Demand and Demand Models in Revenue
Management Applications
E. Andrew Boyd
Senior Vice President and Senior Scientist
PROS Revenue Management
Heralded as the single most important factor
effecting revenue management, demand forecasting is neither simple in concept
nor practice. Using the airline industry as an illustrative example, we begin
by tracing how demand models have been influenced by data availability,
computational limitations, optim-ization models, user requirements, and
disagreements about the fundamental nature of demand itself. In doing so, we
establish a foundation for examining a demand model that provides an analytical
solution to a perceived industry problem arising from the cyclic interaction of
forecasting and optimization. Computational results are presented demonstrating
the potential for substantial revenue loss when applying a demand model that
doesn't coincide with consumer behavior. A property of robust demand models for
revenue management applications is proposed, and research directions with
practical applicability are discussed.