Skip to main content
UX Design Engineering
Settings

Settings

Theme
Text size
100%
Typeface

Chapter 2

Investigating the problem behind the request

Observe a workflow, separate evidence from interpretation, and identify the questions a feature request leaves unanswered.

“Assignment takes too long” sounds specific until you watch someone do it. An administrator reads a case, checks its status, opens another screen for eligibility information, returns to the list, and chooses a specialist. The final submission occupies only part of the task.

Your decision is what to investigate before committing to bulk assignment. Start by locating uncertainty in the workflow.

Turn the request into questions

Keep the original request visible: “Allow administrators to assign several cases together.” Record the assumptions it contains. Cases can be grouped meaningfully. Administrators can identify eligible work before selecting it. Repeating the submission is a significant source of delay. A combined action would remain understandable when only some assignments succeed.

Each assumption suggests something to observe or test. Ask how administrators identify eligible cases, where they leave the workflow, and what happens after a rejected assignment. A question such as “Would bulk assignment save you time?” invites an opinion about the proposed solution before you understand the task.

Gather a small amount of useful evidence

Work with a researcher where one is available. Ask for access to people who do the task, describe what decision the study supports, and agree on what you can record. If live customer records cannot be shown, use approved sanitized examples while acknowledging that they may omit real complexity.

Ask participants to walk through a recent case or complete a realistic task. Listen to their explanation, but also record actions and interruptions. When someone leaves the screen, ask what information they need. When they abandon a case, ask what made them stop.

Inspect support issues and existing product signals as additional perspectives. A support ticket can reveal a failure, but it does not tell you how common that failure is. A click count can show repeated actions, but it rarely explains their purpose by itself.

Separate what you saw from what you think

The following material is supplied fictional evidence for the Casework exercise. No participant sessions occurred.

Supplied observationPossible interpretationNext question
An administrator opens a case detail page before choosing an assigneeThe list may omit eligibility informationWhich detail changes the decision?
An assignment is rejected after the selector was enabledPermission or status may have changedWhat changed, and how could the user recover?
Another administrator assigns several straightforward cases individuallyRepeated submission may be costly for some workHow often does this occur, and can those cases be grouped safely?

All three observations deserve attention. The first two do not invalidate the third. They suggest that a combined action could still be valuable after eligibility and recovery are understood.

Write observations as concrete actions with context. “The user was confused” is an interpretation. “They reopened the record twice and asked whether the assignment had saved” gives a reviewer something to examine.

Map the task and its dependencies

For the worked case, the task is: find a case, establish eligibility, choose a specialist, submit, confirm the outcome, and continue. Beside each step, record the information required and the system or person supplying it.

This reveals a dependency early: if eligibility depends on service rules the product team cannot inspect, a more informative interface requires help from the service owner. That is a product constraint as well as an engineering concern.

Do not wait until implementation to discover who can explain the rules. Ask for the smallest useful conversation: “Which conditions can make this action unavailable, and which can change after the screen loads?”

Handle missing access honestly

Large organizations may require scheduling, consent review, or approval before you can speak with users. Request the access through the appropriate channel and continue independent work: inspect existing studies, review service behavior, and list competing explanations.

Colleagues can help you find problems with a prototype, but a colleague unfamiliar with the real task is not a substitute for an experienced administrator. Label that evaluation accurately. If you cannot reach users in this exercise, analyze the supplied observations and leave a real-user study as a proposed check.

Write a problem brief

The worked brief reads:

Administrators need to determine whether a case can be assigned and understand the outcome of an attempt. Supplied observations suggest that missing eligibility information and unclear recovery interrupt the task. Repeated submission may create additional effort for straightforward work. We will first test whether clearer eligibility information improves decisions, while investigating the frequency and safety of batch assignment.

This brief preserves uncertainty. It states what the team needs to learn before choosing a larger implementation.

Practice: investigate a competing explanation

Create a task map and a short evidence table for your workflow. Write at least two explanations for the reported difficulty and identify an observation that would distinguish them. If conducting a real session, agree on an appropriate task and recording practices with your research partner.

Write a brief separating the requested feature, the user problem, supporting evidence, missing evidence, and the next decision. Include a successful case so your account does not describe only failures.

Review your work

A peer should be able to identify which statements are observations, which are interpretations, and which are assumptions. They should understand whose experience is represented and whose may be missing. Your next study should answer a decision-relevant question instead of asking people to approve the proposed feature.

The next chapter turns the brief into success criteria and a scoped recommendation.

Found an error? Report it.