What is Decision Hygiene?

Organisations pay a premium for brilliant judgement, then act surprised when the same brilliant professional, given the same case on two different days, returns two different answers. The scatter is not a character flaw. It is a property of human judgement, and it has a name.

In Noise: A Flaw in Human Judgment (2021), Daniel Kahneman, Olivier Sibony and Cass Sunstein separate two kinds of error. Bias is a shared tilt: judgements that err in the same direction, like optimism about delivery dates. Noise is scatter: the same case drawing different answers depending on who judges it, when, and in what mood. Bias gets the headlines. The book’s uncomfortable argument is that in repeated professional judgements (pricing, estimating, underwriting, hiring, scoping), noise is routinely the larger source of error, and the less examined one, because it hides inside case-by-case plausibility. Every individual judgement looks reasonable. Only the spread is damning, and nobody is looking at the spread.

Their remedy is not better geniuses. Decision hygiene is the discipline of removing error you cannot see: independent judgements before discussion, structure before debate, intuition last. The analogy with handwashing is deliberate. Hygiene is preventative rather than corrective, it targets no single named error, and you never find out which mistake you avoided. Which is also why it is neglected: nobody gets credit for the infection that did not happen.

The core disciplines

Four practices carry most of the weight:

• Independent judgements first. Collect each person’s view, in writing, before anyone speaks. Groups do not average away their members’ errors; they amplify their earliest signals. The first confident voice anchors the room, seniority and eloquence carry weight they have not earned, and views cascade. Independence preserves the genuine spread, and the spread is information: where independent judgements disagree most is where the decision needs the most work.
• Structured decomposition. Break the decision into the handful of dimensions that actually matter and assess each one separately, on its own evidence, before forming any overall view. Kahneman, Dan Lovallo and Sibony formalised this as the Mediating Assessments Protocol. Decomposition stops a strong impression on one dimension from quietly colouring all the others: the halo effect that lets a confident sponsor make every number look better.
• Relative rather than absolute judgements. People compare more reliably than they score. Asking whether this engagement is riskier than the last three is a better question than asking for a risk rating out of ten.
• Intuition delayed, not banished. Expert intuition is real, and hygiene does not discard it. It sequences it. The holistic verdict comes last, after the structured assessments are on the table, so that intuition integrates the evidence instead of replacing it.

Why hygiene beats brilliance

In a single, unrepeatable decision, you might get away with talent. Organisational decisions are not like that. Pricing, scoping, hiring, estimating and reviewing repeat weekly, and a process that repeats amplifies scatter into cost: quotes that vary with the quoter, estimates that vary with the estimator’s calendar, reviews that vary with the room. Hygiene compounds across those repetitions in a way brilliance cannot, because a modest improvement applied every time outperforms an occasional stroke of genius.

Hygiene also survives its practitioners. Brilliance is not transferable; when the exceptional partner leaves, their judgement leaves with them. A hygienic process is teachable, inspectable and auditable: you can verify that judgements were collected independently, but you cannot verify that someone was brilliant. For an organisation that intends to build judgement that belongs to the organisation rather than to individuals, that inspectability is worth more than any single gift.

One concrete example

Clearly illustrative, with no customer implied. A services firm must choose between two delivery plans for a difficult transition. The habitual version: the sponsoring director presents plan A with conviction, the room reacts, the most senior sceptic speaks second, and forty minutes later the group has negotiated its way to the average of its loudest voices. The hygienic version: before the meeting, each reviewer independently scores both plans on three mediating assessments (staffing readiness, dependency risk, client readiness), in writing. The meeting opens with the spread. Staffing and dependency scores cluster; client readiness scatters wildly, which tells the room where its real uncertainty lives. Discussion concentrates there, the missing evidence is named, and only then does the group form its overall judgement. Same people, same evidence, different machinery. The second version is slower by an hour and better by a decision.

Hygiene built into the machinery

Most hygiene fails not because anyone disagrees with it but because it depends on a chairperson remembering to enforce it under time pressure. Decision intelligence builds the disciplines into the machinery instead. Evidence is labelled before anyone argues from it: every fact carries a quality state from the evidence hierarchy (measured, modelled, inferred, stated or unmeasured), so an assumption cannot impersonate a measurement no matter how confidently it is spoken. Decisions are decomposed by construction: a scenario run prices candidate options against each binding constraint separately, which is mediating assessment as machinery rather than as protocol. And the sequence is enforced: the structured evaluation exists before the group convenes in a Decision Room, so discussion starts from the assessments rather than from the most confident voice, and the human verdict comes last, recorded alongside everything it overruled.

Brilliance is welcome anywhere. Hygiene is what makes it repeatable.