Repair-vs-Buy Outcome Accuracy Rate (RBOAR)

GMI-AVI-024AviationIntegration-Required

Definition

Percentage of AI-recommended repair routing decisions where the component was actually repairable at shop induction: measuring how often the repair-vs-buy model correctly predicts repairability before the component reaches the shop.

The Measurement Gap

The measurement gap RBOAR addresses: aviation MRO tracks pool availability, turnaround time, and fill rate, but none of these metrics evaluate whether the AI repair-vs-buy recommendation was correct. A component scrapped at shop induction after an AI repair recommendation shows up in TAT reports as a long TAT event and in fill rate reports as a gap, but neither report attributes the root cause to a model misclassification. The AI's accuracy in predicting repairability is never examined.

For high-value rotable components (engines, APU cores, landing gear assemblies, flight control actuators) this prediction accuracy matters significantly. A wrong repair recommendation doesn't just mean a delayed component; it means the rotable pool loses an asset it was counting on, a new-buy procurement is initiated at spot-market pricing or with weeks of lead time, and aircraft routing may be disrupted to avoid an AOG. The cost of a single wrong repair-routing decision for a major engine module can exceed $500,000 in expedite costs and revenue disruption.

The decision RBOAR supports: should the repair-vs-buy model's confidence threshold be raised for specific component types? Should a minimum inspection requirement be added before routing decisions are finalized? Should new-buy safety stock be increased to buffer against BER discoveries? These decisions require knowing the AI's actual accuracy at predicting repairability, which current reporting does not provide.

Formula

Plain language: Of all components the AI recommended for repair routing, count how many were actually repaired and returned to serviceable condition. Divide by total repair-routed components.

Notation:
RBOAR = Successfully_repaired / Total_repair_routed

Variables:

  • Successfully_repaired = components routed for repair by AI recommendation that completed repair and were returned to serviceable condition (count, from shop completion records)
  • Total_repair_routed = all components routed for repair based on AI recommendation in the measurement period (count, from materials/routing system)

Units: Percentage. 100% = every repair-routed component was actually repairable. The complement (1 − RBOAR) is the BER-at-induction rate: components found beyond economical repair only after reaching the shop.

Also track:

  • Scrap_at_induction_rate = 1 − RBOAR (percentage of repair-routed components scrapped at shop induction)
  • Average_TAT_penalty = additional days of AOG or aircraft-on-ground exposure incurred when a repair-routed component is scrapped and a new procurement must be initiated

Edge case handling:

Repairs completed with additional findings or cost overruns: count as "repaired" for RBOAR (the component is back in service). Track cost variance separately from RBOAR: cost overruns are a different problem than BER.

Components awaiting parts in the repair cycle: do not count as BER until final disposition. Long TAT from parts shortages is a supply chain issue, not a prediction accuracy issue.

Multiple repair attempts: count the final outcome (repaired or scrapped). Intermediate dispositions are not RBOAR events.

Segment by component type: aggregate RBOAR may mask high BER rates for specific high-value component families. Major component types (engines, landing gear, APUs) should each have their own RBOAR tracking.

Proposed formula, pending validation pilots.

At a Glance

Category
Predictive Life Management Accuracy
← Prediction Source Accuracy Differential (PSAD)Alert Economic Timing Index (AETI) →

Assess Your Organization

The AI-GPS Assessment evaluates your AI governance maturity and recommends KPIs like this one for your implementation roadmap.

Take the Free AI-GPS Assessment