording | Google Meet, Zoom free tier, Microsoft Teams, Jitsi |
| Local screen and webcam capture | OBS Studio, QuickTime, Windows Game Bar, Loom free plan |
| Mobile session recording | Built-in iOS and Android screen recording, phone on a cheap desk stand |
| Note taking | A spreadsheet with one row per task and one column per participant |
| Transcription | Built-in meeting captions, or Whisper-based free tools |
| Prototype to test | Figma free plan, or the staging site itself |
| Supporting behaviour data | Free tiers of heatmap and session replay tools, GA4 funnels |
Do not buy a research platform for your first test. Dedicated tools like Maze, UXtweak or Userfeel are genuinely useful once you run tests monthly and need panels, heatmaps and automatic metrics. For test number one, a call link and a recorder are enough.
Step 5: Prepare a Simple Moderator Script
Reading from a script keeps every session comparable and stops you from unconsciously coaching people. A workable structure:
- Welcome and consent (3 min): explain the purpose, confirm recording permission, state clearly that you are testing the website and not them.
- Warm-up questions (5 min): what they do, how they normally research this kind of purchase, what device they use.
- First impression (2 min): open the homepage. “Without clicking anything, tell me what this company does and who it is for.”
- Tasks (25 to 30 min): one at a time, read aloud and paste into chat.
- Wrap-up (5 min): what was most frustrating, what was missing, would they contact this company.
Phrases every moderator needs
- “Please think out loud as you go. Tell me what you are looking at and what you expect to happen.”
- “What are you thinking right now?”
- “What did you expect to happen when you clicked that?”
- “If I were not here, what would you do next?”
- “There is no wrong answer, take your time.”
The hardest skill is silence. When a participant is stuck, count to ten before saying anything. That awkward pause is where the most valuable findings live.
Step 6: Record the Sessions Properly
- Ask for screen share of the full screen, not one tab, so you catch tab switching and search behaviour.
- Keep the participant webcam on if they agree. Facial reaction at the moment of confusion is your best clip for client presentations.
- Start recording after consent is given on camera and say the date and participant number aloud.
- Name files consistently: clientname-round1-P03-2026-09-18.mp4.
- Have a second person take notes if possible, so the moderator can concentrate on the conversation.
- Store recordings securely, delete them after the agreed retention period, and mention this in your consent statement.
Step 7: Turn Observations Into a Prioritised Issue List
Within 24 hours of the last session, list every observed problem, then score it. A simple severity scale keeps clients focused on what matters:
| Severity | Definition | Action |
|---|---|---|
| 4 – Blocker | User cannot complete the task at all | Fix before launch, no discussion |
| 3 – Major | Task completed but with significant delay or frustration | Fix in the current sprint |
| 2 – Minor | Brief hesitation, user recovers quickly | Backlog, bundle with related work |
| 1 – Cosmetic | Noticed but no impact on the outcome | Fix if cheap |
For each issue record: what happened, how many of the five participants hit it, severity, a 15 second video clip, and a recommendation. The clip is what ends internal debates.
Step 8: Report, Fix, Retest
Keep the deliverable short. A one-page summary plus a prioritised table beats a 40-slide deck nobody reads. Structure it as:
- What we tested, with whom, and when
- The three findings that cost the most money
- Full issue table with severity and frequency
- Recommended changes, split into quick wins and structural work
- Link to a highlight reel of 4 to 6 clips
Then book round two. Retesting the same tasks after the fixes is what proves the value of the work and keeps usability testing in the budget next time.
Real Problems These Tests Uncover on Client Sites
These are the patterns that appear again and again on small and mid-size business websites. None of them show up in analytics with an explanation attached.
1. Navigation labels invented by the marketing team
A B2B services site used “Capabilities”, “Sectors” and “Insights”. Four of five participants looking for pricing or a service list clicked “Sectors”, bounced back, then used the search box. Renaming to plain service names cut the path to the quote form from five clicks to two.
2. The hero section that says nothing
“Empowering growth through intelligent solutions” over a stock photo. Asked what the company sells, three of five participants guessed wrong, and one guessed “a bank”. The company installs commercial kitchen equipment.
3. Prices hidden behind a contact form
Participants routinely open a second tab and check competitors instead of filling in a form for a price. Watching someone silently abandon a site to compare elsewhere is far more persuasive to a client than any best-practice argument.
4. Forms that reject valid input
Phone fields rejecting spaces, postcode fields rejecting lowercase letters, error messages displayed above the fold while the user is scrolled to the bottom. One participant tried four times, said “this is broken”, and closed the tab.
5. Mobile CTAs below the fold and under a cookie banner
On a phone, the primary call to action often sits behind a sticky cookie bar, a chat bubble and a newsletter popup. Test on a real phone, not a resized browser window, and you see this in the first minute.
6. Filters that reload the page and lose scroll position
On ecommerce and property sites, users apply a filter, get scrolled back to the top, lose their place, and give up after the second attempt.
7. “Contact us” pages with no phone number
For local and urgent services, participants look for a phone number first. A form-only contact page routinely sends them to Google to find the number, where competitors are waiting.
8. Trust gaps at the decision moment
Participants pause before submitting and start hunting for something reassuring: a real address, named team members, review counts, delivery and returns terms. When they cannot find it quickly, they hesitate and often stop.
What a Five-User Test Actually Costs
| Item | Low-budget approach | Typical cost |
|---|---|---|
| Recruitment | Client list plus on-site prompt | 0 EUR |
| Incentives | 5 x 40 EUR voucher | 200 EUR |
| Software | Meet or Zoom plus OBS | 0 EUR |
| Preparation | Goals, tasks, script, screener | 3 to 4 hours |
| Sessions | 5 x 45 min plus buffer | 5 hours |
| Analysis and report | Issue table plus clip reel | 4 to 6 hours |
Total: roughly two working days and 200 EUR. Compare that with the cost of rebuilding a checkout that nobody tested.
Eight Mistakes That Ruin a First Usability Test
- Testing with colleagues or the client’s staff. They already know the site structure.
- Helping too early. Every rescue destroys a finding.
- Asking “Do you like it?” Preference is not usability.
- Cramming 12 tasks into 30 minutes. Rushed users behave nothing like real ones.
- Only testing desktop when most traffic is mobile.
- Skipping the pilot session. Run one practice session with a colleague to catch broken links, unclear wording and timing problems.
- Treating one person’s opinion as a pattern. Count how many of the five hit each issue before recommending anything.
- Delivering a report with no clips. Video evidence is what gets budget approved.
Your Pre-Session Checklist
- Research goals written down and agreed with the client
- 4 to 6 task scenarios with defined success criteria
- Moderator script printed or on a second screen
- Screener used, six participants booked for five slots
- Consent wording ready, recording permission requested on camera
- Test environment stable, with test accounts and dummy payment data if needed
- Pilot session completed and tasks adjusted
- Note-taking sheet prepared, one row per task
- Recording software tested on the exact setup you will use
- Calendar invites with clear joining instructions and a reminder 24 hours before
FAQ: Usability Testing in Web Design
What are the five types of usability testing?
The five most commonly used types are moderated remote testing, unmoderated remote testing, in-person moderated testing, guerrilla testing and comparative testing. Teams also use related methods such as A/B testing, tree testing and first-click testing to answer narrower questions.
What is usability testing in UX design?
In UX design, usability testing is the observation of real users performing realistic tasks with a design, in order to identify where the interface fails to match their expectations. It is evaluative research, used to validate or challenge design decisions before and after launch.
What is the difference between UX testing and UAT?
UAT (User Acceptance Testing) checks whether the build matches the agreed requirements and functions correctly. Usability testing checks whether real people can actually use it to achieve their goals. A feature can pass UAT perfectly and still fail usability testing because the label is confusing or the flow is illogical.
How many users do I need for a usability test?
Five per audience group is the standard recommendation for qualitative testing, uncovering roughly 85% of usability problems. If you need statistics such as task success rates, you need 20 or more participants.
Can I run usability testing for free?
Yes. Using a free video call tool, a free screen recorder and participants recruited from the client’s own customer base, the only real cost is your time plus optional incentives. Free tiers of dedicated usability testing platforms also exist for small unmoderated studies.
How much do website testers get paid?
Participants in unmoderated remote tests are typically paid around 5 to 15 EUR per short test on public panels. Recruited participants in moderated 45 minute sessions usually receive 25 to 100 EUR depending on how specialised the profile is.
How long should a usability testing session last?
Between 30 and 60 minutes. Beyond an hour, fatigue changes behaviour and the quality of think-aloud commentary drops sharply.
When should usability testing happen in a web design project?
Ideally three times: on the current site before a redesign to find what is broken, on the prototype before development to validate the new structure, and after launch to confirm the fixes worked.
Start With One Test
You do not need a research department to make usability testing part of your process. You need clear goals, honest task scenarios, five people who match the real audience, a recorder and the discipline to stay quiet while someone struggles.
Pick your least favourite client page, write four tasks, book five sessions and watch what happens. The first recording usually changes how the whole team talks about the design.
Want help planning your first round of usability testing? The team at awesome-screens.com runs lean, evidence-based testing for small and mid-size websites. Get in touch and we will help you decide what to test first.
