Steady Rides, Steady Profits: Detecting Customer Routines to Optimize CRM Strategies

Journal of Marketing Research Scholarly Insights are produced in partnership with the AMA Doctoral Students SIG – a shared interest network for Marketing PhD students across the world.

Imagine two customers: one gets a rideshare five times a week at random intervals, and another gets a rideshare every weekday at the same time. Is one of these customers more valuable than the other?

A Journal of Marketing Research study discusses the significance of routines and their impact on customer relationship management (CRM) in the context of the ridesharing industry. The authors note that although previous research has emphasized the role of habits in shaping consumer behavior, scant attention has been directed toward comprehending the ramifications of routines—defined as repetitive behaviors characterized by recurring temporal structures—on customer management.

The study reveals that routineness correlates with heightened future usage and engagement rates, lower price sensitivity, and greater resilience to service disruptions such as pickup and drop-off delays. Furthermore, the research shows how the nature of customers’ routines (e.g., leisure- or work-oriented routines) influences these outcomes, thus suggesting the relevance of routines in customer segmentation and targeting strategies.

Routineness correlates with heightened future usage and engagement rates, lower price sensitivity, and greater resilience to service disruptions such as pickup and drop-off delays.

While these results may sound intuitive, gauging routines using transactional data is a big challenge, especially considering the substantial variability of customer routines. To address this, the authors develop a new methodology that utilizes Bayesian nonparametric Gaussian processes (GPs) at the customer level to estimate routines using transactional data. This approach surpasses traditional methods by employing a pioneering kernel that allows for flexible and precise routine estimation. By integrating these GPs within heterogeneous Poisson usage processes, the authors can estimate customer routines and separate their usage patterns into routine and nonroutine components.

This research emphasizes the importance of routines in consumer behavior and loyalty in the ridesharing industry. By comparing customers with similar historical usage or purchase rates, the study indicates that those who incorporate the service into their routines tend to demonstrate greater long-term value and overall favorability. These findings underscore the significant role of routines in shaping consumer behavior and highlight their implications for effective CRM.

Highlights of this research:

Routineness profoundly impacts customer value.
The routineness of your customers are can be segmentation strategy and an overall KPI in understanding your customer base.
The authors developed a new method to quantify this phenomenon of routineness using standard transactional data.

We caught up with the authors to get more insights into their study:

Q: How did you identify that routines could be a useful input for research in terms of CRM?

A: Habits have often been studied in the literature, but from our own experiences, routines frequently shape our consumption. In conversations, our partner company also identified routines as important drivers of ridesharing behavior. However, the company didn’t know how to measure these routines and assess their downstream consequences.

Q: Can the research predict which customers are more likely to develop ridesharing routines on the basis of their interaction patterns?

A: Our model leverages past riding behavior to identify routine customers. We also find that routine customers exhibit certain ride characteristics such as lower prices and convenience. While not causal, these results indicate potential predictiveness. More research is needed to firmly establish which interventions more generally would cause the emergence of routines.

Q: Is there value in considering routines not only in terms of timing but also in terms of the specific actions customers take during their interactions with the service?

A: We explored this in several ways, most notably by location. Our initial analysis found minimal incremental value from modeling location routines beyond temporal routines. However, future work on joint “when and where” modeling could prove fruitful.

Q: In what ways are routine customers shown to be better customers beyond just lifetime value?

A: This research offers a method to quantify this phenomenon using plain vanilla transactional data. This approach is broadly applicable across industries and sectors. By measuring the impact of routines on customer lifetime value, the study offers actionable insights for businesses seeking to enhance customer relationship management strategies. Beyond lifetime value, routine customers displayed lower price sensitivity and higher resilience to service failures, which are key indicators of being “better” customers.

Q: How do routines affect customer value and retention?

A: We find that routines positively predict both value and retention. Importantly, this held even when controlling for past usage and other transaction timing summaries such as clumpiness. This speaks to the distinct explanatory power of routines. While it’s possible that routines cause customer value, our paper did not test this assumption, as we did not change routineness exogenously. Instead, our paper focuses on the development of routineness as a metric and how to capture it from transactional data.

Q: What role do routines play in segmentation and targeting strategies for businesses?

A: We found that different routine types exhibited systematically different behaviors and value levels. Accordingly, routineness can be a useful base for segmentation and targeting. Because routines can be identified from basic transaction data, they may prove fruitful for targeting in other contexts, even with limited data.

Q: How can firms optimize the provision of services based on an understanding of customers’ routines?

A: We highlighted price optimization as one possibility, as different routines seem to exhibit different price sensitivities. More broadly, as routines increase predictability at a very granular level (hours), they can be used to better forecast demand and optimize the provision of drivers. Broader research on interventions tailored to customer routines could further assist service personalization. We have only scratched the surface of the optimization possibilities.