When will a customer purchase again?

Video content: https://www.youtube.com/watch?v=n91gSSS4mRA

Companies are often clueless about when their consumers will purchase again. New research by Evşen Korkmaz from Rotterdam School of Management, Erasmus University (RSM) will help businesses determine how much time and effort individual customers are worth investing in. Korkmaz reveals that by using existing customer base analysis models but extending their analytic capabilities, companies can predict the specific interests, and purchasing behaviour of customers at an early stage in order to develop individual marketing strategies.

Recent years have seen many advances in quantitative models in the marketing literature. These advances enable model building for a better understanding of customer purchase behaviour and customer heterogeneity in a way for firms to develop optimal targeting and pricing strategies. Still, it has been observed that not many of the advanced models have found their way into business practice. With her dissertation Understanding heterogeneity in hidden drivers of customer purchase behaviour, Evşen Korkmaz bridges the gap between advanced models and their business applications by systematically extending the use of models.

Sometimes a company hasn’t heard from a particular customer for a while, due to matters such as a changing family situation, holiday leave, or the loss of interest in your product. While a customer does not often share this information, companies should know whether that customer is still active and worth approaching. To map this out, businesses use probabilistic models that utilise customer data from the past to predict their future behaviour. However, in her dissertation, Korkmaz exposes that much more can be read with these existing models.

Korkmaz focuses on probabilistic customer base analysis models that deal with understanding customer heterogeneity and predicting customer behaviour. These models specify a customer's transaction and defection processes under a non-contractual setting. Through this study, she shows that the timing of the next purchase for each customer can be predicted using these models. This information can help a company decide whether the customer is worth investing time, money and effort in.

Building two new models, Korkmaz interprets customer heterogeneity in a more flexible and insightful way. As a result, managers can obtain a refined segmentation. Customers are no longer analysed as one heterogeneous group, but rather as several heterogeneous groups, so-called customer segments, so that one can analyse within and across heterogeneity, for whom both individual and segment-specific marketing techniques can be determined.

Businesses can learn from this research how to get a hold of more – and more accurate – information on their customer’s behaviour by using data they already have.

Dissertation abstract

Recent years have seen many advances in quantitative models in the marketing literature. Even though these advances enable model building for a better understanding of customer purchase behavior and customer heterogeneity in a way that firms can develop optimal targeting and pricing strategies, it has been observed that not many of the advanced models have found their way into business practice. This thesis aims to bridge the gap between advanced models and their business applications by systematically extending the use of models. Korkmaz first focuses on probabilistic customer base analysis models that deal with understanding customer heterogeneity and predicting customer behaviour. These models specify a customer's transaction and defection processes under a non-contractual setting. Through this study, she shows that the timing of the next purchase for each customer can be predicted using these models. Korkmaz also extends them by modeling customer heterogeneity in a more flexible and insightful way. As a result, managers can obtain a refined segmentation. Based on the customer heterogeneity insights, Korkmaz then focuses on pricing strategies for online retailers who derive their revenues from delivery fees and sales. In order to come up with optimal pricing strategies for delivery fees, she uses ideas from the two-part tariff literature.

Given the time and costs associated with implementing advanced models and theories in managerial practice, marketing executives need to be convinced  to implement advanced models by clearly demonstrating the contributions of such models. Korkmaz’ study serves as a step toward bridging advanced models and business practice by empirically demonstrating their extended contributions.

About the author

Evşen Korkmaz (1983, Turkey) completed a BSc in Industrial Engineering in 2005 and a MSc in Industrial Engineering with a specialisation in operations research in 2008 from Istanbul Technical University with distinction. She started her PhD research at Rotterdam School of Management, Erasmus University in 2009. Her main research interest lies on the interface of marketing modeling and operations research. Her research has been presented at several international conferences such as INFORMS Annual Meeting, Production and Operations Management Conference and ISMS Marketing Science Conferences.

Rotterdam School of Management, Erasmus University (RSM) is ranked among Europe’s top 10 business schools for education and among the top three for research. RSM provides ground-breaking research and education furthering excellence in all aspects of management and is based in the international port city of Rotterdam –  a vital nexus of business, logistics and trade. RSM’s primary focus is on developing business leaders with international careers who carry their innovative mindset into a sustainable future thanks to a first-class range of bachelor, master, MBA, PhD and executive programmes. RSM also has offices in the Amsterdam Zuidas business district and in Taipei, Taiwan. www.rsm.nl

For more information about RSM or on this release, please contact Ramses Singeling, Media Officer on +31 10 408 2028 or by email at singeling@rsm.nl.

Share this article: