Definition
Proportion of assets receiving active predictive alerts from multiple platforms simultaneously where the recommendations conflict.
The Measurement Gap
The Measurement Gap
Most organizations deploying predictive analytics adopt platforms incrementally. Vibration monitoring arrives first, then oil analysis gets its own predictive layer, then a thermal imaging vendor adds anomaly detection. Each platform generates alerts independently, and each vendor calibrates confidence thresholds to maximize their own detection rates. The result is a growing probability that two or more platforms will issue simultaneous, contradictory recommendations for the same asset-component combination.
Legacy metrics track each platform's accuracy in isolation. A vibration platform might report 85% alert accuracy while the oil analysis platform claims 90%. Neither metric captures what happens when both platforms flag the same gearbox but one recommends immediate intervention and the other suggests monitoring for another 60 days. The planner receiving both alerts has no systematic way to arbitrate, so resolution defaults to whoever shouts loudest or whichever platform the planner trusts more.
This hidden conflict creates two problems. First, planning time inflates as planners manually reconcile contradictions. Second, inconsistent arbitration degrades the value of both platforms because neither platform's recommendation is followed reliably. Without PPCR, organizations cannot quantify how much multi-platform overlap costs in planning overhead or identify which platform combinations generate the most friction.
Decision Supported:
How often are maintenance planners forced to reconcile contradictory AI recommendations, and what is the planning overhead of multi-platform conflict?
Formula
Plain Language: Count the number of times two or more predictive platforms have active alerts on the same asset-component where those alerts recommend incompatible actions. Divide by the total number of instances where two or more platforms have overlapping active alerts on the same asset-component. Multiply by 100 for a percentage. Track planning overhead separately by multiplying mean additional planning hours per conflict by the number of conflicts per period.
Notation:
PPCR = (Conflicting_alert_sets / Total_multi_platform_alert_sets) x 100
Planning_overhead = Mean_additional_planning_hours_per_conflict x Conflicts_per_period
| Variable | Description | Source |
|---|---|---|
| Conflicting_alert_sets | Instances where two or more platforms have active alerts on the same asset-component with incompatible recommendations. Incompatibility is defined as: timing recommendations differing by more than 50%, severity scores differing by more than 3 points on a normalized scale, or contradictory recommended actions (e.g., replace now vs. continue monitoring). | Cross-platform alert correlation |
| Total_multi_platform_alert_sets | All instances where two or more platforms have active alerts on the same asset-component, regardless of whether recommendations align or conflict. | Cross-platform alert correlation |
| Mean_additional_planning_hours_per_conflict | Average planner time spent resolving a single conflict instance beyond normal alert processing. | Time tracking or planner survey |
| Conflicts_per_period | Count of conflicting alert sets within the reporting period. | Derived from Conflicting_alert_sets |
Proposed formula, pending validation pilots.