Decision Quality Model™

Purpose

The Decision Quality Model™ helps leaders judge whether a project decision was well made at the time it was made, independent of how it later turned out. It supports decisions about whether to proceed, redesign or reverse major commitments by testing six dimensions: framing, alternatives, information, values, reasoning and commitment to action. Its purpose is to separate luck from quality in outcome reviews.

How to Use It

Convene the decision team before a major commitment and score each of the six dimensions explicitly. First, confirm the frame: are objectives and context correctly defined? Second, demand genuine alternatives, not one option with a straw man. Third, audit the information base: distinguish verified data from assumption, and log key uncertainties. Fourth, surface values: make trade-offs between cost, schedule, safety and scope explicit rather than implicit. Fifth, stress-test the reasoning: does the logic connecting evidence to recommendation survive challenge from a dissenting reviewer? Sixth, confirm commitment to action: are the people who must execute genuinely signed up? Proceed only when all six pass; otherwise identify the weakest link and repair it before committing.

Worked Example

Applied to the JPMorgan Chase 'London Whale' episode of 2012, in which the bank's Chief Investment Office reported trading losses exceeding $6 billion, the model exposes multiple weak dimensions. The frame was blurred: a unit presented as hedging risk was in effect taking large, concentrated positions in credit derivatives. Alternatives to unwinding or restructuring the position early were not seriously entertained as losses grew. Information was unreliable — subsequent US Senate investigation findings described mismarked positions and risk models altered in ways that understated exposure. Values were conflicted: short-term reported results competed with prudent risk limits. Reasoning failed as escalating positions were justified rather than questioned. Even commitment to action was fractured, with traders and senior management not aligned. The eventual loss was not bad luck; it was the predictable output of low-quality decisions.

Limitations

The model assesses process quality, not outcome probability; even excellent decisions can fail under adverse conditions, and poor ones sometimes succeed. Scoring the six dimensions involves judgement and can itself be gamed by teams seeking approval rather than rigour. It also requires access to how the decision was actually made — information frequently unavailable to outside analysts, who must rely on investigations and testimony.

Relationship to PMOS

PMOS is being designed to operationalise this model as a decision-gate checklist: each major commitment recorded in the system will require evidence against the six dimensions before approval. Until PMOS is released, the model remains a manual discipline PIA applies in its investigation reviews.

Version 1.0Published 2026-08-01Review date 2027-02-01