List the unknowns, then classify them:
- must be answered before action;
- can be tested through a draft;
- depends on another person;
- can wait until a later decision.
Choose the cheapest action that reduces the most important uncertainty: request data, create a sample, test one case, or write an assumption explicitly.
Do not research every branch equally. Update the plan when evidence arrives. Progress in uncertain work is often a better question, not a larger amount of output.
Related reading
- Continue with clarifying an unclear task.
- Continue with finding the first workable step.