Write a goal rather than an instruction list
A task prompt should describe what the person needs to achieve. If it specifies every button and menu, the exercise may reveal little about whether the interface supports independent use. Use the context required to make the task meaningful without disclosing the intended route.
The plan should also define what counts as a useful endpoint. Reaching a screen may not be enough if the person misunderstands what the screen means.
Make intervention deliberate
A researcher may need to clarify the scenario, resolve a prototype limitation, or help a session continue. Those interventions should be recorded because they affect how an outcome is interpreted. Completing a task with a hint is different from completing it independently.
Decide in advance what kinds of questions can be answered without teaching the product. A consistent approach helps the team compare observations and reduces the influence of an enthusiastic explanation.
Capture behavior and interpretation separately
Record actions, pauses, route changes, and the person’s explanation where it is relevant. A statement about confidence can be informative, but it is not the same as an observed successful decision. The two can disagree in ways that matter to the design.
A session should avoid collecting more personal detail than the study needs. Use invented or redacted task material when the question can be answered without real operational content.
Retain the context needed for analysis
An observation should identify the task, the material shown, the relevant user context, and any assistance provided. That record makes it possible to compare sessions without converting every difference into a broad conclusion.
The output can include a task guide and an observation structure. Recruitment, recording arrangements, and the specific session method belong in the agreed study scope. A website description does not imply that a participant panel or research facility is already in place.