Dynamic Pricing and Bidding for Display Advertising Campaigns.

This article addresses the complex operational problem faced by digital advertising agencies in jointly optimizing pricing for new ad campaigns, bidding for viewers on ad exchanges, and allocating viewers to campaigns under uncertainty. The authors develop a Markov decision process framework to determine optimal dynamic pricing and bidding policies, providing exact solutions for finite-horizon and steady-state cases, along with scalable heuristics for large-scale problems. Key managerial insights include the identification of three critical levers—campaign pricing, bidding strategy, and queue capacity—and the demonstration that dynamic policies outperform static ones in maximizing profits and controlling delays. The study also reveals significant economies of scale from combining campaigns or merging agencies, leading to higher profits, lower prices, and reduced delays. Numerical analyses validate the effectiveness of the heuristics and highlight sensitivity to parameters such as queue capacity, arrival rates, and viewer action probabilities.