Optimal keyword auctions for optimal user experiences

Poor user experiences with search advertisements can lead to ad avoidance thus reduce search engine's long-term revenue. We capture the effect of negative user experiences on search engine's future revenue in a new variable called “shadow costs” and examine the optimal keyword auction mechanisms (KAMs) in a general model that takes into account advertiser-specific and position-specific shadow costs. We show that the optimal KAMs can be implemented in an ex-post equilibrium with a “progressive second price” payment rule. Furthermore, under a few special but practically relevant cases, the optimal KAM takes the form of relatively simple scoring auctions. We show that minimum bids in these scoring auctions may be advertiser- or position-specific and the allocation rule may or may not be greedy. Our results highlight impact of shadow costs on keyword auction designs and hold implications for search engines, advertisers, and internet users.