Maximally representative allocations for guaranteed delivery advertising campaigns (Q2375891)

This article presents a model for pricing of online advertising banners which is based on randomized bidding and uses the relative scarcity of each advertisement which is well-behaved under uncertainty. The problem tackled by authors is that when the variety of goods for sale is very large, inverse-elasticity methods to allocate prices cannot be used. The article begins with an overview of the proposed model including the necessary notation, followed with a presentation of the optimization model, which considers minimizing the weighted squared distance of the within-supply pool market shares from representative market shares. The model is then optimized and the economic interpretation of the solution is analyzed. The practical applications of the model are also described, including an implementation of the model in the Yahoo! advertising mechanism.