Decisions that keep slipping

What is decided and by whom

A choice cannot be made while it stays fog. First we formulate it: what exactly is being decided, who the deciding subject is — you, a partner, a family, a company — what the decision costs, what not deciding costs, and where a reversible trial is possible and where the choice is final. The last two questions do most of the work. An unmade decision is also a decision, except that its price accumulates silently: options expire, partners decide for you, the team reads the silence as an answer. And reversibility changes the required certainty: a step that can be undone needs enough analysis to start, not enough to be sure.

• The formulation of the choice, instead of the fog
• The subject of the decision: you, a partner, the family, the company
• The price of the decision — and of its absence
• Reversibility: where you can try, where the choice is final

Values into criteria

While criteria stay unnamed, every option looks equally arguable, and the last conversation wins. We unfold values into checkable criteria and weights: what may not be broken for any gain — the red lines; what is desirable and tradeable; and what you are ready to pay with — money, time, status, comfort. A criterion you cannot check is a wish, so each one gets a measure or a test. Comparison then runs along axes, not along the intensity of the argument: each option is scored against the same criteria, and the conflicts between criteria — safety against speed, income against autonomy — are negotiated openly instead of being settled by mood. Three scenarios close the section: best, working, worst — and the honest question whether the worst one is survivable.

• Hard limits (red lines) and soft preferences
• Criterion weights — and an honest refusal of "everything at once"
• Three scenarios: best, working, worst — and the survivability of the worst
• Pre-mortem: imagine the failure in advance and find its cause

Tools

The methodology of risk and the psychology of choice work in one direction. The tools are not used for the sake of formulas; they are used so that uncertainty becomes listable and discussable, and the method is matched to the decision — some need a calculation, some only clarity about what you want.

A decision tree lays the choice out as branches: options, the events that may follow each, their probabilities and their outcomes; the expected value at each node shows which branch carries the best average result. It is the right tool when the options and their consequences can be enumerated — an investment sequence, a hiring chain, a phased market entry.

A probability × impact matrix and a risk register do the same work for risks rather than options: each risk is scored, given an owner, early indicators and a prepared response — avoid, transfer, reduce, accept — so the decision includes its own protection.

Where probabilities are simply unknown, decision theory offers four classical criteria — four different, honest answers to the question “rational according to what?”:

Wald (maximin) scores each option by its worst outcome and chooses the option whose worst outcome is best. It is the criterion of survival: applied when the choice is irreversible and a bad outcome is not a loss but an ending — signing a personal guarantee, concentrating the whole business on one client.

Savage (minimax regret) first asks, for each scenario, what the best option would have been; regret is the gap between that and what your option gives. The chosen option is the one whose maximum regret across scenarios is smallest. It is applied when you will inevitably judge yourself against the road not taken — selling a business, leaving a partnership — and need the choice you can defend to yourself afterwards.

Hurwicz refuses both extremes and makes your risk posture explicit: each option is valued as α × its best outcome + (1 − α) × its worst, where α between 0 and 1 is your declared optimism. The value of the criterion is that the optimism stops being a mood and becomes a number you chose — and can be argued with.

Laplace proceeds from the position that unknown probabilities give no ground to favour one scenario over another: scenarios are treated as equally likely, and the option with the highest average outcome wins. It fits repeatable, symmetric choices, where the average over many similar decisions is what actually accumulates.

Two further instruments complete the set. A pre-mortem — “assume we have already failed; what happened?” — surfaces the risks politeness hides. And a pilot converts an irreversible-looking decision into a reversible first step with a check date: the decision is made, and its first stage is a test. Where no participant holds the whole picture, estimates are gathered by a structured expert round (the Delphi method) and revised as evidence arrives, instead of being frozen at the first meeting.

• Decision tree and expected value
• Probability × impact matrix, reserves and early indicators
• Criteria for choice under unknown probabilities: Wald, Savage, Hurwicz, Laplace
• A pilot and a reversible step instead of an all-or-nothing bet

Inside the choice: why a mature decision stalls

Behind a stalled decision usually sits not a lack of data but a conflict between ego states: reason argues one way, the inner Parent voices prohibitions — "don't risk", "what will people say" — the inner Child wants, or fears, and the Adult is left with last week's spreadsheet. The script decisions made early in life — how much I may earn, whom I may partner with, how visible I may be — quietly pre-select the options before any analysis starts; naming them does not cancel them, but it moves them from the steering wheel to the table, where they can be examined like any other input. And games stretch the choice for years: a decision is postponed because making it would end the familiar game — the grievance, the suspense, the bargaining — and the payoff of the game is real, which is why "just decide" never works.

• Sunk costs: "I have invested too much"
• Loss avoidance and postponement as a way to regulate anxiety
• Endless gathering of information; waiting for permission from an outside figure
• An impulsive decision as a way to end the tension

Execution and living with the result

The end of the work is not insight but movement. The decision is made and fixed in writing — the choice itself, the criteria it passed, and the red lines it respects; a decision that exists only in conversation dissolves at the first objection. The first step is scheduled as an artefact, not an intention: a letter, a term sheet, a meeting on the calendar, with a date. And a review date is set at which the choice is re-checked against the criteria and the new evidence — not against Monday's mood, and not against the opinion of the last person you spoke to.

• A bad decision ≠ a bad outcome
• Review point: when, and by what data
• Communicating the decision without excuses
• Recovery after a hard choice

Boundaries

Not investment, legal or medical advice. I don't make the decision for you.