Definition
Percentage of proactively rebooked passengers who would have actually misconnected on their original itinerary: measuring the accuracy of the AI's misconnect prediction when it triggers a proactive rebooking action before the connecting flight departs.
The Measurement Gap
Proactive rebooking (rerouting passengers before their connecting flight departs, based on a prediction that they will misconnect) is one of the most passenger-impactful tools in an airline's IROPS toolkit. Done well, it prevents the worst passenger experience: arriving at a gate just as the door closes and finding no one to help. Done poorly, it disrupts passengers who would have made their connections, forces them onto worse itineraries unnecessarily, and erodes trust in the system.
Traditional IROPS metrics measure misconnect rate (how many passengers missed their connections) and rebooking rate (how many passengers were rebooked) but not the accuracy of the proactive rebooking model's predictions. An airline can have a high rebooking rate while making large numbers of unnecessary proactive rebookings: improving one metric while degrading the passenger experience for passengers who would have connected naturally.
Decision supported:
Is the proactive rebooking model correctly identifying passengers who will misconnect, or is it unnecessarily rerouting passengers who would have made their connections? PRP answers this question by validating predictions against outcomes: for each proactive rebooking, did the passenger actually need it? This measurement enables calibration of the model's misconnect probability threshold. If PRP is 55%, the threshold is too low and is triggering proactive rebookings for too many borderline cases; raising the threshold will improve PRP without materially increasing the missed-misconnect rate.
Formula
Plain language: Of all passengers the AI proactively rebooked before their connecting flight departed, determine how many would have actually missed their connection (the connecting flight departed before they could have reached the gate given actual inbound arrival time and terminal walk distance). Divide by total proactive rebookings.
Notation:
PRP = True_misconnects_avoided / Total_proactive_rebookings
Variables:
| Variable | Definition | Source | Unit |
|---|---|---|---|
| True_misconnects_avoided | Proactively rebooked passengers where, at the time their original connecting flight departed, they had not arrived at the connecting gate (confirmed via DCS gate activity data or timing vs. minimum connect time) | DCS + flight timing comparison | Count |
| Total_proactive_rebookings | All passengers rebooked by the AI before their connecting flight departed, based on predicted misconnect | Rebooking system / PSS | Count |
Units: Percentage. 100% = every proactive rebooking was correct (the passenger would have missed their connection). Lower = more unnecessary rebookings. The complement (1 − PRP) is the unnecessary rebooking rate: passengers placed on worse itineraries when their original connection was achievable. Track both PRP and its complement explicitly; each matters to different stakeholders.
Proposed formula, pending validation pilots.