Choose what you can learn.
Name what you cannot.

A method is useful when its evidence matches the question and its limits remain visible. These are three possible approaches to the same fictional access-information task.

Moderated task walkthrough

A moderated conversation

What it can contribute: Seeing the route and asking a neutral follow-up can distinguish an action from the reason the person gives for it. The moderator may still influence behavior.
Inputs and collection: Record route choices and the participant’s words in separate notes. Ask “What are you looking for?” only when necessary; do not tell them where to click.
Who is included: Recruit intended first-time visitors whose relevant planning needs fit the research question. Agree recruitment criteria without collecting unnecessary sensitive details.
What it cannot establish: This is a small qualitative inquiry plan, not a representative estimate of how often a problem occurs.

Use these conditions in the protocol

Task response with a route account

An unmoderated written response

What it can contribute: A written response can show the answer and the route a person reports. It cannot reliably reveal every action or an unspoken hesitation.
Inputs and collection: Provide the task without hints, ask for the answer and the page used, then ask what remained uncertain. Keep reported actions distinct from directly observed actions.
Who is included: Recruit intended visitors able to use the supplied format. Offer an accessible response method and account for people excluded by a written-only approach.
What it cannot establish: Self-reported routes and explanations may omit steps. Absence of a reported difficulty is not evidence that no difficulty occurred.

Use these conditions in the protocol

Content and task review

No participant access yet

What it can contribute: A desk review can identify missing information and assumptions to test. It cannot stand in for evidence of what visitors actually do.
Inputs and collection: Trace the task against the supplied navigation and content. Record the exact page or wording supporting each observation, then list the questions that require visitor evidence.
Who is included: No participants in this stage. Define who should be involved later and what evidence their involvement would add.
What it cannot establish: The output is an expert review and a set of hypotheses, not a usability finding or a claim about visitor behavior.

Use these conditions in the protocol

Do not combine unlike evidence silently.

A recorded action, a person’s explanation and a reviewer’s judgment can all inform a study. Label them distinctly. If one source changes your interpretation of another, explain that reasoning instead of turning the combination into an unsupported certainty.

Return to the protocol

Choose what you can learn.Name what you cannot.

Do not combine unlike evidence silently.

Choose what you can learn.
Name what you cannot.