What is Prototype Testing?

Imagine after months of design and engineering, and a confident launch, your users stare blankly at the screen, unable to find the button that was supposed to be obvious.

Prototype testing lets you watch real people use your product while it is still made of sketches and clickable mockups, when changing direction costs hours instead of sprints.

What is prototype testing?

Prototype testing means presenting a prototype in front of users and observe how they interact with it. Participants attempt realistic tasks while researchers watch where they succeed, struggle, hesitate, or get confused. The goal is to evaluate usability, validate design decisions, and uncover problems early.

Prototype fidelities

Low-fidelity

Paper sketches, wireframes, basic clickable screens. They are fast and cheap, and attention is focused on structure and flow rather than visual polish.

Mid-fidelity

Digital wireframes with realistic layout and basic interactions.

High-fidelity

Polished, interactive mockups, or coded prototypes that look and feel close to the real product.

The prototype does not need to be complete. It only needs to be realistic enough for participants to complete the tasks you want to study.

Moderated vs. unmoderated

Moderated testing

A researcher is present, live, who guides the participant, asks follow-up questions, and probes unexpected behavior. It gives you rich insight into the "why", and works well for early, rough, or complex prototypes where things may need explaining. The trade-offs are higher cost, more scheduling, and fewer participants per round.

Unmoderated testing

Participants complete tasks on their own, usually through a platform that records their screen and answers. It is faster and cheaper and reaches more people. The downside is that you can’t ask follow-ups, and any prototype glitch or unclear task can derail a session with no one there to help. It suits polished prototypes and simple, well-defined tasks.

When to use it

Prototype testing is most valuable when the design is solid enough to react to but still cheap to change. It fits especially well when you are:

• Validating a new concept or feature before committing to development.
• Comparing design directions, such as two navigation structures or checkout flows.
• Iterating between design rounds, testing, fixing, and testing again.
• Checking a complex or unfamiliar flow, like onboarding or a multi-step form, where confusion is likely.
• Building stakeholder buy-in with evidence of how real users behave rather than opinions in a meeting.

How to prototype test

1. Define your goals

Decide what you want to learn, and write a clear question to ask users.

"Can users complete checkout without help?" is a better goal than "see if the design is good." Clear questions shape everything that follows.

2. Choose your prototype and structure

Make sure the prototype supports the full path of each task you plan to test, and match the fidelity to your questions.

Testing information architecture? Low-fidelity is fine. Testing visual hierarchy or micro-interactions? You will need something more polished.

Choose which structure of prototype testing you’d like to use that matches your circumstances. Will a live, moderated test work best, or a unmoderated test?

3. Recruit participants

Aim for participants who resemble your actual users.

Five to eight participants per round is typically enough to surface the major usability issues, especially if you plan to test again after making changes.

4. Craft realistic tasks + a testing script

Craft your questions as scenarios rather than as instructions.

Instead of "Click the Settings icon," try "You want to change how often you get notifications. Show me how you would do that."

For the script, prepare an introduction, the tasks, and follow-up questions, and keep your wording neutral so you do not lead participants.

If running an unmoderated session, decide the tool you want to use and set up your test appropriately (Tip: UserSage is a perfect tool for this!)

5. Conduct the session

When conducting moderated sessions, explain that you are testing the design, not the person, and encourage participants to think aloud. Then mostly stay quiet. Observe where they hesitate, backtrack, or misinterpret things, and resist the urge to help.

Ask open follow-ups like "What were you expecting to happen there?" Record sessions (with consent) and have a note-taker if possible.

6. Analyze results

Look for patterns across participants. Group issues by theme and rate them by severity, considering how many people hit the problem and how badly it blocked them.

Examples of observations to keep an eye out for:

• Task success rates
• Time on task
• Errors
• Recurring comments

Use these insights to iterate on your prototype, and retest to confirm the fixes worked!

Prototype testing with UserSage

1. Define your goals

Create a solid goal for what you want to learn through this test.

2. Set up your test

Choose your prototype and structure

UserSage is the perfect place to test your prototype! Once you decide what fidelity to test, you can connect your Figma prototype directly, or upload your designs manually.

Set your start and end screens

The start screen is what the participant sees first. End screens are any screen that counts as a successful finish.

Mark more than one if your prototype has multiple valid paths to success (like two different confirmation screens), so a participant landing on either one is scored as successful.

Craft realistic tasks

A clear, realistic task will set your test up for success. What are your users trying to do? This is shown to the participant as their instructions.

Good examples:

• "You're buying a birthday gift for a friend and want it to arrive by Friday. Find a pair of running socks under $20, and complete your purchase."
• “You're planning a weekend trip to Chicago next month and want a hotel with free cancellation near the city center. Find an option you'd feel comfortable booking."

Optional: set a step limit

You can set how many clicks a simulated person can try before the walk ends, so a lost person doesn't wander forever.

Recommended: add trailing questions

These are optional questions asked once someone finishes, e.g. how confident they felt, or what they'd change.

3. Recruit participants and run your test

Get a quick first read from simulated people with an AI Panel. Then run the same test with real participants to confirm it.

You can run your initial AI Panel scan with simulated people matched to the users you’re targeting.

5. Analyze results

Review important metrics like success rate, misclick rate, average steps and drop off rate right at the top.

Analyze path flows, heatmaps, per-person paths, and feedback!

Ready to get started?

Run your first prototype test with User Sage. Sign up for free and get a first read today.