Definition
Proportion of systemic root causes that AI-prioritized individual-mode investigations fail to identify.
The Measurement Gap
The Measurement Gap
AI-driven reliability prioritization excels at ranking individual failure modes by frequency, cost, or predicted occurrence. The algorithms identify that "bearing failure on pump X" is the top priority, then "seal leak on conveyor Y," then "gearbox wear on crusher Z." Investigations follow these rankings, and each investigation produces findings specific to its failure mode.
What this approach systematically misses are root causes that cut across multiple failure modes simultaneously. A contaminated lubricant supply line might be causing bearing failures on pumps, seal degradation on conveyors, and accelerated gear wear on crushers. Investigated individually, each failure mode gets its own local explanation: bearings replaced with cheaper alternatives, seals exposed to temperature cycles, gearbox overloaded. The systemic root cause, contaminated lubrication, only becomes visible when someone steps back and asks why three different failure modes across three different asset types are all trending upward at the same time.
Legacy reliability programs rely on experienced engineers to notice these cross-cutting patterns. AI-driven prioritization, by directing investigation resources toward the highest-ranked individual modes, can actually reduce the likelihood of systemic discovery because investigators are focused on mode-specific questions rather than fleet-wide patterns. Without SPDG, organizations have no way to measure how much systemic insight they are losing to mode-by-mode optimization.
Decision Supported:
Is the AI's failure-mode-by-failure-mode prioritization missing systemic issues that drive multiple failure types simultaneously?
Formula
Plain Language: Count the total number of systemic root causes identified through any means during the reporting period. Then count how many of those systemic root causes were discovered through AI-prioritized investigations. Subtract the AI-discovered fraction from 1 to get the gap. A result of 0 means the AI's prioritization led investigators to every systemic pattern. A result of 1.0 means the AI's prioritization failed to surface any systemic patterns, and all were found by other means.
Notation:
SPDG = 1 - (Systemic_from_AI / Total_systemic_identified)
| Variable | Description | Source |
|---|---|---|
| Systemic_from_AI | Count of systemic root causes that were identified through investigations triggered or prioritized by the AI reliability platform. A systemic root cause is one that contributes to three or more distinct failure modes across two or more asset types. | RCA database with investigation trigger tagging |
| Total_systemic_identified | Count of all systemic root causes identified during the reporting period, regardless of how the investigation was triggered. Includes AI-prioritized, engineer-initiated, incident-driven, and any other discovery path. | RCA database with systemic classification |
Proposed formula, pending validation pilots.