How Service Quality Variability Hurts Revenue When Customers Learn: Implications for Dynamic Personalized Pricing

Abstract

We formalize an understudied mechanism through which quality variability hurts firm revenues, analyze when and why this mechanism is important, and generate new insight into how dynamic personalized pricing strategies can mitigate the negative effects of quality variability. To do so, we model a firm that sells a service repeatedly with variable but stationary quality. Customers update their quality beliefs based on their experiences (exponential smoothing), and their purchase probability in each period increases with their respective beliefs about mean quality (logit choice). For any fixed price, we show that quality variability reduces the firm's revenue and leads to a downward bias in customer beliefs about quality. These effects arise even when customers are risk neutral and update their beliefs symmetrically after good and bad experiences. We then investigate whether the firm can improve revenues through dynamic personalized pricing. We find that a "fixed perceived surplus" pricing policy--charging a lower price when a customer believes the quality is lower to induce a constant purchase probability--is not only optimal but can also match the optimal revenue when quality is not variable. The revenue gain from implementing this pricing policy (as compared with fixed pricing) is greatest when quality variability is large, customers react strongly to recent experiences, and/or mean service quality is high. Our numerical results further show that firms can achieve significant revenue gains through dynamic pricing even in information-poor or low-price-flexibility environments. Finally, we extend our model to consider social learning as well as competition between two firms.

Keywords: service quality, customer choice dynamics, stochastic models, customer learning, behavioral operations, dynamic pricing