A Balanced View for Customer Segmentation in CRM

Introduction and the Research Problem

Jongwook Yoon, Seok H. Hwang, Dan J. Kim, and Jongsoo Yoon proposed a study titled “A Balanced View for Customer Segmentation in CRM” published in the Ninth Americas Conference on Information Systems in 2003. The authors emphasize that customer segmentation is essential for enterprise customer relationship management (CRM) implementation. However, existing studies have primarily focused on customer segmentation in terms of profitability as the end goal. For addressing this research gap, the research proposes a balanced view for customer segmentation. In addition, the study was conducted using extensive data with demographic and transactional information. Furthermore, the authors state that the monetary measures may include customer satisfaction and loyalty as critical indicators of customers’ expected value, which can be essential for customer segmentation. Besides, the authors suggest that balanced segmentation ensures that customers stay loyal to a brand for a more extended period.

In recent years, companies have increasingly invested in CRM, but the success rate is not satisfactory. According to the study’s findings, about 35% of the CRM applications are considered a success, and the estimated failure rate stands at 55%, which is on the higher side. The authors suggest that the CRM applications are continually implemented without carefully evaluating the critical components in the CRM lifecycles as the segmentation views are currently specific to the profitability aspects of the companies. As a result, the typical concern observed with CRM implementation is the lack of emphasis on customer retention during segmentation procedures and strategies.  In addition, the study reflects on the unclear paradigm of CRM as there are different definitions concerning academic and industrial fields.

Research Problem Solution

First, the study summarizes the companies’ key terms, such as expected value or profit (EVCO) and the customers’ expected value or profit (EVCU). The findings disclosed an imbalanced approach that includes more EV (CO) utilization, and only limited research has addressed EV (CU) considerations for segmenting customers. Therefore, the study proposed an approach beneficial for customers and companies to make the segmentation balanced. Then the authors divide the approach into three ways to specify that marriage is an integral status for balanced segmentation, and the worst-case is the parting where the customer opts for another company. Furthermore, one-sided is the scenario that deteriorates the relationship and may lead to parting. Finally, to collectively measure the success rate, the study implements both EV (CO) and EV (CU) methods for evaluation (see Figure 1).


The figure above illustrates that customer segmentation has been performed frequently by EV (CO) measure, whereas EV (CU) is primarily non-existent. Another critical point is that very few industries possess the potential to implement EV (CU) as a monetary measure. However, using EV (CU) ensures competitive advantage and offers customer insights regarding transactional data and demographic information. Therefore, the reason for conducting the review is to pinpoint that an appropriate ratio should include EV (CU) to obtain customer satisfaction and loyalty information. Moreover, the customer segmentation is fair with a well-balanced ratio involving EV (CO) and EV (CU) to generalize the market into well-tiered customer segments.

Furthermore, the experimental work of the research model involves a well-refined data mining analysis which comprises 10,243 transactional records of the customers from a “D” Security company. The preprocessing step consists of organizing the transactional records according to individual customers, including the variables required for the cluster analysis for customer segmentation. On the one hand, the missing values within the 27 variables were excluded from the experimentation. On the other hand, 16 variables inclusive of the RFM index were used as input variables. To perform the demographic cluster analysis, these variables were divided into primary and auxiliary variables (See Figure 2).

Figure 2: Input Variables for Customer Segmentation

The result of the analysis provided eight clusters which were assigned after careful evaluation of the attributes and characteristics of each cluster. Moreover, the cluster analysis helps identify loyal customers measured using EV (CO) and EV (CU). For the EV (CO) measure RFM index has been implemented since it is widely used in the industries and is capable of generating simple and effective results. However, the average price-earnings ratio was obtained to identify customers’ monetary value with a continuous relationship with the company. This factor was found to be an effective measure of EV (CU) (See Figure 3).


Figure 3: Customer Classification using EVCO and EVCU

In Figure 3, cell –I indicate that the percentage of loyal customers is about 2.13%, which belongs to the “D” company, reflecting that security service companies have challenges differentiating their services from competitors. Therefore, it is essential to develop marketing strategies to convince and retain customers belonging to cell-II and cell- IV to move into the cell-I category.

Findings of the Research

The study emphasizes the need to consider the company and the customer perspectives to achieve a successful customer segmentation method. Since customer acquisition is more expensive and difficult to achieve; therefore, customer retention relationship management is increasingly adopted. However, the traditional segmentation methods are criticized as all customers are not equal. Similarly, new approaches have targeted heavy users as an essential component for obtaining more sales. Therefore, this theory introduced a new element known as corporate profit from the customer’s continuous relationship into existing customer segmentation approaches. Additionally, the study suggests that CRM is applied for customer retention as a defensive marketing strategy to identify customer preferences in the existing customer base. On the other hand, the research provides critical findings that CRM should treat every customer differently as all customers do not have equivalent value.

Based on this theory, several studies have implemented this premise to measure customer value, but the authors suggest it is only a fundamental level of valuation. In addition, profitable customers are handled with an exceptional approach to opt for a product or service. However, it is defined that a mere profitability perspective is not sufficient for customer segmentation.

Furthermore, the authors suggest that the expected value indicates customer satisfaction and loyalty, which should be integrated as a monetary measure and as input for customer segmentation approaches. While customer acquisition and customer retention have remained crucial factors in customer segmentation, the study indicates that modern customer segmentation approaches focus more on customer retention. In addition, to achieve highly profitable customers that are frequently targeted with advertising, promotion, sales calls, and other communication approaches, the rationale for CRM for such groups can be achieved in terms of two groups, namely customer retention and differential customer value. Besides, the Pareto principle or an 80/20 rule is indicated as a practical approach for segmentation.

Although there are barriers that prevent EV (CO) and EV (CU) measures from being implemented for segmentation, the author state that both measures should be implemented in balance for effective customer segmentation; however, the study summarizes the reasons why EV (CU) is isolated despite being an effective measure. Furthermore, the authors specify that the primary reason is current measures like customer satisfaction and customer loyalty not falling into the tangible categories. As a result, an incomplete variable set with only limited variables such as corporate revenue will lead to segmentation failure. In addition, the absence of mutually exclusive factors in the segmentation stage results in ambiguous segmentation results that are not purposeful in practice. Besides, there are several additional factors like under-skilled personnel in the segmentation process. These problems lead to the low success rate of segmentation efforts of the companies. However, in terms of customer segmentation, the authors highlight that the telecommunication and banking industries are ahead of other industries. This is because they follow a more straightforward process of capturing customers’ transactional and demographic data and providing crucial customer insights.

The benefit of a balanced view for customer segmentation provides company and customer perspective in the segmentation step, which forms a mutually exclusive and collective set for analysis. The experimental analysis emphasizes that balanced segmentation efforts can increase corporate profit and long-term customer relationships leading to customer loyalty in the future. At the same time, the selection of more profit-generating customers can be identified as it is based on customer loyalty. Additionally, the multilateral viewpoint of the study allows companies to understand the customer perspective and appeal to a profit-producing customer base efficiently.

Future Work Suggestions and Implications for Practitioners

The rapid increase in CRM investments is evident, but a single-dimensional approach towards customer segmentation with profitability as the focal point is disrupting the effectiveness of CRM implementations. Thus, future research must focus on equal considerations of company and customer perspectives. During the CRM lifecycle, it is crucial to treat customers with differential values to the company. Maintaining the EV (CO) and EV (CU) ratio is an approach commonly observed in banking or telecommunications. However, this balance provides proper customer segmentation and insights. Therefore, including the EV (CU) and EV (CO) measures, including multivariate statistical methods such as cluster analysis and discriminative analysis, can help develop a generalizable method to make well-tiered customer groups.