Home / Guides / Decision-Making Under Uncertainty
Guide

Decision-Making Under Uncertainty

Practical ways to make and document decisions when evidence is incomplete, outcomes vary, and delay also has consequences.

By Adrian M. FenwickReviewed August 3, 2026

Uncertainty cannot always be removed before a decision is required. A good process makes objectives, options, assumptions, evidence, trade-offs, authority, and review conditions visible.

Frame the decision

State the decision, objective, deadline, constraints, stakeholders, and what happens if no decision is made. Separate facts, estimates, assumptions, and preferences.

Compare options consistently

Evaluate benefits, risks, dependencies, timing, reversibility, resources, and consequences under several scenarios. Include the status quo and the cost of delay where relevant.

Preserve options

Phased commitments, pilots, checkpoints, modular design, contingency clauses, and staged investment can reduce irreversible exposure while information improves.

Use ranges and confidence

Where values are uncertain, show plausible ranges and the assumptions that drive them. Avoid selecting only the most convenient point estimate.

Record review conditions

  • What evidence would change the decision?
  • Which assumptions must be monitored?
  • When will the decision be reviewed?
  • Who owns implementation?
  • What threshold triggers pause, escalation, or reversal?
Use with judgmentRisk methods support decisions; they do not remove uncertainty. Record assumptions, limits, and acceptance authority.

Choose methods that fit the decision

Not every decision needs a formal risk register. A short assumptions check may be enough for a reversible, low-consequence choice. Larger or less reversible decisions may benefit from scenarios, ranges, sensitivity analysis, pre-mortems, expert challenge, staged commitments, or explicit decision gates.

Preserve options where possible

When information is incomplete, a useful response may be to delay an irreversible commitment, pilot on a smaller scale, create an exit condition, obtain additional evidence, or build capacity that works across several scenarios. The cost of waiting should be considered alongside the value of learning.

Decision record prompts

  • What objective and decision are being considered?
  • Which assumptions matter most?
  • What could change the preferred option?
  • Which effects are reversible and which are not?
  • Who accepts the remaining uncertainty and on what basis?