Leveraging Loyalty Programs Using Competitor Based Targeting

Abstract

Loyalty programs are widely used by firms but their effectiveness is subject to debate. These programs provide discounts and perks to loyal customers and are costly to administer, and with uncertain effectiveness at increasing spending or stealing business from rivals. We use a large new dataset on retail purchases before and after joining a loyalty program (LP) at the customer level to evaluate what determines LP effectiveness. We exploit detailed spatial data on customer and store locations, including locations of competing firms. A simple analysis shows that location relative to competitors is the strongest predictor of LP effectiveness, suggesting that LPs work primarily through business stealing and not through other demand expansion. We next estimate what variables best predict LP effectiveness using high-dimensional data on spatial relationships between customers, the focal firm's stores, and competing stores as well as customers' historical spending patterns. We use LASSO regularization to show that spatial relationships are more predictive of LP effects than are past sales data. Finally, we show how firms can use this type of predictive analytics model to leverage customer and competitor location data to substantially increase the performance of their LP through spatially driven targeting rules.

Keywords: loyalty programs, predictive analytics, spatial models, retail competition, machine learning