Decision skills fields compared: statement, constraints, criteria and reversibility

Most decisions are never made. They are arrived at, by a process nobody chose, and then explained afterwards. Decision skill is the ability to make that process visible while it is still running: to know what the decision is, what would settle it, and whether the moment to settle it has arrived.

The tool for that is a record. Not a long one. Eight fields, written before the decision rather than after, each of which has a near neighbor it is constantly mistaken for. The rest of this page is those eight fields and their neighbors.

What to take away

Field 1: the decision statement

A decision commits to one course of action and excludes others; a choice is the set of available courses; a judgment assesses how things are.

The three nest: judgment informs choice, and choice is resolved by decision.

Trouble starts when a meeting produces a judgment, such as "the supplier is unreliable," and everyone leaves believing a decision was made.

The statement should say what will be done and what will not, in a form that a stranger could check later. "Improve supplier reliability" is a hope. "Move the second production line to supplier B from the start of the next quarter, and keep the first line where it is" is a decision.

Field 2: the objective

An objective is what you are trying to achieve. A constraint is a

Confusing them produces two errors. Treating a constraint as an objective ("we want to spend as little as possible") turns a limit into an aim and pushes every decision toward the cheapest option. Treating an objective as a constraint ("it must improve reliability") removes the ability to trade it against anything.

Write one objective. If there are two, decide now which gives way if they conflict, because they will.

Field 3: the constraints

A hard constraint cannot be violated: a legal requirement, a physical limit, money you do not have. A soft constraint is a strong preference presented as a limit. Most constraints offered in a discussion are soft.

To find out, ask what happens if it is breached. If the answer is that someone will be unhappy, not that it cannot be done, it is a preference. It belongs in the criteria.

An assumption is something being treated as true that has not been checked. It is not a constraint, but it behaves like one until it is written down, at which point it usually turns out to be checkable in an afternoon.

Field 4: the alternatives

An alternative is a genuinely different course of action. An option is usually a variant of the same course. Three suppliers is one alternative (outsource) with three options. Making it yourself, outsourcing it, and not doing it at all are three alternatives.

Two of them are missing from almost every decision record.

The habit of writing the choice out with all its alternatives is worth practicing on decisions that are already made. The entries on corporate jargon supply the other half of it: a decision stated with no actor in it cannot be assigned to anyone.

Field 5: the criteria

A criterion is a property of an alternative you use to compare them, and a preference is which direction on that criterion you like. A weight is how much that criterion matters relative to others.

Most decision arguments are disagreements about weights disguised as disagreements about facts. Both people agree supplier B is cheaper and slower, but they disagree about whether cheap matters more than fast.

Separating the three makes that visible. It also exposes the tactic of adding criteria until the preferred option wins, which is easier to spot once everyone has agreed the list in advance. The pull toward evidence that supports the option you already like has a name, confirmation bias, and agreeing the criteria first is the cheapest available protection against it.

Field 6: the evidence

Data is a recorded observation. Information is data arranged to answer a question. Evidence is information that bears on this decision. A great deal of data is not information, and a great deal of information is not evidence, and the second gap is where decisions drown: a report of everything known about the topic, none of which changes which alternative is best.

The working test is a sentence: "If this were true, I would choose X; if false, Y." Anything that cannot be fitted into that sentence is background. Anything that can is evidence, and it is worth finding out.

A fact has been checked. A claim has been asserted. In a meeting they sound identical. Ask which.

Field 7: uncertainty

Risk is a spread of possible outcomes whose likelihoods you can estimate, which is what makes an expected value calculation meaningful at all. Uncertainty is a spread you cannot put likelihoods on. Ambiguity is not knowing what the outcomes even are. They are handled differently. Risk can be priced and hedged. Uncertainty is reduced by finding out, or accepted by keeping the decision reversible. Ambiguity means the decision is not yet framed, and the next step is framing, not deciding.

Then the pair that does the most damage. A good decision is one that was sound given what could be known at the time. A good outcome is one that turned out well.

Chance sits between them. A sound decision can end badly and a reckless one can end well, and if you grade decisions by outcomes you will learn to be reckless whenever you were lucky. The record exists so that the decision can be judged on what it was, not on what the dice did afterwards.

Field 8: reversibility

A reversible decision can be undone at a cost you would willingly pay. An irreversible one cannot, or only at a cost that would itself be a crisis. This one field sets the pace of the whole process.

For reversible decisions, the cost of deciding slowly usually exceeds the cost of deciding wrong. Decide fast, on partial evidence, and watch. For irreversible ones, the opposite, and the delay alternative from field 4 comes into its own.

Two further pairs belong here.

Sunk cost is money or effort already spent and not recoverable. It is not an input to the decision, because no alternative gets it back. Opportunity cost is the value of the best alternative you give up, and it is an input to every decision, including the one to do nothing. People routinely count the first and ignore the second.

Consensus means everyone agrees. Consent means nobody objects strongly enough to block. Consultation means people were asked, and someone else decided. Which of the three a decision needs should be stated in the record before the discussion, because a group that thinks it is reaching consensus and a chair who thinks she is consulting will both leave believing the other has behaved badly. How to run that conversation so that each person's position is actually heard is the subject of communication skills.

The record, assembled

One page, filled in before the meeting rather than after it, does more for the quality of a decision than any amount of analysis added later. Analysis answers the question you asked. The record is what makes sure you asked the right one; the same discipline applied to the numbers a decision rests on is the subject of business acumen.

Common questions

Is this overkill for small decisions?

For reversible ones, yes, and the point of field 8 is to tell you so. The record is for decisions that are expensive to undo. For the rest, decide and watch.

What if we cannot agree on the objective?

Then the decision is not ready, and no comparison of alternatives will make it ready. Disagreement about the objective is a different conversation, usually a more important one, and it is worth having openly rather than fighting it out through the criteria.

How do I know if a past decision was good?

Reconstruct what was knowable at the time and ask whether the choice was sound on that basis. If you cannot reconstruct it, that is the first lesson: the record was not kept.

Who should make the decision?

Whoever will carry its consequences and has the authority to commit. Where those are different people, the record should say which one is deciding and which is being consulted, and it should say so before the discussion starts. The words for those roles are defined more closely under business vocabulary.