Current Trends in Digital Marketing
(A2020-62468)

AUTHORS

Edlira Shehu, Copenhagen Business School; Nadia Abou Nabout, WU Vienna

ABSTRACT

The special session aims to discuss current trends in digital marketing. Each of the unpublished papers provides a new substantive and/or methodological contribution. Below, we briefly describe each of the papers and their respective contributions in more detail:

#1: The value of cursor information in predicting purchase behaviour (Atalay, El Kihal & Ellsaesser): Retailers are interested in predicting customers’ purchase probability to target customers in real time. The authors develop a Long Short Term Memory neural network and use cursor flow data to show that purchase likelihood can be predicted very accurately using only the first 20 seconds of customer visits. A simulation study illustrates how retailers can use this model to individualize offers and considerably increase their margins.

#2 Placement disclosure in ad auctions: Evidence from a policy change (Ada, Abou Nabout & McDonnell Feit): The authors analyze how information disclosure of ad placements affects ad auction outcomes. They analyze data from a leading European ad exchange, which sells about 170M impressions weekly via real-time bidding auctions. Using data before and after a policy change, they show that average revenue per impression rose due to information disclosure. This effect was stronger for sites with thick markets.

#3 Horizontal and vertical software multihoming for entertainment platforms: Impacts on hardware sales and market share (Wiegand, Peers & Bleier): The authors analyze how horizontal and vertical multihoming strategies affect sales and market shares of different entertainment platforms. Using monthly hardware sales data and over 8,800 software releases of the leading consoles, PlayStation, Xbox, and Wii, they show positive effects of horizontal software multihoming, whereas the impact of vertical multihoming depends on the specific platform. In addition, the authors highlight specific software characteristics that influence the multihoming release effects.

#4 Understanding Multi-Channel Dynamics (Shehu & Naik): Given the highly interactive multichannel multimedia environment, firms face the challenge on how to allocate advertising budgets across multiple media, while accounting for interdependencies between sales channels. The authors propose a multimedia multichannel model, which accounts for channel specific dynamics and multichannel interdependencies. They use weekly data from a leading retailer over a period of three years, and estimate the model via Kalman filter. In addition, they analytically derive a closed-form solution for the optimal budget allocation, and demonstrate that ignoring channel dependencies leads to sub optimal managerial decisions.