For every $1 spent on UX, companies can earn back $100. That wild return is exactly why How to do User Testing is not optional anymore. Yet teams still ship products no one understands and call it bad luck. One quick test with real people can save months of wasted work and money. Learn to test early and your ideas stop failing quietly and start winning loudly.
TL;DR
- Don’t just “test.” Have a goal so you aren’t just watching strangers click on things.
- Five users are often enough to spot most issues. After that, you start seeing the same problems repeat.
- Do not guide them. Give them a task and let them figure it out.
- Keep your hands off. The moment you step in, the test stops being real.
- Collect the rough moments and turn them into a clear list your team cannot brush aside.
How to Do User Testing: The Process
The user testing process involves 5 dedicated phases:
- Phase 1: The Strategy
- Phase 2: The Participants
- Phase 3: The Script & Scenarios
- Phase 4: Facilitation & Moderation
- Phase 5: Synthesis & Action
Phase 1: The Strategy
Before you ask people to try your app, slow down and make a plan. Jumping into a test without a clear goal just creates noise. You end up collecting random feedback that does not help you move forward.
Setting Objectives: What are you Actually Hunting for?
You aren’t just testing the site. You are looking for friction. Are you trying to see if people can actually find the checkout button? Or are you wondering why they keep clicking on the non-clickable logo of a cat?
Start with the “Big Why.” Identify the three most important actions a user takes. If you are an e-commerce site, your objective is “Can a user buy a red shirt in under two minutes?” Everything else is just background noise.
Identifying Top Tasks (The 80/20 of user behavior)
Most users only use 20% of your features 80% of the time. Don’t waste time testing the terms and conditions page. Test the login flow, the search bar and the primary call-to-action. These are the lifeblood of your product.
Choosing your Metrics (Time on Task vs. Subjective Satisfaction)
Numbers are great but feelings tell the real story.
- Time on Task: How long did it take them to find the “Help” button? (If it’s over 30 seconds, your button is basically invisible).
- Subjective Satisfaction: Did they look like they wanted to throw their laptop out the window? That’s a low satisfaction score.
The Methods: Moderated, Remote or Guerrilla?
In 2026, you have options.
- Moderated: You sit there (virtually or in person) and watch. Best for complex stuff.
- Remote/Unmoderated: They do it at home while wearing pajamas. This is great for getting honest, “real-world” results.
- Guerrilla: You take your laptop to a coffee shop and bribe a stranger with a latte to try your app. It’s fast, cheap and surprisingly effective.
Phase 2: The Participants
If you test your app with your mom, she will tell you it’s “wonderful, dear.” This is useless. You need people who will tell you your baby is ugly professionally, of course.
Recruitment: Finding Your “Real” Users
We know that the right feedback from the wrong person is still wrong feedback. If you are building a tool for brain surgeons, don’t test it with a middle-schooler unless that kid is a prodigy (actually, even that isnt 100% the slamdunk you think it is).
The “Magic 5” Rule (and when to break it)
The legendary Jakob Nielsen discovered that you only need 5 users to uncover about 85% of your usability issues. Why? Because after the fifth person, you start seeing the same mistakes over and over. Unless you have a massive budget and a lot of free time, stop at five.
The Screener Trap: How to Spot Professional Liars
There are people out there who make a living doing user tests. They are professional testers. They will lie and say they own a private jet just to qualify for your $50 study.
- The Trap: Ask a dummy question. Do you use the software ‘XylophoneConnect’ for your daily accounting? (XylophoneConnect doesn’t exist). If they say yes, show them the door.
B2B vs. B2C Testing: The Hidden Nuances
Testing a game for kids? You need high energy and short sessions. Testing a tax software for accountants? You need a quiet room and a high tolerance for talk about spreadsheets.
Phase 3: The Script & Scenarios
The way you ask a question can ruin the entire test. It’s like a lawyer leading a witness.
Writing Scenarios that Don’t Lead the Witness
- Bad Task: Go to the menu, click on settings and change your password. You just told them exactly how to do it!
- Good Task: You think someone hacked your account. What do you do to make it safe again? Now you are testing if they can actually find the settings.
The Anatomy of a “Good Task” vs. a “Bad Task”
A good task is a story. It gives the user a “why.” Instead of saying search for a flight, say “You want to go to Paris for your anniversary next month. Find a flight that leaves on a Friday and costs less than $800.”
Why You Should Never Ask “Do You Like this Button?”
People are polite. They will say “Yes, it’s a very nice blue.” They are lying. Instead, watch their face. If they squint and lean into the screen, they don’t like the button or they can’t see it.
The Think Aloud Debate: To Prompt or Not to Prompt?
Getting users to talk through what they are thinking can be very helpful. Still, some people change their behavior when they explain things out loud. If they pause or struggle, do not rush to help. Give them time. Small reactions like silence, sighs or frustration often tell you more than their words.
Phase 4: Facilitation & Moderation
Being a moderator is like being a ghost. You should be seen (maybe) but never heard.
The Art of Being Invisible
Your job is to be a human tape recorder. If the user asks, “Where do I click?” you respond with, “What would you expect to do here?” It’s annoying for them but vital for you.
Handling the Silent User and the Over-Talker
- The Silent User: I see you’re looking at the top right of the screen. What are you thinking about right now?
- The Over-Talker: That’s very interesting. Let’s try to see if you can complete the next task while keeping that in mind.
When (and when NOT) to intervene
If the user has been stuck for five minutes and is starting to feel anxious, help them. If they are just mildly annoyed, let them suffer. The suffering is where the insights are.
You don’t need a $10,000 lab. Use Zoom or Google Meet to share screens. Record the session (with their permission). For a completely free asynchronous test, you can use the free tier of tools like Microsoft Clarity to watch session replays of real users on your site.
There are numerous other tools that can help you facilitate users and aid in your research.
Phase 5: Synthesis & Action
Congratulations! You have 5 hours of video of people being confused. Now what?
From Raw Video to Actionable Insight
Don’t write a 50-page report that nobody will read. Create a top 10 list of things to fix. If it’s not in the top 10, it doesn’t exist yet.
Watch the videos at 1.5x speed. Every time a user struggles, make a note. If three out of five users struggled with the same thing, that is a red flag issue.
Creating “Highlight Reels” for Stakeholders (The Power of Pain)
It is one thing to tell a developer, “The checkout is hard.” It is another thing to show them a video of a user yelling, “I hate this website!” at their screen. One minute of video is worth a thousand emails.
Using the PXL Matrix for Prioritization
Not all bugs are equal.
- Critical: They can’t finish the task. (Fix this yesterday).
- Major: They finished but it was a struggle. (Fix this next).
- Minor: They didn’t like the color of the icon. (Fix this when you’re bored).
The UX Debt Framework: Selling the Fix to the CEO
Fixing a problem after release can cost 100 times more than fixing it during the design phase. When you talk to your CEO, don’t talk about user delight. Talk about how fixing this button will save $50,000 in support calls next month. That is a language they speak fluently.
The “Cheating” Way: How Articos Automates Your Suffering
If you just read the last five phases and thought, “I have to find five strangers, write a script, watch hours of video and then fight my boss? I have a life,” then welcome to the future. Traditional user testing is like hand-washing your laundry in a river, it works but your back is going to hurt. Articos is the industrial-strength washing machine.
Instead of waiting weeks to find a “real human” who isn’t a professional liar, Articos uses AI-powered personas to simulate your audience. You give it your landing page or your brief and within 30 minutes, it runs the tests for you. It’s like having a team of 500 robot users who never get tired, never lie on screeners and don’t require you to buy them coffee.
Why Articos is the “Easy Button” for Beginners:
- Instant Gratification: Turn a six-week research cycle into a 30-minute coffee break.
- Zero Recruitment Drama: No more hunting for “brain surgeons who like gardening.” Articos already has the AI personas ready to go.
- No “Think Aloud” Awkwardness: The AI tells you exactly why a design fails without you having to sit in a dark room observing a stranger’s sighs.
Conclusion: This is How to Do User Testing Right
User testing isn’t about proving you’re right; it’s about finding out where you’re wrong before your customers do. 88% of users will never come back to your site after a single bad experience.
We believe that great design is just a series of guesses until a user touches it. So, go find five people, buy them some coffee and watch them break your product. It’s the most painful, expensive and rewarding thing you’ll ever do.
FAQs on How to Do User Testing
Five is the sweet spot. It helps you find about 85% of your biggest usability problems without wasting your entire budget on redundant feedback.
Don’t ask questions; give tasks. Instead of “Do you like the search bar?” say “Find a pair of blue socks for under $10.”
Not quite. User testing checks if people actually want the product (market fit), while usability testing checks if they can actually use it (functionality).
Use “Guerrilla Testing.” Take your laptop to a public place, offer to buy someone a coffee and ask them to try to complete one specific task on your site.
User testing is the “Big Picture” (Do they need this?), while usability testing is the “Nitty Gritty” (Can they click the button?).
