In this section, we present some previous works related to customer life time value. We also present some backgrounds that we utilized in our research, including RFM method, Naïve Bayesian classification, potential value, and market segmentation.
2.1 The Definition of LTV
Customer lifetime value is commonly used to identify profitable customer and to develop strategies to target customers (Irvin,1994). Different methods have been used to find out customer’s value to the company, and these were done based on customer lifetime value, customer equity, and customer profitability etc. Many of the researches focused on the Present Value of customers over the lifetime of transactions, through observing the behavior of their customer (Gupta & Lehmann, 2003).Berger and Nasr(1998) defined LTV as the net profit or loss to the firm from a customer over the entire life of transactions of the customer.
Pearson(1996) stated LTV as the net present value of the stream of contributions to profit that result from customer transactions and contacts with the company. Gupta and Lehmann(2003) specified LTV as the present value of all future profits generated from a customer. LTV can also be considered as the sum of revenues gained from company’s customers over the lifetime of transactions after the deduction of the total cost of attracting, selling, and servicing customers, take into account the time value of money( Hwang, Jung,Suh, 2004).
2.2 RFM method
RFM is an important method in measuring customer’s value to the company. The
purpose of RFM scoring is to project future behavior (driving better segmentation decision).
In order to allow projection, it is important to translate the customer behavior into numbers which can be used through time.( Miglautsch,2000)
Bult and Wansbeek defined the terms as: (1) R (Recency): period since the last purchase;
a lower value corresponds to a higher probability of the customer’s making a repeat purchase;
(2) F (Frequency): number of purchases made within a certain period; higher frequency indicates greater loyalty; (3) M (Monetary): the money spent during a certain period; a higher value indicates that the company should focus more on that customer.
Several studies have discussed about the approaches and advantages of RFM.
Goodman(1992) suggested that the RFM method avoided focusing on less profitable customers, allowing resources to be diverted to more profitable customers. Hughes (1994) proposed a method for RFM scoring that involved using RFM data concerning to sort individuals into five customer groups. Different marketing strategies could then be adopted for different customers. Stone (1995) suggested that different weights should be assigned to RFM variables depending on the characteristics of the industry.
Miglautsch(2000) ranked its customers based on their purchase behavior. The proposed method specifically turns the three vectors (R,F,M) separately into five segments each. For instance, the R value is classified by 1-3 months, 4-6 months, 7-12 months, 13-24 months, and 25 months and up, which are assigned with values of 5, 4, 3, 2, 1. After obtaining the values in the RFM, the values are added up, and if the RFM value of a certain customer is greater than the average value, the customer is considered to be an important customer.
The RFM model is frequently applied in the Customer Relationship Management of
could perform further analysis to learn about the cause and decide on the appropriate actions.
2.2.1 Weighted-RFM method
There is an advantage to finding a single variable for the RFM scores: they can be easily segmented or queried in a single field in the relational databases. In the book Libey on RFM by Donal R.Libey, the method of adding up Recency, and Frequency, and Monetary scores was suggested. Different weights are applied accordingly to the relative importance of R,F,M of each company. Liu, Shih, 2004, employed the AHP to evaluate the weighting of each RFM variable, and specifically asks decision-makers to make intuitive judgments about ranking ordering to make pair wise comparisons. The RFM values of each customer are normalized.
The normalized RFM values of each customer are then multiplied by the relative importance of RFM variables, WR, WF and WM, which are determined by the AHP.
2.3 Naïve Bayesian Classifications
Naïve Bayesian Classification is a simple and statistical classifier which can predict the probability that a given sample belongs to a particular class. This classification method is based on Baye’s Theorem, a statement of the definite relationship between an event A and B even with the fact that the probability of the event A conditional on event B is generally different from the probability of B conditional on A.
The naïve method assumes the concept of class conditional independence, that the effect of an attribute value on a given class is independent of the values of the other attributes.
Naïve Bayesian classifiers are, in many cases, considered to have the minimum error rate in comparison to other classifiers, including decision tree and neural network classifiers.(Han, Kamber,2001).
2.4 Potential Value
In order to have an effective CRM, it is essential to have information on the potential value of customers. Although it is difficult to have a precise evaluation of future outcomes, we have performed an estimation based on the probability of customers’ past historical data.
Reducing the uncertainty of the future is always a perceived benefit to the company.
Furthermore, based on the information of potential value and realized value, managers can devise customer specific strategies.
The potential value of a customer refers to the profitability of a customer if that customer buys all purchased products or services from the supplier. By performing an analysis on customers’ historical records, companies can realize their potential clients, and invest in (potentially) valuable customers. At the same time, they can minimize their investments in non-valuable customers. Their profitability is often used as segmentation variables to distinguish between valuable and non-valuable customers (Verhoef and Bas Donkers, 2001). In the works of Kim, Hwang, Jung & Suh, potential value of customers was defined as expected profits obtained from a certain customer when a customer uses the additional services of a wireless communication company. Past researches have applied this concept in telecommunication companies. In the research, we applied this concept in the field of retail markets, shopping mall in particular.
2.5 Market Segmentation
Several clustering techniques have been applied to segment markets, including K-means, hierarchical, fuzzy, and c-means etc. K-means clustering is the one of the most well-known method (J.B. McQueens,1967) to partition data sets into different groups. This algorithm takes the input parameter k, and partitions the set of n objects into k clusters, such that the distance from data sample di to cluster kj is the minimum in all k clusters. Cluster similarity is measured in regard to the mean value of the objects in a clusters, also viewed as the “center”
of the cluster kj.