A Hidden Markov Model of Customer Relationship Dynamics

To profitably manage their customer base, firms need to capture and analyze the dynamics of customer relationships, over time.
March 9, 2009
Print this page

The Idea

To profitably manage their customer base, firms need to capture and analyze the dynamics of customer relationships, over time.

The Research

A good customer relationship management (CRM) tool tracks each and every touch point between the customer and the firm, including the exposure of customers to marketing campaigns. 

Current CRM tools, however, often produce static segmentation, which assigns customers to different loyalty levels such as not loyal, somewhat loyal and very loyal. These tools are limited in their ability to capture dynamic segmentation — or how the firm’s marketing efforts influence the customer’s movement between those levels over the lifetime of a customer’s relationship with the firm.

Professor Oded Netzer, with coresearchers James Lattin and V. Srinivasan of Stanford University, devised an econometric model for an improved CRM tool that analyzes a customer’s history with a firm and recognizes customer movement between levels of loyalty in response to marketing efforts. 

The researchers observed alumni donations at Stanford University over a period of 27 years, identifying three levels of loyalty based on the frequency of donations: dormant, those who never donate; intermediate, those who donate infrequently; and active, those who donate frequently. 

Universities focus the majority of their marketing efforts, such as reunions and e-mails, on active alumni. The researchers’ results showed that active alumni are likely to donate frequently even when their attendance at events or frequency of contact from the school is low, suggesting that increased efforts to cultivate this group would not lead to additional gains. Alumni in the intermediate state are more likely to donate after an interaction with the university — when interactions increase, so do their frequency of donations. 

By capturing dynamic segmentation, firms can avoid misdirecting marketing efforts and can, instead, target the optimal customer — one whose loyalty can increase with targeted marketing efforts. 



Practical Applications

Marketing managers

You can use this enhanced CRM tool to better understand your customers’ level of loyalty over the lifetime of their relationship with your firm. This allows you to determine your customers' expected loyalty dynamics and assess which customers have the most potential to increase their loyalty and with which marketing efforts they should be targeted. 


Oded Netzer is associate professor of marketing at Columbia Business School.

Oded Netzer

Professor Netzer's research centers on one of the major business challenges of the data-rich environment of the 21st century: developing quantitative methods that leverage data to gain a deeper understanding of customer behavior and guide firms' decisions. He focuses primarily on building statistical and econometric models to measure consumer preferences and understand how customer choices change over time, and across contexts. His research...

View full profile

Read the Research

Oded Netzer, James Lattin, V. Srinivasan

"A Hidden Markov Model of Customer Relationship Dynamics"


Download PDF View abstract/citation