ArticlePDF Available

Bank Customer Classification based on Recency, Frequency, Monetary (RFM)

Authors:

Abstract and Figures

Customer relationship management improves the institutes’ interaction with customers. Identification of customers’ characteristics and optimal allocation of resources to them have been mentioned as the major concerns at the area of customer relationship management. The present research intends to propose a model for customer classification. In the proposed process of this research which was performed in one of Banks in Iran , 32211 customers were segmented to five clusters using “K-Means” and “Two Step”. The results from this research paved the way for analysis of characteristics of bank customers and better interaction with them. Keywords: CRM, Customer Classification, RFM model, Two Step algorithm, K-means algorithm
Content may be subject to copyright.
International Research Journal of Finance and Economics
ISSN 1450-2887 Issue 149 July, 2016
http://www.internationalresearchjournaloffinanceandeconomics.com
Bank Customer Classification based on
Recency, Frequency, Monetary (RFM)
Mirza-Hassan Hosseini
Professor, Department of Management, Economics and Accounting
Payam e Noor University, Po. Box: 19395-3697, Tehran, Iran
Yaser Pourdavar
PHD. Candidate for Business Administration at Payam e Noor University
Po. Box: 19395-3697, Tehran, Iran
E-mail: yaser.pourdavar@gmail.com
Abstract
Customer relationship management improves the institutes’ interaction with
customers. Identification of customers’ characteristics and optimal allocation of resources
to them have been mentioned as the major concerns at the area of customer relationship
management. The present research intends to propose a model for customer classification.
In the proposed process of this research which was performed in one of Banks in Iran
1
,
32211 customers were segmented to five clusters using “K-Means” and “Two Step”. The
results from this research paved the way for analysis of characteristics of bank customers
and better interaction with them.
Keywords: CRM, Customer Classification, RFM model, Two Step algorithm, K-means
algorithm
Introduction
Customer relationship management refers to an infrastructure that reveals and increases the customer
value. To have an effective customer relationship management, data collection in the context of
customer value and classification to meet customers’ needs is required. In banking industry, the factors
such as supply of services in form of e-banking and rising of competition among banks due to
increasing their numbers have increased the probability for exclusion of valuable customers. On the
other hand, banking industry deals with numerous customers which recognition of them one by one is
not possible, thus their classification to homogeneous groups and overview of characteristics of each
group to propose a specific marketing plan will be a suitable approach to meet customers’ needs. Yet,
this classification should be made regarding unique conditions and characteristics of groups and factors
affecting banking
2
.
Problem Statement
In industrial countries, prioritizations of customers and customer relationship management have
intertwined with banking industry. Yet, in the countries worldwide, customer maintenance and
1
. Tosee Taavon Bank (TT Bank)
2
. Gholamian MR, Niknami Z, providing a model adapted for segmenting customers based on their lifetime value”,
JOURNAL OF EXECUTIVE MANAGEMENT, 1391 (2012)- Issue 7
International Research Journal of Finance and Economics - Issue 149 (2016) 34
exploration into their behavior and needs have been considered as unresolved problem in the banking
sector. Prioritization of customers and arrangement of suitable customer relationship management
system are required for maintenance of current customers and increase of new customers in banks so
that discovery of knowledge from databases is a suitable tool to resolve this problem.
If it is looked into the current status of banks from customers’ Insight, it can be observed that
most of banks don’t have suitable design for providing services for the customers. Providing services
for the customers has a little speed without strategic desire to resolve problems pertaining to the
customers which this is due to the bureaucratic and cumbersome rules in banks. Some banks cannot
provide a suitable prediction for the period of time that the customers must spend to receive any
service. Further, banks do not differentiate customers in terms of creation of a suitable product or
service
3
.
Accordingly, design model of services in the large number of banks is based on the model of
"Mass production", thus their attempt is made to meet needs of all the customers with one type of
product or service. Yet, the global approach has changed to “Customized production” with the content
of design of custom services based on customers’ various needs.
A large number of customers have bad feeling about this point that their feelings and needs
have not been drawn into attention by the banks. Bank staffs often do not understand customers’ needs.
From perspective of innovation in information technology (IT), banks have faced this problem that
they have sought trappings of modernity, yet trappings of modernity do not raise modernity. For
instance, they have used information technology to increase security and auditing accuracy, while
information technology is a strategic tool that must be used to meet customers’ needs and create value
added in a long term. Therefore, most of customers are satisfied with their connection to bank,
resulting in reduction of their loyalty to bank.
Research Objective
The present research intends to identify the scientific basis and experiences at the area of bank
customer classification and conduct the proposed research plan in form of a project so as to assist for
improving quality of services provided for the customers well suited to their importance and value
extent.
Literature Review: Definitions, Models and Theories
Definition for Customer Relationship Management
Customer relationship management can be defined as follow:
Customer Relationship Management (CRM) refers to a strategy in the bank area which is used
to optimize income, profitability, and customer satisfaction through organizing process based on
different groups of customers, expansion of satisfactory behavior and relationship of processes from
customers to suppliers. Investment in CRM causes better understanding, more access and more
effective interaction with customer through various channels.
CRM refers to a business process which addresses all aspects of customer characteristics,
creates the customer knowledge, forms customer relationship and creates their impression from bank
products or services
4
.
CRM as a Strategy
CRM refers to a technique, system or most importantly a strategy in business which intends to classify
and manage the customers so as to optimize the customer value in a long term. Indeed, CRM appears
3
Amir, Al-Badawi et al, “use of data mining technique to implement customer relationship management in the
banking industry of Iran, Journal of Business Review, 1384 (2005), Issue 14
4
. Amir , Al-Badawi, et al
35 International Research Journal of Finance and Economics - Issue 149 (2016)
in finding customer, approaching to customer, managing and creating satisfaction in the customer.
These processes entitled “customer life cycle” have been elaborated in this way: acquiring customers,
increasing the customers’ value and maintenance of good customers. From strategic perspective, it
should create specific banking products or services for each group of customers; supply of pre-
designed products is not proper for all the customers. To make suitable relationship with customer,
bank requires recognizing all the customers and giving answer to the questions below about each
customer:
1. Whether the customer is profitable?
2. Why the customer engages in this business with bank?
3. What the customer loves about bank?
4. Whether the customer engages in this business with the competitors of bank?
Significance of CRM to bank is in its assistance to respond to the above questions and customer
interactions management
5
.
Strategic Selection: Acquiring New Customer or Current Customer Maintenance
Studies have shown that some institutes only with 5% further effort in maintenance of existing
customers have increased their profit to 100%. The secret behind this has lied in repetition on referrals
of current customers. Further, acquiring new customer is more precious than maintenance of existing
customer. On the other hand, acquiring customers’ satisfaction needs huge investment and cost which
is not sometimes economically effective. Thus, the important point to the banks has lied on this fact
that which customers enjoy suitable capabilities
6
.
3. Customer Classification Models
1. Boston Consulting Group Method: BCGC Model” for customer classification
This model was first proposed by Boston Consulting Group to analyze and evaluate status of
investments aiming to detect activity range of SBUs, to provide strategic plan and allocate required
credits to these units based on strategic plan. The leading institute constantly measures conditions
undergoing these units so as to make decision about growth, maintenance or depletion of their
lifespan
7
.
In Boston Consulting Group method, a matrix has been considered which has two dimensions
including “Market growth rate” and “Relative market share”. With regard to the SBUs status in this
matrix, decisions are made about their future:
1. Question marks: The units in this sector are young which bank must invest in to increase
their relative share than the competitors so that this might be expensive.
2. Stars: These units require much investment to face the competitors. If a particular attention
is paid to these units, they will turn to cash cows and gain profit.
3. Cash cows: These units bring much money into the bank which are profitable due to the
superiority to the same competitors’ units. To develop these markets, there is no need to
invest because their market has been saturated.
4. Dogs: These units have little profit. Bank must make attempt to cancel these units or sell
them unless there is a convincing reason to hold them.
If we seek to use BCGC model to prioritize customers, this matrix must change to figure below.
5
Amir , Al-Badawi, et al
6
. Amir , Al-Badawi, et al
7
Amin, Fereshte et al, Increasing the efficiency of banking services by prioritizing customers using quantitative
techniques”, Research and Economic Policies, 1384 (2006) - Issue 36
International Research Journal of Finance and Economics - Issue 149 (2016) 36
Table 1: Customer prioritization matrix
Stars
(Light: accelerate)
Question marks
(Ambiguity : use caution)
Cash cows
(Good: use caution)
Dogs
(Risk: keep distance)
1. Question marks: This includes the clients who have made relationship with bank more
recently and have undergone the early screenings by bank, but bank has not set any specific
status for them. In other words, bank is at ambiguous status and does not know whether the
clients will be troublesome or might be profitable?
2. Stars: if the customers who are at the stage of question mark exclude from this stage with
success under supervision of bank, they will enter into the stage of stars. These customers
can gain profit for the bank. The difference between this group of customers and the
customers at the stage of question mark lies on this fact that the volume of transactions with
bank increases; in other words, the condition to move from stage of question mark to stage
of star lies on increase of transactions with bank or increase of extent of bank’s interest
received from these customers than rest of customers. Bank must invest for acquisition of
them and detect their needs to avoid them inducing to the competitor banks and transform
them to loyal customers.
3. Cash cows: there is a group of customers who enter much money into bank mentioned as
old bank customers with high loyalty level to bank. This group of customers has no need to
displace the bank of which they receive service. They give priority to quality of
services(speed, accuracy and timeliness) instead of new services.
4. Dogs: the customers at the category of dogs are the low profit customers. These customers
have high expectations and waste bank energy but they will have no value added to bank.
They try to make win-lose relationship with bank and strive to use bank services at least
price as much as possible. If the clients who are at question mark category have no benefit
for the bank or the clients who are at stars category strive to make relationship with bank
only aiming at utilization of bank services and make no value added for bank, they will be
gradually induced to this stage.
To gain maximum value added, the bank should organize the service orientation based on
customers’ position.
2. The Customer Pyramid
The Customer Pyramid (see Figure 1) is another conceptual customer classification and has 4
levels
1. The Platinum Tier: describes the company's most profitable customers, typically those
who are heavy users of the bank products and services, with low Maintenance Cost, and
retold positive advertising of bank products by word of mouth.
2. The Gold Tier: differs from the Platinum Tier in that profitability levels are not as high,
perhaps because the customers want price discounts that limit margins. They might not
be as loyal to the firm even though they are heavy users in the product category they
might minimize risk by working with multiple banks rather than just one bank.
3. The Iron Tier: contains customers that provide the volume needed to utilize the bank’s
capacity but whose spending levels, loyalty, and profitability are not substantial enough
for special treatment.
4. The Lead Tier: consists of customers that are costing the bank money. They demand
more attention and their profitability is low, and their banking services usage is low, and
they have no loyalty and satisfaction about bank services. They are sometimes problem
customers; complain about the firm to others and waste the bank’s resources.
37 International Research Journal of Finance and Economics - Issue 149 (2016)
Customer pyramid model is an one-dimensional model while the BCGC model is a two-
dimensional model that only in one dimension, ie profitability is similar to customer pyramid model
8
.
Figure 1: The Customer Pyramid diagram
3. Customer Classification Based on Loyalty
Customer Loyalty to bank is another important aspect about customer which enjoys high
complicatedness. The two-dimensional map below indicates that how combination of the strategies
based on product (service) and customer affects customer loyalty.
1. In this diagram, customer (A) is satisfied with his/her purchase, but his relationship with
bank is not satisfactory. This customer is vulnerable. If the bank enables to improve its
relationship with customer, the existing problems will diminish. Customer/bank
relationship affects customer’s behavior to bank. By proper management and suitable
service, this status will reach to the loyalty.
2. Customer (B) refers to a customer that any bank wants it. He /She is satisfied by bank
products and his/her relationship with bank.
3. Customer (C) is a bank nightmare. He /She has had an improper product (service)
purchase experience from bank, and no proper relationship has been made by bank. In this
case it can be guaranteed that this customer will not purchase from this bank again.
4. Customer (D) is not satisfied with the product, but he /she hopes a satisfactory interaction
in next time. A good relationship causes these customers to hope receiving required
quality in next transactions.
8
. Amin ,Fereshte et al
International Research Journal of Finance and Economics - Issue 149 (2016) 38
Figure 2: Customer classification based on loyalty
Figure above indicate that how it can perceive customer better and increase his satisfaction. As
shown in diagram above, if customers (C) and (D) do not receive a proper product from bank,
customer will not lose his hope to receive better product from bank. CRM indicates that how a bank
can transform the customer’s unpleasant experience to loyalty. In doing so, the bank requires
coordination of activities
9
.
4.
Customer
classification based on the satisfaction level
In a general outlook, there are two groups of customers: external customers who are out of
bank and purchase product (service) and internal customers who are bank staffs. Internal customers are
considered as the next persons in the working process. In this regards, all the working activities in the
bank can be defined as a series of relationships between staffs or internal customers and internal
suppliers. Thus external customers are those who use the bank services and internal customers are
those who receive work results of others which this might include report, order or services. Bank must
put an emphasis on both types of internal and external customers and make attempt to meet their needs.
To be informed of customer’s opinion on the services received from bank and estimate his/her
satisfaction, a scale has been proposed. This scale divides the extent of customer’s satisfaction and
dissatisfaction into the categories as follow. The scale ranges from -2 to +2 proposed as follow:
Figure 3: Rating customers’ satisfaction with products/services
9
. Youssefi P. and Syed Javadein, S. R., “Factors affecting the loyalty of Export Development Bank of Iran customers
using the concepts of Customer Relationship Management”, Business Review magazine, Summer 1385 - Issue 18
39 International Research Journal of Finance and Economics - Issue 149 (2016)
1. Satisfied customer (0): Satisfied customers refer to the customers who do not abandon bank
but they can abandon it. Customers’ satisfaction with bank products (services) is the lowest
level of goodness feeling. Bank must give priority to this group of customers in order that they
remain as bank customers.
2. Dissatisfied customer (-1): Dissatisfied customer is seeking proactively other resources to
replace the services received from bank.
3. Angry customer (-2): Angry customer seeks destruction of bank and wants to damage it.
4. Happy customer (1): Happy customers are customers who have high expectation from bank
and seek their expectations to be met.
5. Intrigued customer (2): This group of customers are fan of bank knowing the bank as their
interest.
With regard to the theoretical rating of customers, a bank should attract satisfied customers by
supply of better products (services) and seek the reason for some of customers’ dissatisfaction with
bank activities and then analyze their reasons and make attempt to resolve them. Ultimately, the reason
for customers’ satisfaction should be known in the factors such as adoption of suitable methods,
technology factors, personal characteristics of staffs and managers, social relations and customers’
need
10
.
5. Customer classification based on Customer Lifetime Value (CLV)
Using tools entitled “customer classification” causes the bank put a huge emphasis on
detection, acquiring and maintenance of more profitable customers. Customer lifetime value as a
criterion to evaluate customers can be a suitable framework.
A variety of methods to calculate customer lifetime value
Some researchers have defined Customer Lifetime Value (CLV) as customer benefits minus customer
costs. A variety of methods (models) to calculate CLV include “RFM, probability, econometrics,
sustainable, computerized”. With regard to the subject of the present research, it is sufficed to
introduce RFM model.
Research Methodology
RFM Model
RFM model refers to the most prominent model and effective method to determine Customer Lifetime
Value. In this research, RFM classification model has been used which is a model to classify the
customers. This model has three variables which includes:
1. Recency (R): It’s the time from the last customer transaction with bank.
2. Frequency (F): Number of transactions in a given period. In this research, transaction number
of customer’s accounts in the first six months of the year 1393 (2014) is known as Frequency.
3. Monetary (M): It refers to the amount of customer transactions during a specified period.
In this research, it is the sum of amounts in customer's accounts at the end of summer in 1393 (2014).
Table 1: Definition of research variables
Variable Definition
Customer Id Customer Number- It’s an unique number that identifies each customer.
M Customer's accounts balance at the end of given period
F Number of transactions during given period
R Number of days passed from the last customer’s transaction
10
. Memarzade, GR et al, “Customer Relationship Management (CRM) methodology”, Journal controller, autumn and
winter 1387 (2008) - Issue 22
International Research Journal of Finance and Economics - Issue 149 (2016) 40
Research Data
Considering the research variables in table above, the real data of the customers’ accounts in one of the
banks in the country has been used in the present research. Each of the bank customers has their own
specific Number which is considered customer ID, helping to find the data allocated to that customer.
The collected data relate to the early six months of 1393(2014).
Data Analysis Software
SPSS Software has been used to analyze data of this research.
Data Analysis Algorithm
1. According to this algorithm, at the first step, the transactional data of legal customers was
cleansed. Data cleansing was made to prepare data for extracting knowledge. Format of some
data might be different (because of different reasons), thus data cleansing is necessary at the first
stage.
2. In this step, to specify the number of clusters, the “Two Step” clustering method was used.
3. After determining optimal number of clusters, K-means algorithm was used for customer
classification into homogeneous groups. (In order to get better result, classification was made
twice: by Two Step and K-Means algorithm). It should be noted that there is no rule for optimal
number of clusters in K-means algorithm and it depends on research problem (Clustering
algorithms intend to separate input data into clusters or homogeneous groups in order to
maximize similarity of records inside the clusters and minimum similarity of records out of the
clusters).
4. Since some of collected data are large, the results from clustering are influenced by them, so that
a customer might be in one cluster and rest of them might be in other clusters, whereby a precise
clustering might not be made. Therefore, at this stage, large data are excluded from the records
Transactional data
Data cleansing
Two Step
Clustering
K-Means
Clustering
Efficiency of clusters
regarding the
research problem
41 International Research Journal of Finance and Economics - Issue 149 (2016)
and rest of data is entered into the clustering process, whereby the process might be repeated
several times so as to obtain more homogenous results.
Classification Results
On the whole, there are 32211 legal customers in bank divided into five clusters. Table below displays
the result from clustering.
Cluster
(Class)
Number of
customers
Percent
(%)
Account balance Intervals
(Million Rials )
11
Cluster Average
Low bound Up bound M (Million
Rials)
F (Number of
transactions) R (Day)
2 3,153 117,000 3,000,000 16,000 0.2 89 Cluster 1
6 893 6,000 16,000 3,000 0.8 263 Cluster 2
9 774 1,600 3,000 900 2 717 Cluster 3
14 129 297 900 100 11 3,479 Cluster 4
47 32 10 100 0 86 27,663 Cluster 5
- - - - - 100 32,211 Sum
Features of Clusters
1. Cluster (1): This cluster included 89 legal customers who had on average 117 billion Rials
account balance at the end of sixth month and had 3,153 transactions with bank and the last
transaction of this group was 2 days before the end of period. The account balance of this group
of customers ranged from 16,000 to 3,000,000 million Rials.
2. Cluster (2): This cluster included 263 customers; and on average their account balance was 6
billion Rials; they had 893 bank transactions and the last transaction of this group was 6 days
before the end of period. The account balance of this group of customers ranged from 3,000 to
16,000 million Rials.
3. Cluster (3): This cluster included 717 customers; and on average their account balance was 1.6
billion Rials; they had 774 bank transactions and the last transaction of this group was 9 days
before the end of period. The account balance of this group of customers ranged from 900 to
3,000 million Rials.
4. Cluster (4): This cluster included 3,479 customers; and on average their account balance was
297 million Rials; they had 129 bank transactions and the last transaction of this group was 14
days before the end of period. The account balance of this group of customers ranged from 100
to 900 million Rials.
5. Cluster (5): This cluster included 27,663 customers; and on average their account balance was
10 million Rials; they had 32 bank transactions and the last transaction of this group was 47
days before the end of period. The account balance of this group of customers ranged from 0 to
100 million Rials.
Conclusion
The present research elaborates the position and significance of the customer classification to pave the
road to success of bank as well as the common models in this field. Further, using RFM model and
common classifying algorithms (Two Step, K-Means), the project of customer classification was
performed during early six months of the year 1393 (2014).
1- The classification results indicate that huge number of customers belongs to 4
th
and 5
th
classes
or categories (about 97%). According to RFM factors, this group of customers has low loyalty
to the bank.
11
International Research Journal of Finance and Economics - Issue 149 (2016) 42
2- Since maintenance of old customers has lower cost than acquiring new customers, it is better to
give priority to representing strategies to develop and maintain customers loyalty to the bank.
3- Since a substantial number of customers are inactive, representing promotional strategies to
activate and develop this group of customers has great importance.
References
[1] Gholamian MR, Niknami Z, providing a model adapted for segmenting customers based
on their lifetime value”, Journal of Executive Management, 1391 (2012)- Issue 7
[2] Amir, Al-Badawi et al, use of data mining technique to implement customer relationship
management in the banking industry of Iran, Journal of Business Review, 1384 (2005),
Issue 14
[3] Amin, Fereshte et al, Increasing the efficiency of banking services by prioritizing
customers using quantitative techniques”, Research and Economic Policies, 1384 (2006) -
Issue 36
[4] Youssefi P. and Syed Javadein, S. R., “Factors affecting the loyalty of Export Development
Bank of Iran customers using the concepts of Customer Relationship Management”,
Business Review magazine, Summer 1385 - Issue 18
[5] Memarzade, GR et al, “Customer Relationship Management (CRM) methodology”, Journal
controller, autumn and winter 1387 (2008) - Issue 22
ResearchGate has not been able to resolve any citations for this publication.
Chapter
This chapter discusses customer relationship management (CRM) as a customer-focused business strategy enhanced by technology that automates and enhances business processes to proactively manage profitable and long-term customer relationships. CRM solutions span a continuum of implementations from a narrow tactical implementation of a specific technical solution to a broad strategic implementation of a customer centric solution. Furthermore, the authors hope that understanding the underlying assumptions and theoretical constructs through the use of CRM will not only inform researchers of a better CRM design for studying e-commerce and Internet marketing, but also assist in the understanding of intricate relationships between different factors.
providing a model adapted for segmenting customers based on their lifetime value
  • M R Gholamian
  • Z Niknami
Gholamian MR, Niknami Z, "providing a model adapted for segmenting customers based on their lifetime value", Journal of Executive Management, 1391 (2012)-Issue 7
use of data mining technique to implement customer relationship management in the banking industry of Iran
  • Al-Badawi Amir
Amir, Al-Badawi et al, "use of data mining technique to implement customer relationship management in the banking industry of Iran, Journal of Business Review, 1384 (2005), Issue 14
Increasing the efficiency of banking services by prioritizing customers using quantitative techniques
  • Fereshte Amin
Amin, Fereshte et al, "Increasing the efficiency of banking services by prioritizing customers using quantitative techniques", Research and Economic Policies, 1384 (2006) -Issue 36
Factors affecting the loyalty of Export Development Bank of Iran customers using the concepts of Customer Relationship Management
  • P Youssefi
  • S R Syed Javadein
Youssefi P. and Syed Javadein, S. R., "Factors affecting the loyalty of Export Development Bank of Iran customers using the concepts of Customer Relationship Management", Business Review magazine, Summer 1385 -Issue 18