First Pass Yield (FPY) is among the most widely reported operational quality metrics, yet its interpretation degrades once delivery is distributed across sites, shifts, contractors, or jurisdictions. Observed dispersion then reflects structural endowments, contextual exposures, measurement artifacts, and managerial practice at once, and conventional benchmarking treats the mixture as a single signal, producing misattributed accountability and improvement programmes aimed at variables no local manager can move.
The paper develops the Controllable-Variable Isolation (CVI) framework, separating the portion of FPY variation actionable at a defined organizational level from the portion that is not. It proceeds through four stages: normalization across heterogeneous sites, variance decomposition and driver attribution, controllability gating through a three-test screen, and prioritization by expected yield elasticity per unit of implementation cost. Two constructs anchor it: a five-dimension controllability score built from authority, latency, cost, reversibility, and observability; and a level-shifting rule reassigning high-leverage but locally uncontrollable drivers to the level where authority resides.
The framework is synthesized from conceptual and review scholarship across ten streams, spanning lean process improvement, multi-site performance management, procurement and supplier governance, reliability engineering, safety analytics, audit and measurement integrity, predictive decision support, service and IT operations, regulated health and aviation delivery, and asset-intensive infrastructure. Eight propositions are stated and an illustrative multi-site application is worked through. CVI is a discipline for establishing whether a defect driver can be acted upon before it is assigned to someone to act upon.