Definition
Average degradation in passenger journey quality between the original booked itinerary and the AI-assigned rebooking itinerary during IROPS, measured across four dimensions: additional connections added, additional elapsed travel time, cabin class downgrade, and involuntary carrier change.
The Measurement Gap
Standard IROPS metrics count passengers recovered (the number moved to an alternative itinerary) but do not measure the quality of what they were moved to. Completion Rate confirms the airline got passengers to their destination but is blind to whether they arrived with two additional connections, in a lower cabin, on a partner carrier, and eight hours later than planned. IROPS delay attribution scores time against the original schedule but cannot capture the compounding harm of connection additions and class downgrades on the passenger's actual experience. Customer satisfaction surveys capture emotional aftermath but cannot be attributed back to the specific AI rebooking decision: surveys arrive days later and don't distinguish AI-assigned recovery from agent-assisted recovery. The result is that an AI rebooking engine can systematically produce technically "successful" recoveries that are experientially poor, and no existing metric will surface the pattern. PJDI closes this gap by scoring recovery quality at the individual itinerary level, making visible the trade-offs the AI is making across recovery events and passenger segments.
Decision Supported:
Is the IROPS rebooking AI producing acceptable passenger outcomes, or is it optimizing for journey completion at the expense of journey quality?
Formula
Plain language: For each rebooked passenger, score the degradation across four dimensions: additional connections added, additional elapsed travel time, cabin class downgrade, and carrier change. Normalize each dimension to 0–1, apply dimension weights, and average across all rebooked passengers in the period.
Notation:
PJDI = (1/N) × Σ [ w₁ × ΔConnections_i + w₂ × ΔTime_i + w₃ × CabinDrop_i + w₄ × CarrierChange_i ]
Variables:
| Variable | Definition | Unit |
|---|---|---|
| N | Number of passengers rebooked by AI in the measurement period | Count |
| ΔConnections_i | (Connections in new itinerary − connections in original) / max_connections_cap, capped at 1.0 | 0–1 |
| ΔTime_i | (Elapsed time new − elapsed time original) / max_time_cap, capped at 1.0; proposed cap: 12 hours | 0–1 |
| CabinDrop_i | 1 if passenger was downgraded from booked cabin, 0 otherwise | Binary |
| CarrierChange_i | 1 if rebooked to a different operating carrier, 0 otherwise | Binary |
| w₁, w₂, w₃, w₄ | Dimension weights: proposed starting: 0.30, 0.35, 0.20, 0.15 (time matters most, connections second) | Decimal |
Weights should be validated against passenger complaint correlation data during the pilot. Airline-specific complaint patterns may shift weights depending on the passenger mix.
Proposed formula, pending validation pilots.