Automated Bidding for Sponsored Ads on E-commerce Platforms

<p><span>E-commerce platforms such as Amazon offer sellers an automated targeted bidding system for sponsored ads known as dynamic bidding that optimizes a seller's bid based on a consumer's predicted likelihood of buying from that seller. We examine the implications of dynamic bidding on an e-commerce platform with horizontally differentiated sellers. We find that dynamic bidding may cause sellers to set higher prices to strategically limit their sponsored ad competition. Consequently, contrary to conventional wisdom, dynamic bidding need not lead to higher conversion rates for sponsored ads. Moreover, the platform's main source of value from dynamic bidding may come not from its ad revenue but from the impact on its sales revenue. In fact, in contrast to prior work on standalone auctions, dynamic bidding on e-commerce platforms may hurt ad revenue even when there are many sellers. Nevertheless, dynamic bidding can benefit the platform by increasing its sales revenue. These results are due to the endogenous interaction between seller pricing decisions and sponsored ad valuations that we identify in this paper. We also show that dynamic bidding can be more profitable than a pay-per-conversion scheme. Further, platforms may find dynamic bidding more valuable when sponsored ads are less attention-grabbing.</span></p>