Airline revenue management is a picture-book application domain for analytics, where market analytics, forecasting and optimisation meet. While many existing revenue management models and systems assume the capacity of sold products, such as the number of seats in an aircraft, to be fixed, practice shows that this is not true.
In a recent study by Catherine Cleophas, member of the Centre for Marketing Analytics and Forecasting, and her colleagues, they examine empirical data and find that operational processes cause capacity changes of more than 10% on 40% of flights during the booking horizon. When ignoring this, revenue management can suffer from spill and spoilage, either selling too many tickets at too low a price or reserving too many seats for valuable demand that fails to arrive.
To overcome this problem, the authors propose a first scenario-based model to integrate the idea of several potential capacities into leg-based revenue management. As a result, revenue increases by more than three percentage points, depending on the demand constellation, the probability and extent of capacity updates, as well as their timing. The computational results highlight the need to consider particularly the possibility of late capacity decreases triggered by operational failures. This approach also provides benefits concerning overbooking, where not capacity, but its utilisation from tickets, is uncertain.