Bidding for an optimal portfolio of keywords in sponsored search advertising: From generic to branded keywords

Abstract

With the increasing prominence of digital media, retailers attempt to attract consumers to their websites by investing in sponsored search advertising. However, due to stiff competition among retailers, sponsored search advertising can be expensive. This paper develops a multi-period, dynamic programming model that provides a retailer with an optimal portfolio of generic and branded bids. We model two critical aspects of consumer search behavior: (i) the spillover effect of generic searches leading to branded search arrivals in subsequent periods and (ii) the memory effect that leads to a decline of consumer awareness of a brand over time. We find that the retailer can effectively shuffle his investments on generic and branded keywords depending on several consumer parameters, e.g., awareness level, brand retention and reservation price variances. We develop a bidding policy framework to highlight the shift in bid shares from generic to branded at different levels of consumer awareness. We find that harnessing the benefits of spillover from generic to branded keywords allows the retailer to save on generic bids at higher awareness and retention levels and lower variance in consumers’ reservation prices. Further, we extend our model to different consumer purchase situations/ product classifications, viz., Convenience, Shopping and Specialty purchasing. Our analysis suggests prevalence of generic bids for certain purchase/product situations, whereas branded bids remain salient in other situations.