What Deming Established
Deming's contribution was precise. Variation is inherent in every process — not as a failure condition, but as a structural property. The question is not whether variation exists but whether a given observation represents common cause variation (noise within the process's natural range) or special cause variation (a genuine signal that something has changed).
Natural process limits — not arbitrary targets — define the boundary between signal and noise. An organisation that responds to common cause variation as if it were a signal is tampering with its own process. Deming called this the most common and most damaging management error. The XmR chart is the operational instrument: a moving average with calculated upper and lower natural process limits, updated with each new observation.
A threshold set by management preference is not a measurement instrument. It is a target. Targets produce target-chasing. Natural process limits produce signal detection. These are not the same activity, and they do not produce the same organisational behaviour.
Why the Field Looked Elsewhere
Deming's work was absorbed into quality management and manufacturing. The Total Quality Management movement carried his methods into production environments. Within those domains, the influence is traceable and substantial.
Performance measurement developed within a different disciplinary silo — strategic management. The BSC appeared in 1992 and dominated the next two decades. The academic infrastructure around performance measurement organised itself around strategy execution, not statistical process control. Cross-disciplinary reading was not incentivised. The Deming literature and the performance measurement literature accumulated in parallel without intersection.
The Timeline the Field Chose to Ignore
Drucker introduces management by objectives. Performance measurement emerges as a management discipline inside strategic thinking.
Deming publishes Out of the Crisis. Natural process limits, XmR methodology, the tamper/signal distinction. The statistical foundation is available.
Kaplan & Norton publish the Balanced Scorecard. The field accelerates — inside strategic management, away from Deming's statistical foundation.
Bititci asks the discipline question directly: where does performance measurement belong? The structural problem is formally identified. The statistical foundation is still not integrated.
The MetriqOne Trilogy closes the argument. Deming's statistical foundations are integrated into the stack at the PI layer. Trackable replaces Time-bound. Natural process limits replace arbitrary targets. The MCA — Meta, Challenges, and Aspirations — establishes context as the precondition for stack configuration.
The Cost of Not Listening
Performance measurement frameworks without statistical foundations produce target-chasing, not signal detection. Arbitrary thresholds replace calibrated process limits. Managers respond to variation as if it were a signal. Deming called this tampering — and he was precise about its consequences: tampering increases variation, it does not reduce it.
The field accumulated sophisticated strategy tools and inadequate measurement instruments. The BSC, OKRs, and their descendants are goal alignment systems. They tell organisations what to aim at. They do not tell organisations whether their operational processes are within the bounds required for the aim to remain achievable. That question requires statistical foundations. Deming provided them. They were not borrowed.
What Listening Would Have Required
Integrating Deming's statistical foundations into performance measurement architecture requires four structural commitments — and one precondition. The precondition is context: the MCA. Before a single PI threshold is set, the organisation's actual operating conditions must be documented through the Meta, the specific Challenges whose deterioration would constitute a governance failure, and the Aspirations that give the deployment its directional orientation.
The four structural commitments then follow: replacing Time-bound with Trackable — converting a completion test into a longitudinal measurement instrument. Building the moving average and natural process limits into the PI layer. Treating miscalibration as the primary failure mode. Maintaining a calibration log — an immutable record of every threshold decision, every revision, and the causal argument behind each change.
A public health intervention programme set monthly targets for case detection. Targets were based on the programme manager's expectations, not on process history. Variation below target triggered escalation. Variation above target triggered commendation. Neither response was warranted — both were responses to common cause variation within a process that had not yet established its natural limits.
After twelve months, the XmR analysis of the data revealed that eleven of the twelve monthly responses had been tamper events. One genuine special cause signal had occurred in month nine. It had been obscured by the noise of the other eleven responses. Deming had described this failure mode in 1982. The methodology to avoid it was available. It had not been applied.
Takeaway
Common cause variation is not a signal. Responding to it as if it were a signal is tampering. Tampering increases variation. The XmR chart identifies the boundary between noise and signal before the response decision is made.
Arbitrary targets are not thresholds. A threshold calibrated from process history converts observations into evidence. A target set by management preference converts observations into performance scores.
Time-bound is a deadline test. Trackable is a calibration instrument. The difference determines whether a PI produces a longitudinal signal or a completion record.
The moving average is not optional. Without it, natural process limits cannot be established. Without natural process limits, the threshold is arbitrary.
The disciplinary silo problem is not an excuse. Deming's work was available. The cost of not reading across disciplinary boundaries was borne by every organisation that spent forty years chasing targets instead of detecting signals.