A decision is a commitment among alternatives under constraints, evidence and uncertainty.
Choosing a lunch venue, approving a medical intervention, selecting a school pathway, allocating capital, publishing an article, shutting down a system and changing a regulation are all decisions, but they differ enormously in stakes, reversibility, authority, time horizon, uncertainty and downstream consequence.
Quick answer: how should decisions be categorised?
- Stakes: how much can be gained or lost?
- Reversibility: can the decision be undone?
- Time horizon: immediate, short-term, long-term?
- Recurrence: one-off or repeated?
- Authority: who is allowed to decide?
- Evidence: what information supports the choice?
- Uncertainty: how much remains unknown?
- Alternatives: how many viable options exist?
- Dependencies: what other decisions or systems constrain it?
- Consequence: who or what is affected afterward?
This article applies the general framework from How to Categorise Anything to decisions as structured acts of choice.
1. Define the decision unit
One strategic programme can contain many operational decisions. Do not classify a whole decision chain as though it were one choice.
2. Separate decision from outcome
A good decision can have a bad outcome under uncertainty, and a poor decision can sometimes get lucky.
3. Separate decision from recommendation
An adviser may recommend while another person or body holds final authority.
4. Stakes define consequence scale
Low-stakes decisions tolerate faster heuristics; high-stakes decisions justify stronger evidence, review and documentation.
5. Stakes are multidimensional
Financial, safety, legal, educational, reputational and human consequences should not be compressed blindly into one score.
6. Reversible decisions permit experimentation
When a decision can be cheaply undone, small tests and rapid iteration may be appropriate.
7. Irreversible decisions require stronger gates
Permanent deletion, major capital commitment or high-impact policy change warrants deeper evidence and review.
8. Reversibility can be partial
A decision may be technically reversible but still create sunk cost, delay, trust loss or path dependence.
9. Time horizon changes evaluation
A choice that improves this week can weaken next year. Classify immediate and long-term effects separately.
10. Urgent decisions differ from important decisions
Urgency concerns time available; importance concerns consequence. They should not be treated as synonyms.
11. One-off and recurring decisions differ
Repeated decisions can be standardised, automated or improved from feedback in ways one-off decisions cannot.
12. Policy decisions create rules for later decisions
A policy decision changes the decision environment itself by constraining future choices.
13. Strategic decisions allocate direction
They shape positioning, priorities, capabilities and long-term commitments.
14. Tactical decisions translate direction
They choose methods, sequencing and resource use within broader strategy.
15. Operational decisions govern immediate execution
Routing, scheduling, approval and exception handling are common operational decision types.
16. Authority is a core dimension
Who has the right to decide can matter as much as what option is chosen.
17. Delegated authority needs boundaries
Decision rights may depend on amount, risk level, geography or subject matter.
18. Collective decisions differ from individual decisions
Boards, committees, juries and teams may use voting, consensus, delegation or quorum rules.
19. Evidence-rich and evidence-poor decisions differ
Some choices rely on extensive data; others must be made under sparse or rapidly changing evidence.
20. Unknowns should be preserved
Decision records should distinguish missing evidence, contested evidence and genuinely unpredictable outcomes.
21. Risk and uncertainty are not identical
Risk may have estimable probabilities; uncertainty can include cases where probabilities themselves are unclear.
22. Alternatives define choice structure
Binary, ranked, multi-option and continuous allocation decisions require different methods.
23. “Do nothing” can be an alternative
Status quo should be represented explicitly rather than treated as absence of decision.
24. Constraints remove options
Law, budget, time, capacity, ethics and prior commitments can reduce the feasible choice set.
25. Dependencies create sequencing
Some choices can only be made after prerequisite decisions or evidence become available.
26. Path dependence raises switching cost
An early choice can reshape later options even when it was initially reversible.
27. Decision criteria should be explicit
Cost, quality, safety, speed, fairness and strategic fit are different evaluation dimensions.
28. Weighted scores are models, not truth
If criteria are combined numerically, weights and thresholds should be justified and tested.
29. Trade-offs should remain visible
A single score can hide that one option is faster while another is safer or fairer.
30. Decisions can be rule-based
Routine cases may use explicit thresholds and decision trees.
31. Decisions can be judgement-based
Novel or ambiguous cases may require expert interpretation rather than one fixed formula.
32. Decisions can be automated
Automation is suitable when rules are stable, evidence is machine-readable and error costs are acceptable.
33. Automated decisions need review gates
Low confidence, high impact, novelty and disputed inputs should trigger human review.
34. Decision quality is not outcome quality
Evaluate whether the decision used appropriate evidence, logic and process given what was knowable at the time.
35. Outcome feedback improves recurring decisions
Compare predicted and observed consequences to recalibrate criteria and thresholds.
36. Decisions need provenance
Record who decided, under which authority, from what evidence, under which rule or model version.
37. Reconsideration is a decision state
Pending, approved, rejected, deferred, escalated and under review describe lifecycle rather than decision type.
38. A practical decision record
- decision ID;
- decision type;
- stakes;
- reversibility;
- time horizon;
- recurrence;
- authority;
- alternatives;
- criteria;
- constraints;
- evidence;
- uncertainty;
- dependencies;
- decision;
- expected consequences;
- observed outcome;
- review status;
- version.
39. Categorisation should determine decision process
High-stakes irreversible decisions should receive stronger evidence and governance than low-stakes reversible ones.
40. The deeper idea
A decision category is useful when it tells us how carefully, how quickly, by whom and with what evidence a choice should be made.
The important distinction is not merely what was decided, but what kind of commitment the decision created and how difficult it will be to recover if it is wrong.
Final answer
Categorise decisions by stakes, reversibility, time horizon, recurrence, authority, evidence, uncertainty, alternatives, constraints, dependencies and consequences. Separate decision quality from outcome quality, preserve trade-offs, and use stronger review for choices that are consequential, novel or difficult to reverse.