A good outcome can follow a poor decision, and a poor outcome can follow a sound decision. Evaluate the process using what was knowable at the time.
1. Separate outcome quality from decision quality
| Good outcome | Bad outcome | |
|---|---|---|
| Sound process | Learn without becoming overconfident. | Examine uncertainty and execution without automatic blame. |
| Weak process | Do not let luck certify the method. | Fix the process; do not learn only from pain. |
This is not an excuse to ignore outcomes. Outcomes matter deeply; the point is to improve learning by examining both result and reasoning.
2. Build a decision record
| Field | Question |
|---|---|
| Decision | What choice is being made, by whom, and by when? |
| Objective | What outcome matters—and for which stakeholders? |
| Options | What are the realistic alternatives, including waiting or gathering evidence? |
| Evidence | What is known, unknown, and assumed? |
| Probability | What outcomes are plausible, and how likely do they seem? |
| Tradeoffs | Who benefits, contributes, carries cost, or bears risk? |
| Trigger | What new evidence would change the plan? |
| Review | When and how will the decision be evaluated? |
Keep a knowledge tracker
- Facts: supported observations or data.
- Interpretations: what the facts may mean.
- Assumptions: beliefs the plan depends on but has not established.
- Unknowns: information not yet available or not practically knowable.
- Updates: new evidence that changes confidence or action.
3. Defend against predictable distortions
| Distortion | Countermeasure |
|---|---|
| Resulting | Reconstruct what was knowable before the outcome. |
| Hindsight bias | Record expectations and probabilities prospectively. |
| Overconfidence | Use ranges, base rates, and reasons you could be wrong. |
| Premature closure | Name at least one plausible alternative explanation. |
| Social proof | Ask what evidence supports the consensus in this case. |
| Incentive blindness | Map how each stakeholder is rewarded, constrained, or exposed. |
| Action bias | Compare action with waiting, monitoring, or a reversible test. |
Invite independent views before group discussion, ask the most junior relevant voice early, distinguish facts from interpretations, and assign someone to articulate the strongest alternative.
4. Use counterfactuals to learn
- What other outcomes could reasonably have followed the same decision?
- What different process could have produced the same outcome?
- Which signal was diagnostic, and which only appears important in retrospect?
- What should become a reusable rule—and what is too context-specific?
- What would we document differently next time?
Decision-quality worksheet
| Prompt | Your notes |
|---|---|
| Decision, owner, and deadline | |
| Evidence and important unknowns | |
| Alternatives considered | |
| Expected outcomes and probabilities | |
| Stakeholder tradeoffs and change trigger | |
| Review date and reviewer |
Further reading: Annie Duke, How to Decide, for practical tools on resulting, hindsight bias, probabilities, decision journals, and counterfactuals. Scope: Educational only; not medical, legal, financial, or admissions advice. No outcome is promised.
Prefer to work offline?
Keep a printable copy.
The web page is the primary guide. The PDF contains the same core material in a format you can save, print, annotate, or carry into a conversation.
Download the printable guide