User research and task analysis

Prerequisites: Basic software design.

Learning objectives

• Separate goals from proposed features.
• Analyze a task and its failure points.
• Identify limits of a research sample.

A feature request is evidence about a need

Human-computer interaction studies how people use computational systems in particular contexts. Start by identifying the task, the people performing it, and the environment. A student asking for a streak counter may want help maintaining a routine, reassurance about progress, or a way to compare with friends. Those needs imply different designs and different risks.

Research methods answer different questions. Interviews can reveal goals and explanations, observation can reveal actual workarounds, and usage data can show patterns among people who already use the system. None alone represents everyone who might be excluded, confused, or absent from the product.

Trace a real task

Consider “resume studying after a week away.” Break it into finding the course, recognizing the last completed work, recalling the relevant concept, choosing the next activity, and confirming that new progress was saved. A design that opens the next page instantly may still fail if the learner has forgotten a prerequisite. A short retrieval prompt could support the actual task better than another dashboard metric.

For each step, record required information, actions, possible errors, and recovery. Observe whether users recognize labels and state, rather than asking only whether they like the screen. A participant saying “looks clear” is weaker evidence than successfully finding and resuming the intended lesson without assistance.

Recruitment shapes the evidence

Recruit participants who reflect relevant variation in experience, devices, access needs, and context. A sample of enthusiastic developers may miss barriers faced by beginners or keyboard-only users. Small qualitative studies can reveal concrete usability failures; they do not estimate population prevalence precisely.

Use consent appropriate to the study, explain recording and retention, and avoid collecting unnecessary sensitive information. Ask neutral questions. “How would you find yesterday’s exercise?” reveals more than “Is our convenient progress panel helpful?” Separate what participants did, what they said, and your interpretation.

Turn findings into testable decisions

Write a finding with its evidence and scope: three observed participants could not distinguish completed lessons from saved drafts. Propose a change, such as separate labels and states, and define a follow-up task that could disconfirm its usefulness. Avoid turning one vivid quote into a universal requirement.

A research artifact should change an engineering decision. Task models can inform navigation, state transitions, error handling, and acceptance criteria. Revisit the model after implementation because real constraints and unexpected behavior often appear only when people use a working system.

Practice problem

Plan a short study of learners returning after a week. Write three neutral tasks or prompts, one observation to record, and one limitation of recruiting only existing daily users.

Worked solution

Ask participants to find the last exercise they attempted, explain what they believe is saved, and choose the next useful activity without instructions about which control to use. Record whether they identify the correct state and recover from a mistaken choice, along with points of hesitation. Existing daily users already know the interface and may have stronger motivation than people who stopped returning. Their success cannot establish that the design works for interrupted or first-time learners. Recruit some returning or lapsed learners and report the sample’s limits.

Check your understanding

Which prompt is least leading?

• How useful is our excellent progress panel?
• Wouldn’t a streak help you study?
• Show how you would resume your last exercise.
• Do you agree the navigation is clear?

Explanation

A task prompt lets observed behavior reveal whether the design supports the goal.

What can a small qualitative study establish well?

• Exact prevalence in the whole population.
• Specific observed failure mechanisms and plausible design problems.
• A guaranteed increase in retention.
• That every unobserved user has the same needs.

Explanation

It can identify concrete issues without supporting precise population estimates.

Primary reading

• Introduction to Human-Computer Interaction: University of Washington CSE 440
• Web accessibility tutorials: W3C Web Accessibility Initiative

Open the study and save your work →