The Decision Audit Index

I.1 — Judgment degrades exactly where the stakes rise

The finding: Founders and chief executives report, at near-consensus, that the pressure surrounding their most consequential decisions actively impairs the decisions themselves.

In a 2023 survey of 230 venture-backed founders conducted by Balderton Capital, 88% agreed that excessive stress results in bad decision-making^[1]. Sixty-four percent said sustained pressure negatively affects business performance. Eighty-three percent identified diminishing returns from additional working hours — a finding whose implication for deliberation is direct: past a threshold, continued effort on a decision does not improve it^[1].

The mechanism is not exotic. Cognitive depletion narrows the range of options a person can hold simultaneously, increases reliance on fast heuristic processing, and raises susceptibility to whichever frame is most immediately available^[2]. The person deciding while depleted is not deciding badly because they are careless. They are deciding badly because the apparatus of deliberation is the thing that has been depleted.

The consequence for the class of decision this Index concerns: irreversible decisions are disproportionately made at the moment the decider is least equipped to make them, because irreversibility and pressure co-occur. The decision that can be undone can also wait. The one that cannot, cannot.

I.2 — The decisions most delayed are the ones most examined by no one

The finding: The decisions leaders consistently postpone are not the ones lacking information. They are the ones carrying personal cost.

Layoffs. Underperforming senior hires. Co-founder conflict. Across the coaching and advisory literature, these cluster at the top of what gets deferred, and they share a structure: each threatens the decider's identity, relationships, or self-narrative, independently of whether the decision is correct^[3].

Delay in these cases is routinely experienced, from the inside, as diligence. I am gathering more information. The timing is not right. The team is not ready. The distinction between genuine deliberation and avoidance is not visible from within the delay: genuine deliberation is bounded — it has a horizon, named criteria, and a commitment to act when those criteria are met. Avoidance is open-ended, and its criteria shift each time they are approached^[3].

What this means: the most consequential decisions are precisely those least likely to receive external examination, because examining them requires the decider to first admit the decision is being avoided — which is the thing the avoidance exists to prevent.

I.3 — The decider is structurally alone

The finding: A majority of founders report having no one who understands the pressures under which their decisions are made, and the isolation is not merely uncomfortable — it is causally implicated in outcomes.

Sixty-one percent of surveyed founders said they find it hard to find support from people who understand the challenges they face^[1]. Only 33% turn to their investors for professional support; most rely instead on personal relationships, which the same research found deteriorate under the pressure^[1]. Peer-reviewed work has found that entrepreneur loneliness increases business-exit intentions through reduced entrepreneurial passion — isolation does not merely accompany difficulty; it changes what the person decides^[4].

The structural problem is not that the decider lacks advisors. It is that every available advisor holds a position. The board member has an interest in the outcome. The co-founder is a party to it. The spouse bears its consequences. The investor is not disinterested and cannot be. The executive coach, at €400 per hour, has a relationship to protect. Each of these people may be honest, competent, and well-intentioned, and none of them is neutral.

What this means: for the highest-stakes decisions, the counsel most needed — from a party with no stake in the outcome and no relationship to preserve — is the counsel least available.

I.4 — Confidence and evidence come apart, and the gap is invisible from inside

The finding: Confidence is not a signal of accuracy. It is a separate variable, and among experts it is systematically miscalibrated in the direction of overconfidence.

Across decades of research on expert judgment, calibration studies have repeatedly found that experts assign probabilities more extreme than their accuracy warrants^[5]. Philip Tetlock's long-running work on forecasting found expert predictions systematically overconfident, with confidence correlating poorly with correctness^[6]. The phenomenon is not restricted to forecasting: overconfidence has been documented in financial decisions, medical diagnosis, and expert judgment generally^[5].

A further finding matters more for the decisions this Index concerns. Tetlock has argued that the problem is compounded by language: experts reach for hedges — a distinct possibility, a serious risk — that can mean anything from a 20% chance to an 80% chance, and that permit the forecaster to avoid being pinned to a number. For high-status decision-makers, he observes, the incentive to make explicit numerical forecasts is weak: at best they tie, and more often they risk status if they are wrong^[7].

Moore and Healy's decomposition distinguishes three forms: overestimation of one's own performance, overplacement relative to others, and overprecision — excessive certainty in one's beliefs, expressed as confidence intervals that are too narrow^[8]. Overprecision is the form most relevant to an irreversible decision, and it is the form least visible to the person exhibiting it, because from the inside, an unwarranted certainty is indistinguishable from a warranted one.

What this means: the decider cannot audit their own confidence using their confidence. The gap between certainty and evidence is not detectable through introspection, and it does not announce itself. It is found only by examination against a standard — or by the outcome, at which point the decision has already been made.

I.5 — The reasons arrive after the decision

The finding: The explanation a person offers for a decision is frequently constructed after the decision, and its coherence is not evidence of its truth.

The mind assembles a rational account for choices made on other grounds, and the account is more convincing the more articulate its author. Tetlock notes the tendency directly: limited predictive ability, followed by great confidence in explanations constructed after the fact, with hindsight bias and retrospective determinism doing the work^[7].

The practical form of this in high-stakes decisions is a set of reasons that are unusually smooth, well-ordered, and available on demand — a narrative that has been rehearsed. The polish of the account is not evidence that the account is the cause. It is frequently evidence that it is not.

What this means: examining a decision by examining its stated reasons is insufficient, because the stated reasons may be the decision's justification rather than its origin. What must be examined is the framing — the question as the decider has posed it — which is the layer at which most consequential errors are made and the layer no fluent, helpful answer to the stated question will ever touch.

I.6 — The composite

Assembled, the research describes a specific and recurring room.

A person is facing a decision that cannot be undone. The pressure surrounding it has measurably degraded the judgment they will bring to it, and they know this, and knowing it does not help. The decision has been delayed for months in language that resembles diligence. There is no one available to them who is both competent and disinterested. Their confidence in their read is not correlated with its accuracy, and they cannot tell from inside whether it is warranted. And the reasons they would offer for their choice, if asked, may be a story assembled after the fact — fluent, coherent, and beside the point.

Every element of this is documented. None of it is unusual. It describes the ordinary conditions under which the most consequential decisions in business and in life are actually made.

The question this Index exists to hold open is narrow: what would it mean to examine such a decision before it is made, rather than after — and what standard would that examination have to meet?

The next edition will begin to answer it with evidence.