Personalized Dynamic Pricing: Hidden Consequences for Customer Churn

Abstract

Leveraging large-scale data sources to employ personalized dynamic pricing has been a prominent practice in the retail industry. However, the outcome of this pricing strategy could be mixed: On one hand, it may increase revenue and sales by charging a price close to consumers' willingness to pay; on the other hand, it may decrease store revenue due to unintended consequences such as consumer churn. Therefore, this study aims to empirically investigate the impact of personalized dynamic pricing on store performance and individual consumer behavior under different conditions. To achieve our research goal, we obtain unique access to a large data set from a leading on-demand delivery platform that has launched a personalized dynamic pricing strategy through price discounts in some of its grocery stores. Employing a quasi-experimental setting, we utilize the difference-in-differences method to estimate the causal impact of personalized dynamic pricing. First, we conduct store-level analyses and find that personalized dynamic pricing decreases store revenue. Then, to understand the underlying mechanism, we proceed with individual-level analyses and find that personalized dynamic pricing increases the individual transaction amount and frequency while increasing consumer churn. Therefore, the reduced store-level revenue could be primarily driven by increased consumer churn. Finally, we investigate the moderating effects of an essential feature of personalized dynamic pricing: the price fluctuation over time. We find that a bit more fluctuation can mitigate the negative consequences of personalized dynamic pricing by reducing consumer churn and increasing consumer transaction activities. However, too much price fluctuation intensifies the negative consequences of personalized dynamic pricing.

Keywords: personalized, dynamic pricing, revenue, churn, quasi-experiment, difference-in-differences