Early preview · Six skills are always free. Paid skills open soon.
Research a question · 2 min read

Turn a broad topic into a researchable question

Define the decision, constraints, comparison, and stopping rule before sending AI off to research.

Editorial illustration of a large night sky viewed through a small shaped telescope aperture.

A researchable question tells you what information would change a decision. Replace a broad topic with a specific choice, constraints, and a timeframe. Then define a stopping rule so research does not expand indefinitely into interesting but irrelevant facts.

Start with the decision

“Tell me about remote work” can lead almost anywhere. A narrower question is: “What should our three-person support team test before changing from two office days to fully remote work?” The narrower version identifies a team, a change, and a next action without assuming the change will succeed.

Define what is in and out

BoundaryFictional example
AudienceThree-person support team
DecisionWhether to run a four-week remote pilot
ConstraintsCoverage hours and escalation response
Evidence neededCurrent workflow, coordination risks, pilot measures
ExcludedBroad predictions about the future of work

A boundary is not a way to ignore inconvenient evidence. It is a way to make the question answerable at the scale of your decision.

Use a scope prompt

Working template

Help turn this topic into a decision-focused research question.
Ask what choice I need to make and what constraints matter.
Propose three narrower questions with different scopes.
For the selected question, define evidence needed, exclusions,
and a stopping rule tied to the decision.
Do not assume the desired answer.
Topic: [topic]
Decision context: [facts]

Decide when enough is enough

A stopping rule might be “We can design the pilot after documenting coverage requirements and identifying the three most plausible coordination failures.” It should not be “after finding ten articles,” because article count does not establish usefulness.

After the first research pass, ask which missing fact could reverse the decision. Prioritize that fact instead of collecting more background. If the question cannot be answered from public sources, shift to an interview, observation, or small experiment. Research can reveal the need for firsthand evidence; it does not have to end with a long report.