UX Design Principles: Usability Testing

Your website looked great at launch. Your designer approved it. Your team signed off on it. Six months later, your contact form submissions are flat and you are not sure why.

Usability testing a website answers that question. It replaces assumptions about how people use your site with direct observation of what actually happens when real visitors try to accomplish a goal. No survey, heatmap, or analytics report gives you the same information.

This is the final part of our UX Design Principles series. If you have not read the previous installment on organizing your site so visitors can find what they need, that foundation is worth your time first.

What usability testing actually is

Usability testing is a structured research method in which you ask real people to complete specific tasks on your site while you observe without helping them. The U.S. Department of Energy describes it as “a research activity designed to evaluate the usability of a website or application by observing representatives from your key audience.”

That last word matters: your audience. Not your coworkers. Not your developer. Not you.

Usability testing is not the same as A/B testing. A/B testing measures which of two options performs better across a large sample. Usability testing explains why people behave the way they do. You need both eventually, but usability testing comes first because it tells you what to fix before you run experiments.

When to run usability tests

You can run usability tests at any stage of a project. The three most useful moments are:

Before a redesign

Test your current site before you touch a single pixel. This gives you a baseline: the problems you are solving, ranked by how often and how severely they affect real visitors. Without it, redesigns often replace one set of problems with a different set.

After prototyping

Test on wireframes or a staging build before you write production code. Catching a broken user flow at the prototype stage takes an hour to fix. Catching it after launch takes a sprint.

After launch

A site that worked well for its original audience may develop friction as the business grows and its visitor mix changes. Quarterly or semi-annual testing sessions catch drift before it erodes conversion.

Moderated vs. unmoderated: which one do you need?

The distinction between these two methods comes down to what you are trying to learn.

Moderated testing uses a live facilitator, either in the room or on a video call. The facilitator can follow unexpected behavior, ask follow-up questions, and probe the reasoning behind hesitation. You learn the “why” behind what you see. This is the right choice for complex workflows, early-stage prototypes, and any situation where you suspect the problem is unclear navigation or confusing copy rather than a broken link.

Unmoderated testing lets participants complete tasks on their own while software records their screen and voice. There is no facilitator, which means lower cost per participant and faster turnaround. You learn the “what” at scale. This works well for validating a specific change or gathering data from a geographically distributed audience.

For most small to mid-size businesses running their first round of testing, start with moderated. Five sessions with a facilitator will teach you more than 20 unmoderated recordings you do not have time to fully review.

How to write tasks that reveal real problems

The single biggest mistake in usability testing is writing tasks that lead participants to the answer. A well-written task describes a realistic scenario without naming the feature being tested.

Weak task: “Use the navigation to find the Services page.”

Strong task: “You are considering hiring this company for a website redesign. Find out what services they offer and what the process looks like.”

The weak version tells the participant exactly where to look. The strong version mirrors how a real visitor would approach the site and exposes whether the navigation labels are clear enough to get them there.

After each task, useful follow-up questions include: what does that term mean to you? What did you expect to happen? Which of these felt easier? How difficult was that on a scale of one to five? Observe during the task. Ask afterward.

Running a test with no budget

You do not need a UX lab or a research agency to run a useful usability test. Guerrilla testing in a public space such as a coffee shop or library can produce valid qualitative data in 15-minute sessions with no lab required.

A practical zero-budget approach for a Houston business:

1. Recruit from your customer list. Email five to seven existing or recent customers. Offer a $20 gift card. Most people who have done business with you are willing to spend 20 minutes helping you improve.
2. Use a free screen recorder. Loom, OBS, and several video call platforms record screen and audio at no cost. You need the participant’s screen and their voice.
3. Write three to five tasks. Focus on the highest-stakes flows: finding contact information, understanding your services, requesting a quote or booking a call.
4. Stay quiet. The hardest part of moderated testing is not intervening when a participant struggles. Silence is the data.
5. Take structured notes. Log the task, the specific point of hesitation, and what the participant said or did. Time stamps help.

Research on participant counts confirms that five users surface roughly 85% of a site’s most common qualitative usability problems. Running three iterative rounds of five users each finds more issues than one single session with 15 participants, because you fix problems between rounds and expose new ones.

What to do with what you find

Findings from a usability test are not a to-do list. They are inputs to a prioritization decision.

After each session, log every observed problem with three pieces of information: what happened, at which step, and how many participants encountered it. Then score each issue by frequency (how many participants hit it) multiplied by severity (how much it blocked task completion). Issues that score high on both dimensions go to the top of your fix queue.

A few patterns to watch for:

• Participants who leave the page to search externally for information that should be on the page
• Repeated hovering over non-clickable elements (signals an expectation of interactivity)
• Task abandonment after a specific step (usually means that step’s copy or layout is the culprit)
• Participants who complete a task but describe feeling uncertain about the outcome

When reporting findings to a client or stakeholder, lead with the behavior observed, then the business implication, then the recommended fix. “Three of five participants could not find the contact form without scrolling to the footer” is a finding. “Consider moving a contact CTA above the fold on service pages” is the recommendation. Keeping them separate makes the report more persuasive.

When usability fixes are in place, conversion tracking closes the loop: it tells you whether the changes actually moved the numbers.

What this looks like in practice

The work shown in our Motion Giraffx project and Detail Exchange project followed this pattern: structured testing at the prototype stage, targeted changes before launch, and tracked conversion improvements afterward. Testing before launch is always less expensive than redesigning after.

Our web design process includes usability validation as a standard phase, not an optional add-on.

The takeaway

You do not need a big budget, a UX lab, or a professional research firm to run usability tests that produce actionable results. You need a realistic task list, five willing participants, and the discipline to watch without interrupting.

If your current site has pages that are not converting the way you expect, a half-day session with your own customers will tell you more than a month of analytics review.

Ready to talk through what a review of your site might look like? Get in touch and we will start there.

Frequently asked questions

How many users do you need for usability testing?

For qualitative testing, five participants typically surface the majority of a site’s most common usability problems. This benchmark traces to foundational research by Nielsen and Landauer. For quantitative studies measuring completion rates or error rates with statistical significance, you need 20 to 40 or more participants. If your site serves multiple distinct audience segments, plan on three to five participants per segment rather than five total.

What is the difference between moderated and unmoderated usability testing?

In moderated testing, a facilitator guides participants in real time, either in person or over video. This captures tone, hesitation, and the reasoning behind behavior. In unmoderated testing, participants complete tasks on their own while software records their screen and voice. Moderated sessions give you richer insight. Unmoderated sessions scale faster and cost less per participant. The research question determines which you need.

How do I write tasks for a usability test?

Write scenario-based tasks that describe a realistic situation without naming the feature you want to test. “You need a quote for a kitchen remodel. Use this site to find out how to contact the company” is a good task. “Click the Request a Quote button in the header” is not: it leads the participant directly to the answer. Avoid asking for opinions or preferences during task completion. Observe behavior first, ask questions after.

Can I run usability testing with no budget?

Yes. Five to seven participants recruited from your existing customer list, a free screen-recording tool, and a video call are enough for a productive qualitative session. Guerrilla testing in a public space such as a coffee shop or library can produce valid qualitative data in 15-minute sessions with no lab setup required. The goal is observation, not perfection.