We study keyword auctions, and formulate an advertiser's bidding problem as a profit maximization problem in which the advertiser has to decide on how much to bid on each keyword. We consider models with budget and time constraints, as well as those when constraints are relaxed. We assume the advertiser can observe only her bid, her rank, the search frequency and the click frequency. We discuss how to determine a bid that will maximize expected profits when the search traffic for each keyword is generated by an independent Poisson process. We provide a formulation and provide an algorithm to maximize profit while estimating problem parameters. Several extensions to the model are also discussed.
Date: 16.12.2019, starting at 2 p.m.
Location: City campus 0505.EG.538