Research, evaluation and learning for better experiences.UX audits · Research · Training
Research practice

A better usability study starts with a better task

Small changes in task wording can change what you learn. Here is how to give participants a goal without giving away the route.

Start with the decision

Before writing a task, name the decision the study will inform. Are you deciding whether people can find a service, understand a price, or recover from an error? A task should make that uncertainty observable. Asking someone to explore a homepage can produce interesting comments, but it may not resolve the team's actual question.

Describe the situation, not the interface

Compare ‘Click Appointments and select Reschedule’ with ‘You can no longer attend your visit on Tuesday. Find a way to move it to another day.’ The first instruction tests whether a person can follow your directions. The second gives them a goal and leaves the route open. Avoid borrowing exact labels from the interface unless those words would naturally be part of the situation.

Make success observable

Define what completion means before the session. Is selecting a new date enough, or must the participant confirm it and understand what happens next? Record what they do, where they hesitate and whether they recognise completion. A confident answer is not automatically a successful action. If participants need information the scenario does not provide, revise the task rather than interpreting that gap as a product failure.

Pilot before recruiting widely

Run the guide with a colleague or pilot participant to catch confusing wording, broken prototypes and unrealistic assumptions. This checks the study setup; it does not replace sessions with intended users. Keep a short record of changes so that later findings can be interpreted in the context of the tasks participants actually received.

Keep the facilitator neutral

Let the participant try before intervening. When you need to probe, ask ‘What are you thinking?’ or ‘What would you expect to happen?’ Record assistance as assistance. Your goal is to understand the interaction, not to help the product pass the test.

EVIDENCE IN VIEWIllustrative data

See where the journey breaks down.

Task completion by stage (%)

Revised designInitial design
Illustrative comparison using invented values. Separate task completion at each stage, not a conversion funnel or client result.
View example data
Task / skillInitial designRevised design
Find90%94%
Choose78%90%
Enter58%84%
Review50%82%
Finish44%80%

Develop this skill with us

Practical usability testing connects these ideas with guided exercises and feedback.