Decision quality and measurement¶
Measuring a decision is not asking whether it “turned out well.” A material decision must be examined at three moments:
- before deciding: framing, alternatives, evidence, values, reasoning and commitment;
- during execution: latency, action fidelity, escalation, overrides and guardrails;
- after the horizon: outcome, deviation from expectation, prudent attribution and next-cycle change.
Sprint 5's candidate system connects these moments without collapsing them into a universal score.
Decision Quality
Review defensibility before the outcome introduces hindsight bias.
CYCLEDecision Measurement System
Select a minimum metric portfolio for priority, quality, action, time, risk and learning.
ARTIFACTMeasurement Record
Preregister expectation and measurement, then add outcome without rewriting what was known.
Problem addressed¶
A dashboard can show results and a meeting can produce a decision, but neither necessarily preserves prior expectation, process weakness, authorized execution, distributed cost, compatibility of outcome with expectation or next-cycle change. Without separation, a good outcome may legitimize a weak process and a bad outcome may punish a reasonable decision under uncertainty.
Unit of measurement¶
The minimum unit is an identified decision cycle, not a tool, person or AI initiative. It states owner, date, options, action, expectation, horizon and review. Portfolio aggregation becomes meaningful only after comparable cycles exist.
No universal metric
Quality depends on stakes, reversibility, frequency, uncertainty and affected people. This system proposes an architecture and measurement contracts—not a universal benchmark or validated organizational-maturity test.
Relationship with PULSE¶
PULSE requires observable value and learning that compares expectation with result. This section operationalizes those requirements for the CDI-BoK; it neither redefines PULSE nor proves that instrumentation improves outcomes.