A rating means something only if it tracks reality — if better-rated decisions, in aggregate and over time, produce more reliable outcomes. That evidence is built, not asserted. CREI maintains a growing record of assessed decisions and their resolutions, and recalibrates the standard against it.
The grounding is not novel: the measurement of forecasting calibration, the auditing of organizational noise, and the decision-quality literature each carry decades of validated research. CREI's work is to bring them into a single standard for the enterprise decision — and to prove it, in the open. Validation is the active frontier, not a finished claim.
A rating's authority comes from observable evidence that it predicts. These are the measurement categories the validation record can track over time — categories, not asserted results.
Across resolved decisions, how closely stated confidence at the time of rating tracked what actually happened.
Whether "70% likely" occurs about 70% of the time across a decision-maker's record, and how that sharpens as the record grows.
Whether higher-rated decisions, in aggregate, resolve more reliably than lower-rated ones.
How far ahead a falsifying leading indicator surfaced before an outcome turned adverse.
Whether the same decision, re-assessed, yields the same grade — the consistency the standard is built to measure.
Whether a preserved decision record measurably improves the quality of later, related decisions.
These are measurement categories, not results. CREI does not publish validation findings or rated-decision case studies without verified, approved material. The record is being built; the standard is published as CREI DQS v0 and open for scrutiny.
Founding participants put real decisions on the record — the evidence that proves the rating predicts begins with them.