Industrial Marketing Managem

Managing customer profitability in a competitive market by

continuous data mining

Hsiao-Fan Wang*, Wei-Kuo Hong 1

Department of Industrial Engineering and Engineering Management, National Tsing Hua University, Hsinchu, Taiwan, ROC

Received 8 October 2004; received in revised form 8 June 2005; accepted 9 June 2005

Available online 11 July 2005

Abstract

In a near perfect market, competitive marketing strategies are often adapted dynamically and rapidly. The changes in customer behavior

are resultant in unpredictable customer profitability and cause inefficient and ineffective marketing planning. In this study, by using data

mining techniques, we develop a Customer Profitability Management (CPM) system to achieve marketing goals by leading customers to

migrate along pre-determined and desirable tracks. The proposed system emphasizes a continuous interplay between the active and reactive

monitoring procedures to identify customer shifts. It has been shown to be an effective approach to help a firm calibrate its marketing tactics

with respect to different types of customers in different situations. The proposed mechanism has been applied to a telecom company with

promising results.

D 2005 Elsevier Inc. All rights reserved.

Keywords: Customer lifetime value; Customer profitability management; Customer relationship management; Data mining

1. Introduction

As corporations increasingly come to see customers as

important assets, methods for estimating Customer Lifetime

Value (the CLV model) have been developed as an

important strategic marketing tool. CLV, which appears

elsewhere in management literature as Fcustomer equity_and Fcustomer profitability_ helps firms quantify customer

relationships (Berger & Nasr, 1998), illustrate the profit-

ability of its customers and provides references for the

allocation of marketing resources to customers and market

segments (Mulhern, 1999). However, existing CLV models

still have limits in applicability for three reasons.

First of all, customer behavior is the result of a complex

interaction among factors including the level of marketing

activity, the competitive environment, brand perception, the

influence of new technologies, and individual needs.

0019-8501/$ - see front matter D 2005 Elsevier Inc. All rights reserved.

doi:10.1016/j.indmarman.2005.06.005

* Corresponding author. Tel.: +886 3 5742654; fax: +886 3 5722204.

E-mail addresses: hfwang@ie.nthu.edu.tw (H.-F. Wang),

hwg@cht.com.tw (W.-K. Hong).1 Tel.: +886 2 33436801; fax: +886 2 33433999.

Therefore, current CLV models, which predict purchase

behavior based on past customer spending patterns or

demographic characteristics, are of limited use in predicting

future behavior. In order to extend the basic CLV model and

effectively apply it to a complicated open market, additional

factors must be discussed and considered, including social

effects, competitive effects, economic environment, product

lifecycle, customer lifecycle, and customers’ purchasing

habits, lifestyle, customer satisfaction, price sensitivity and

brand loyalty (Hogan, Lemon, & Rust, 2002; Jacobs,

Johnston, & Kotchetova, 2001; Mulhern, 1999; Stahl,

Matzler, & Hinterhuber, 2003). Without taking account of

these factors, CLV models are of limited use, as noted by

Libai, Narayandas, and Humby (2002) in which a Markov

transition process was proposed to predict the customers’

dynamic behavior. Therefore, research is needed to inves-

tigate which factors determine customer profitability (Stahl

et al., 2003), and which factors determine the distribution of

profitability among consumers (Jain & Singh, 2002; Raaij,

Vernooij, & Triest, 2003). However, in the framework of

conventional CLV models, it would be too complicated to

associate with these factors because of the structural and

ent 35 (2006) 715 – 723

H.-F. Wang, W.-K. Hong / Industrial Marketing Management 35 (2006) 715–723716

data differences between the factors. Aggregating these

factors in order to capture the timing and stochastic nature of

revenue flows is almost impossible in a CLVappraisal (Bell,

Deighton, Reinartz, Rust, & Swartzet, 2002). Moreover, the

probability-based CLV models only guarantee that the

models predict well within the time horizon of the collected

data, but there is no guarantee for forecasting values beyond

that horizon (Bell et al., 2002). For volatile customer

profitability, it is especially difficult for a forecasting model

to predict dramatic upward or downward trends generated

by, for instance, a new product or service provided by a

competitor. Therefore, firms need a more flexible model,

which would not only able to incorporate managers’

judgment calls and other uncertain factors, but also detect

changes in customers’ behavior. As the result, firms would

be able to monitor and calibrate marketing action in

response to unpredictable customers’ behavior.

Second, existing CLV models provide a static estimate

of customer valuation for a given future period to

segment customers into several levels of a firm’s

customer pyramid, such as profitable, less profitable and

unprofitable (Raaij et al., 2003; Zeithaml, Rust, & Lemon,

2001). However, dynamic markets require a more tactical

view towards these measures. The possible directions in

changes, and the possible volatility over customer profit-

ability have been considered as the effective indices to

trace the customer behavior (Dhar & Glazer, 2003; Stahl

et al., 2003). The direction of customer profitability is a

reliable indicator of the customer’s status (defecting,

upgrading or steady), and volatility represents the possible

risk level of a customer’s profitability for a firm.

Therefore, a mechanism to monitor possible directions

and volatility of customer profitability would enable firms

to dynamically adjust marketing activity towards their

targeted customers.

Third, there is a lack of practical discussion about how

to incorporate customer profitability measures into market-

ing planning (Jain & Singh, 2002). To develop a marketing

activity, a firm needs to draw a picture of its customers by

combining customer profitability with customer accessi-

bility, needs and attitudes. Analysis of customer profit-

ability is often used to indicate possible consumption

patterns of the targeted customers. However, this index

may not be sufficient for identifying the customers a firm

truly wishes to acquire or retain through allocating

additional marketing resources. Further information of

customer accessibility and customer attitudes is needed.

Customer accessibility represents the possibility of accept-

ing a marketing package by a potential targeted customer

or segment, and could be estimated through past marketing

or contact records stored in database or given by account/

product managers based on their experiences and judg-

ments. Moreover, customer attitudes, such as customers’

preferences and their levels of satisfaction, can provide the

information for a firm to offer the right marketing

packages to meet customers’ needs.

In summary, simply relying upon the measurement of

CLV to determine Customer Relationship Management

(CRM) success can be misleading because it ignores the

dynamics of customers’ purchasing behavior. Therefore a

more practical approach to Customer Profitability Manage-

ment (CPM) is needed to monitor customers’ shifts and to

calibrate marketing action for improving customer satisfac-

tion and corporate profitability. The result will be a win–

win situation for both firms and their customers, which is

the aim of our study.

Based on the discussion above, this paper will be

organized as follows. After developing the concept of

CPM in Section 2, the framework of CPM will be

established in Section 3, in which monitoring customer

profitability is emphasized. Then, in Section 4, the criteria

with their levels of operations are proposed to facilitate the

implementation. A case study of a Telecom CPM will be

demonstrated in Section 5. Finally, discussion and con-

clusions will be drawn in Section 6.

2. The CPM concept

CPM is a continuous process to trace and develop a

responsive path for obtaining values from customers, as well

as creating values for customers, according to changes in

industrial conditions. A clear path can guide a firm to make

right strategic choices in determining desired marketing

outcomes and allocating limited resources to marketing

initiatives. Making strategic choices in response to socio-

economic changes from among many possible marketing

initiatives is a difficult, yet crucial task for firms. However,

an important principle of strategic choice is to select

marketing initiatives that can actually raise existing or

create new value for customers. Some firms forget this

principle and wind up trapped in destructive price wars.

Although price reductions may temporarily attract custom-

ers, it fails in CPM because products or services without

improvements and differentiation cannot retain profitable

customers in the long run.

Moreover, the purpose of a responsive path is to respond

to unpredictable customer behavior. Although a customer

database might allow a firm to trace changes in the

profitability of a given customer, strategies derived from

data analysis are often too passive to respond to the rapid

changes in the market. This study proposes an alternative

approach which effectively monitors shifts in customer

profitability to help a firm win back defecting customers,

detect potential customers, and evaluate the performance of

upgrading customers. Such an approach takes a holistic

view of a firm’s marketing efforts, in which the strategic,

tactical and operational efficiencies of Customer Profit-

ability Management (CPM), along with efficient manage-

ment of tangible and intangible assets allow firms to

continually identify value-creating opportunities for increas-

ing customer profitability.

H.-F. Wang, W.-K. Hong / Industrial Marketing Management 35 (2006) 715–723 717

3. The proposed CPM system

In this section, the conceptual framework of Customer

Asset Management (CAM) (Berger, Kumar, Parasuraman,

& Terry, 2002) will be extended to a practical CPM system,

in which a firm’s profit and the value of its customers can be

increased through tailoring its interactions with targeted

customers. Based on Fig. 1, we may observe that the role of

strategic CPM is to reach the firm’s long-term marketing

goals, and that of tactical CPM is to develop applicable

plans with resource allocation based on these goals. The

role of operational CPM is to implement marketing

activities and monitor their results to handle real customer

shifts and exceptional events, such as fault detection. These

three levels of CPM implementation correspond to three

managerial routes to form the CPM framework for sustain-

ing and maximizing customer profitability. A detailed

discussion follows.

3.1. The leading CPM route

The core issue of a strategic plan is determining the

firm’s long-term policy for managing customer relation-

ships to increase overall profits. Based on scenario analyses

and market forecasts, a leading CPM route is formed to lead

strategic planning to determine the marketing mix from a

firm’s resources. To optimize resource utilization, a firm has

to provide the right products and/or services to potentially

high profit customers. The targeted customers are filtered by

customer segmentation, which divides customers into

mutually exclusive groups (Peltier & Schribrowsky, 1997).

Therefore, the needs, behavior and preferences of each

group of customers can be identified to develop appropriate

MarketingPlanning/

Goal

MaFor

TargetedCustomers

MarketingProjects

Leading CPM Route Active CPM Route

CustomerPrediction

MarketingMix

CustomerOutcome

MarketingSchemes

MarketingActivities

MonitoringCustomer

Profitability

CustomerSegmentation

Fig. 1. Customer profitab

marketing activities to maximize customer outcomes. A

CPM database includes demographic, geographic and

psychographic on customers, as well as data on their needs,

attitudes, and purchasing behavior. Armed with this multi-

dimensional profile, a firm not only can improve its

marketing strategies, but also can create different ways to

satisfy different customer groups.

3.2. The active CPM route

The role of the active CPM route is to initiate marketing

action to migrate anticipated customers based on a firm’s

strategic analysis-based policies. Data mining analysis

reveals tendencies for defection or growth among the

targeted customer group, providing a basis for developing

active marketing projects and activities. Comparative

analysis then stores the resultant customers’ outcomes to a

CPM database for further use.

3.3. The reactive CPM route

The role of the reactive CPM route is to analyze the

resultant customer outcomes for up-to-date trends of

customer profitability to verify the results of the data

mining analysis. The mechanism of the reactive CPM

route is executed for two purposes, to alert for possible

fraud whenever a drastic change in customer’s purchasing

behavior occurs, and to provide a means of examining the

reasons behind and level of differences caused by

customer predictions from the customer outcomes, allow-

ing the firm to properly calibrate related marketing

projects. Consequently, by filling the value gaps or

satisfying unmet needs in time, the firm can win back

rketecast

ScenarioAnalysis

FraudDetection &

Handling

StrategicLevel

TacticalLevel

OperationalLevel

Reactive CPM Route Guidance

CustomerAnalysis

CPM Database

ility management.

Table 1

Comparison between CPM routes

Attributes Objective Initial point Length of term Success factor

Routes

Leading CPM route Maximize customer profitability Marketing planning Long term Scenario analysis

Active CPM route Guide customer shifts Customer prediction Middle term Prediction capability

Reactive CPM route Handle customer shifts Customer monitoring Short term Flexible response

H.-F. Wang, W.-K. Hong / Industrial Marketing Management 35 (2006) 715–723718

defecting customers and lead growing customers in a more

profitable direction. Since defecting customers trend

towards profitability declination, and growing customers

trend towards expanded profitability, both are outside the

mainstream of purchasing behavior (Rust, Zeithaml, &

Lemon, 2000). Through reactive monitoring, a firm can

overlay and consolidate these identified customers by

retrieving their background information from CPM data-

bases and calibrate marketing activities targeted specifi-

cally at these customers.

3.4. Summary

Table 1 summarizes the roles of the strategic, tactical

and operational routes by their respective objectives, initial

points, durations and success factors. The leading CPM

route leads a firm to develop marketing strategy goals by

maximizing customer profitability with accurate scenario

analysis. Directed by marketing goals, the active CPM

route guides customers to sustain customer profitability

based on potential trend analysis and prediction. To ensure

customer profitability, the reactive CPM route calibrates

and controls customer shifts through evaluation and

comparison so that unexpected events can be handled

timely and effectively.

In summary, fluctuating and unpredictable customer

purchasing behavior could limit the effectiveness of the

active CPM route, resulting in wasted marketing resources

in the event that market projections differ significantly from

actual market conditions. Therefore, a firm needs a reactive

CPM route to correct marketing projects in time and store

the up-to-date information on customers’ behavior in the

CPM database for fine tuning predictions and guiding

subsequent long-term marketing planning. Since these two

routes are essential in managing customer profitability, we

shall focus on the operation and measurement of each

activity along these two routes.

4. CPM criteria

In competitive markets, the goal of managing customer

profitability is to form a stable and profitable customer base

by leading and correcting customer shifts. First of all, we

need to consider the factors that can identify the status of

customer profitability. This normally can be considered

from two perspectives: quantity and quality.

4.1. Amount identification

The quantity of a customer’s profitability can be

measured by the net contribution within certain observa-

tion period. Most of the firms adopt the present or future

CLV of a customer as an indicator to measure the amount

of customer profitability (Mulhern, 1999). However, a

more effective measure is the whole spending amount of

every customer on a specific product or service of a firm

and its competitors. Some firms have developed methods

to estimate the share of customer’s total purchasing to

arrive at the whole amount of customer profitability

(Reinartz & Kumar, 2002), while others get the informa-

tion from the common database of a specific industry, in

which every firm agrees to regularly input purchasing

records (Bell et al., 2002). The whole amount of customer

profitability gives a firm a clearer picture allowing it to

focus its efforts on profitable customers, shift potentially

profitable customers, and reduce contact costs with

unprofitable customers. By denoting the amount of

contribution as ‘‘P’’, we may apply a Neuro–Fuzzy

classification technique (Wang & Guo, 2001) to categorize

each customer as ‘‘Loss’’ (unprofitable customers), ‘‘Low’’

(less profitable customers), ‘‘Medium’’ (profitable custom-

ers) and ‘‘High’’ (most profitable customers) as a reference

for developing marketing strategies.

4.2. Volatility identification

While the amount of spending is a quantitative

indicator of customer profitability, volatility over time

can be an indicator of the quality of customer profitability.

As mentioned in the Introduction, in a competitive market,

customers should be treated as risk assets. Dhar and

Glazer (2003) explore this concept by using the volatility

of customer profitability to define a ‘‘customer beta’’ for

evaluating the relative volatility of a customer or segment

to complete a customer’s portfolio, a factor designated

‘‘V’’. The customer beta is expressed by the formula

bc=covariance (cash flow from customer, cash flow from

portfolio) / variance (cash flow from portfolio) (Dhar &

Glazer, 2003). If the customer’s )bc)o1 (i.e. A custom-

er’s returns fluctuate less than those of the portfolio) he

would be classified as a relatively steady customer

(‘‘safe’’); otherwise, the customer would be classified as

‘‘significant’’ which would then draw the attention of a

manager.

Table 2

Measures of customer profitability

Measures Indicators Levels Identifications

Amount identification P Loss Unprofitable customers

Low Less profitable customers

Medium Profitable customers

High Most profitable customers

Trend identification T Down Posssibly defecting customers

Flat Possibly stable customers

Up Promising customers

Volatility identification V Significant Relatively unsteady customers

Safe Relatively steady customers

Accessibility identification A High Close-relationship customers

Moderate Moderate-relationship customers

Low Remote-relationship customers

H.-F. Wang, W.-K. Hong / Industrial Marketing Management 35 (2006) 715–723 719

4.3. Trend identification

The concept behind trend identification is to identify the

movement between past and current customer profitability.

Let’s denote this indicator as ‘‘T’’. If T is positive,

purchasing behavior is trending up and the customers’

profitability is promising, thus the customer is marked as

‘‘Up’’. If T is negative, profitability is trending down,

indicating that the customer may be defecting, and is

therefore marked as ‘‘Down’’. A stable customer will be

marked ‘‘Flat’’. However, such rudimentary indicators

could result in false signals due to seasonal or cyclic

fluctuations of customer profitability. To avoid false signals,

a firm could evaluate the relative volatility of a customer’s

year-on-year profitability as a complementary indicator

through the ‘‘customer beta’’ formula mentioned in Section

4.2. The customer beta is expressed by the formula

bp=covariance (cash flow from identical periods year-on-

year, cash flow from the first period) /variance (cash flow

from the first period). A value for )bp)>1 would confirm the

trend identification—otherwise it could be attributed to

seasonal or cyclic fluctuations of customer profitability.

4.4. Accessibility identification

With these indicators for amount, volatility and trend

providing a clear and classified picture of a given customer’s

profitability, firms are better able to develop one-to-one

Table 3

Examples of active tactics in CPM route

Attributes Objectives Required analysis Initial

Active tactics

Win back tactics Win back customer share Churn analysis P t� 1

A tmL

Upgrade tactics Increase customer share Upgrade analysis P tmL

Loyalty tactics Broaden market spaces Loyalty analysis P tRM

V t =S

Reactivation tactics Increase customer base Customer base

analysis

P t� n

TtmU

Cost down tactics Decrease contact cost Channel analysis P t =L

customer relationships and marketing strategies. By ap-

propriately allocating marking resources to targeted cus-

tomers, firms are better positioned to increase profits.

However, these indicators can’t accurately show the

possible impact of marketing activity on a customer’s

purchasing behavior. Therefore, we need to take customer

accessibility—the possibility of a potential customer or

segment accepting a particular marketing package—into

consideration. Information on accessibility, designated

‘‘A’’, can be derived from marketing or contact records.

An individual customer who rarely responds to a firm’s

marketing programs over the customer lifecycle or a

corporate customer who is an investor from competitors

will be viewed as having ‘‘low’’ customer accessibility,

whereas a customer who is responsive and/or has close

relationship with service clients is viewed as having

‘‘high’’ customer accessibility, and other customers will

be viewed as having ‘‘moderate’’ customer accessibility.

Table 2 summarizes the symbols with their leveled

meanings for identifications and interpretations.

4.5. Identification of targeted customers

These indicators are combined as classification criteria to

identify target customers for the active CPM route by AND/

OR operations, allowing for the development of the

corresponding active tactics as shown in Table 3. Different

clusters of customers entail different aims in the develop-

classification criteria Customers targeted

for active CPM

Active marketing

proposals

=High & Tt =Down &

ow

Defecting customers Responsive actions

oss & T t =Up Growing customers Total solutions

edium & TtmDown &

afe

Steady customers New product

provisioning

RMedium & P toLow &

p & A tm High

Inactive customers Reactive activities

oss & TtmUp Unprofitable

customers

Economic channel

services

H.-F. Wang, W.-K. Hong / Industrial Marketing Management 35 (2006) 715–723720

ment of marketing tactics. Commonly adopted tactics

include ‘‘Win back’’ tactics for reducing defection rates;

‘‘Upgrading’’ tactics for increasing the share of wallet of

promising customers; and FLoyalty_ tactics for keeping

steady customers. In addition, while ‘‘Reactive’’ tactics

stimulate inactive customers and increase the customer base,

‘‘Cost down’’ tactics provide economic channels to service

unprofitable customers. Other tactics can also be developed

by combining these criteria for leading customer shifts in

directions conducive to the firm’s goals.

5. Implementation of CPM

To implement these indicators in the CPM framework,

Fig. 2 shows the interactions between active and reactive

CPM procedures. Depending on the purpose of tactical

analyses, we can first identify targeted customers through

the initial classification criteria as shown in Table 3. Data

mining tools processes, such as neural networks as proposed

by Wang and Guo (2004), can then be used to isolate

behavior patterns of the targeted customers, allowing for the

verification of further classification criteria, and the weight-

ing of their relative importance. Therefore, other character-

istics of targeted customers can be extracted from

geographic, demographic, attitudinal or behavioral informa-

tion, which will help firms design more suitable marketing

programs and activities. Besides, data mining tools can help

us establish a learning model derived from sampling data

with given classification criteria to separate potential

targeted customers from prospects or new customers about

whom we do not have enough behavioral data. Active

Customers

Evaluation Criteria

AccommodatingCustomers

M

ResistantCustomers

UnpredCusto

Customer Surveys to Determine Valu

Purposes of Analysis

TargetedCustomers

Customer Outcomes

Active Marketing Activity No

N

Customer Classification Criter

TargetedCustomers

N

No MReactive Marketing Activity

Fig. 2. Active/reactive

marketing activities will then be applied to targeted

customers to achieve tactical marketing goals.

Over time, an evaluation of a customer’s response to

active marketing activities reveals the accessibility of

Faccommodating_ customers (those who accepted the

promotion) and potential increases in their customer profit-

ability, along with some resistant customers who don’t

accept active marketing programs and have no increase of

customer profitability. Those who have no increase of

customer profitability are judged by the difference of their

customer profitability (DCP) before and after action.

Decreasing customer profitability is marked as DCPo0.

At the same time, customer outcomes of non-targeted

customers are evaluated by monitoring criteria, as shown

in Table 4, to identify which customers should be targeted

for reactive marketing tactics. These customers are those

who are not identified by the prediction models and have

been denoted as Funpredictable_ customers. For resistant

customers and unpredictable customers, a firm can perform

further customer surveys or interviews to get new insight

about these customers and discover opportunities for

improving existing or creating new value. Then, through

benefit evaluation, we can specify customers targeted for

reactive marketing activities, and allocate resources to

prevent or reverse their migration in accordance with the

firm’s marketing tactics. The outcome of reactive marketing

can be cyclically evaluated and monitored to ensure the final

outcome matches tactical marketing goals.

In such an active/reactive CPM procedure, marketing

activities are designed to lead customer shifts by improving

value or creating new value to satisfy customers’ out-

standing needs. These win–win strategies will help a firm

Note: Reactive CPM procedure

onitoring Criteria

ictablemers

PredictableCustomers

e Creation

CPMDatabase

Marketing Activity

on-targetedCustomers

ia

on-targetedCustomers

arketing Activity

Data MiningModels

BenefitEvaluation

CPM procedure.

Table 4

Examples of reactive tactics in CPM route

Attributes Customers of

active CPM

Evaluation/monitoring criteria Customers targeted

for reactive CPM

New customer

insights

Reactive marketing

proposalsReactive tactics

Win back tactics Targeted DCPo0 Resistant Defecting reasons Responsive actions

Non-targeted P t + 1=High & Tt + 1=Down &

A t + 1mLow

Unpredictable

Upgrade tactics Targeted DCPo0 Resistant Purchasing

motivations

Total solutions

Non-targeted P t + 1mLoss & Tt + 1=Up Unpredictable

Loyalty tactics Targeted DCPo0 Resistant Customer needs New product

provisioningNon-targeted P t + 1RMedium & Tt + 1mDown &

V t + 1=Safe

Unpredictable

Reactivation tactics Targeted DCPo0 Resistant Reasons for

inactivity

Reactive activities

Non-targeted P t� nRMedium & Pt + 1oLow &

T t + 1mUp & A t + 1=Low

Unpredictable

Cost down tactics Targeted DCPo0 Resistant Channel preference Economic channel

servicesNon-targeted P t + 1=Loss & Tt + 1mUp Unpredictable

H.-F. Wang, W.-K. Hong / Industrial Marketing Management 35 (2006) 715–723 721

manage customer profitability well enough to guarantee

stable and long-term customer relationships.

6. Empirical application

We have applied the proposed CPM mechanism to the

international business of a telecom company facing tough

competition from three other carriers. At the time, the

company was engaged in a price war, which had lead

customers ignore all product attributes beyond price. The

company’s profitability had declined for two years running

and its market share had dropped from 85% to 60%. Price

campaigns only brought short-term advantages and were

lost quickly as competitors fought back. Mass marketing

activities were just wasting time and money. Fortunately, the

company had kept reasonably complete customer purchas-

ing records and background information in its databanks,

and the new Vice President of the international business

group made this data and modern data mining tools the

center of the new marketing strategy.

6.1. Establishing the CPM

To avoid intensive price competition, the firm established

a CPM system to integrate its customer data storage system.

The marketing activities emphasized in the CPM system

were focused on the attributes of products that had

competitive advantages to improve existing values and to

create new values, with the intention of emphasizing

customer values by turning the focus from market share to

customer share or new market space. To establish the CPM

system, six task guidelines were established as follows:

? Customized marketing strategy. Based on scenario

analyses and market forecasts, the company established

a customized marketing strategy to consolidate its

customer base, upgrade potentially profitable customers,

and win back defecting customers.

? Identification of targeted customers. To implement this

strategy, active/reactive CPM procedures were initiated

to identify targeted customers by classification, evalua-

tion and monitoring criteria of different active/reactive

tactics as shown in Tables 2 and 3.

? Discovery of unmet needs and new opportunities. To

avoid relying on price competition as well as to extend

customer share and market space, a wider-scale customer

survey and more focused personal interviews were

conducted to discover the unmet needs of targeted

customers and to gauge the seriousness of those needs.

The resulting data on reasons for defection, purchasing

motivation and customer needs were analyzed to update

the CPM database, improve marketing activities and

create new opportunities to deliver valuable products and

services to targeted customers.

? Development of marketing programs. Based on the

classification criteria and the identified needs of the

targeted customers, the size and content of potential

marketing programs were developed to address different

marketing tactics. The costs, competitiveness, feasibility

and the expected benefits of those potential programs

were then evaluated. The resulting outcomes enabled us

to isolate customers targeted for active/reactive market-

ing activities from the general population.

? Execution and follow-up of marketing programs. The

company implemented alert management processes to

identify the status of customer shifts and incentive

mechanisms for its marketing channels. Red, leading,

and reactivating alerts were used to indicate defecting,

upgrading and inactive customers respectively, allow-

ing them to be targeted by different marketing

programs. Moreover, incentive mechanisms were set

to stimulate marketing channels to achieve better

performance.

? Standardization of CPM procedures. To establish con-

tinuous management of customer profitability, CPM

system procedures were standardized to continually

collect and process marketing results and feedback,

H.-F. Wang, W.-K. Hong / Industrial Marketing Management 35 (2006) 715–723722

which also contributed to continuous refinement and

updating of active/reactive CPM procedures. These

standardized procedures also covered the cultivation of

data analysts and marketing experts, the design of

marketing program criteria, channel management regu-

lation, and the establishment of marketing knowledge

base.

6.2. Empirical results

Figs. 1 and 2 show the CPM system implemented along

the guidelines presented above. Based on this monitoring

scheme, 16 million potential customers were considered

from the database. Of these, 1.4 million customers had

purchasing activity every month. These customers were then

classified according to our model’s classification criteria,

and the active marketing activities were initialed with

marketing tactics appropriate to each group.

For example, in October 2003, the company engaged in

win-back tactical marketing, in which 23,724 profitable

and defecting customers were identified according to the

initial classification criteria in Table 3 followed by more

detailed classification criteria through the data mining

model developed by Wang and Guo (2004). This resulted

in the identification of 17,892 resistant customers and

3,732 unpredictable customers. This was followed by

telephone surveys and personal interviews of these

customers to discover their reasons for defecting in order

to win back customer share. This allowed the company to

identify 6541 customers for reactive win-back activity

from among 21,624 defecting customers. Customized

marketing packages were developed to retain these

customers and, in November and December of 2003,

3492 resistant customers and 972 unpredictable customers

accepted the reactive win-back packages.

Upgrade and loyalty tactics were also applied beginning

in October 2003 with significant results. An upgrading

analysis was conducted to identify customers who were

currently less profitable but who were trending up,

allowing the company to provide expanded solutions to

stimulate these customers’ purchasing motivation. These

solutions included filling up value gaps or adding new

value in pre-sale, on-sale and post-sale purchasing pro-

cesses to encourage customers to use the company’s

products and services. To further expand market scale, a

loyalty analysis was carried out on steady customers, so

the company could consolidate its customer base for long-

term sustainable profitability by providing loyal and

profitable customers with needs-appropriate six-month free

trials on a range of new products.

As a result of these innovations, the number of customers

with monthly purchasing activity grew 3%, while average

product usage in this customer group increased 42% after 6

months of implementation. The adoption rate of new

products after free trials increased 36%, and customer

profitability grew 12% after the 6-month implementation of

win-back, upgrade and loyalty tactics, showing the effec-

tiveness of the developed CPM system in customer profit-

ability management.

7. Summary and conclusion

In a competitive market, most customers would rather

have multiple loyalties rather than give exclusive loyalty to

one firm. The dynamic nature of customers’ behavior

increases the unsteadiness and unpredictability of customer

profitability, making marketing planning and tactics devel-

opment inefficient and ineffective. Therefore, marketing

tactics and projects cannot depend entirely on market

forecasts and customer prediction. A firm needs a practical

CPM system to transfer customers’ focus from price to new

valuable product/service attributed to consolidate the over-

all operation of customer profitability management. It is

thus the aim of this study to develop a CPM system to help

firms lead customer migration along desired tracks to

eventually achieve their marketing goals. An efficient and

effective implementation allows a firm to identify appro-

priate customers for active marketing tactics along the

proposed active CPM route, while keeping non-targeted

customers under observation. Then, the subsequent reactive

CPM route calibrates related marketing projects by filling

value gaps or satisfying unmet needs. The interaction

between these two routes illustrates the concept and

realization of effective customer profitability management

with an emphasis on responsive marketing. That is, the

proposed CPM system is a continuous process of customer

reevaluation, through the use of data mining techniques, to

spot trends in customer behavior.

The proposed scheme has been adopted by the Interna-

tional Telecommunication Department of a local telecom

company. With the benefit of huge database, the analysis

and model building could be done in a very short time, and

promising results were achieved after only six months’

intensive operation.

Classification and monitoring criteria are the keys of

managing customer profitability. The combination and

thresholds of the customer profitability measures need to

be developed according to the users’ needs and experi-

ences. Further study is required on how to constitute

systematic rules in this scheme for identifying proper

thresholds according to specific industrial characteristics

and market environments.

References

Bell, D., Deighton, J., Reinartz, W. J., Rust, R. T., & Swartz, G. (2002,

August). Seven barriers to customer equity management. Journal of

Service Research, 5, 77–85.

Berger, P. D., Kumar, V., Parasuraman, A., & Terry, C. (2002, August).

Marketing actions and the value of customer assets: A framework for

customer asset management. Journal of Service Research, 5, 39–54.

H.-F. Wang, W.-K. Hong / Industrial Marketing Management 35 (2006) 715–723 723

Berger, P. D., & Nasr, N. I. (1998). Customer lifetime value:

Marketing models and applications. Journal of Interactive Marketing,

12(1), 17–30.

Dhar, R., & Glazer, R. (2003, May). Hedging customers. Harvard Business

Review, 86–92.

Hogan, J. E., Lemon, K. N., & Rust, R. T. (2002, August). Customer equity

management: Charting new directions for the future of marketing.

Journal of Service Research, 5, 4–12.

Jacobs, F. A., Johnston, W., & Kotchetova, N. (2001). Customer profit-

ability: Prospective vs. retrospective approaches in a business-to-

business setting. Industrial Marketing Management, 30(4), 353–363.

Jain, D., & Singh, S. S. (2002). Customer lifetime value research in

marketing: A review and future directions. Journal of Interactive

Marketing, 16(2), 17–29.

Libai, B., Narayandas, D., & Humby, C. (2002, August). Toward an

individual customer profitability mode: A segment-based approach.

Journal of Service Research, 5, 69–76.

Mulhern, F. J. (1999). Customer profitability analysis: Measurement,

concentration, and research directions. Journal of Interactive Market-

ing, 13(1), 25–40.

Peltier, J. W., & Schribrowsky, J. A. (1997). The use of need-based

segmentation for developing segment-specific direct marketing strat-

egies. Journal of Direct Marketing, 11(4), 53–62.

van Raaij, E. M., Vernooij, M. J. A., & van Triest, S. (2003). The

implementation of customer profitability analysis: A case study.

Industrial Marketing Management, 32(7), 573–583.

Reinartz, W., & Kumar, V. (2002, July). The mismanagement of customer

loyalty. Harvard Business Review, 86–94.

Rust, R. T., Zeithaml, V., & Lemon, K. (2000). Driving customer

equity: How customer lifetime value is reshaping corporate strategy.

The Free Press.

Stahl, H. K., Matzler, K., & Hinterhuber, H. H. (2003). Linking customer

lifetime value with shareholder value. Industrial Marketing Manage-

ment, 32(4), 267–279.

Wang, H. F., & Guo, C. Y. (2001). Hierarchical factor analysis. In Antoniou

G., et al. (Eds.), Advances in signal processing and computer

technologies (pp. 61–66). WSES Press.

Wang, H. F., & Guo, C. Y. (2004). Factor analysis in data mining.

International Journal of Computers and Mathematics with Applica-

tions, 48, 1765–1778.

Zeithaml, V. A., Rust, R. T., & Lemon, K. N. (2001). The customer

pyramid: Creating and serving profitable customers. California Man-

agement Review, 43(4), 118–142.

Hsiao-Fan Wang is the professor of the Department of Industrial

Engineering and Engineering Management, National Tsing Hua University

after she graduated from Cambridge University, England in 1982. Her

research interests are Mathematical Programming, Fuzzy Set Theory,

Multicriteria Decision Analysis and Data Mining. The major journals

published are FSS, EJOR, IJMDM, IEEE Transactions on SMC and Fuzzy

Systems, etc.

Wei-Kuo Hong is a Ph. D. student in the Department of Industrial

Engineering and Engineering Management, National Tsing Hua University.

He is also an instructor of Business Customer Department, Data

Communication Branch of Chunghwa Telecom. He has worked as a

consultant to lead a team for implementing database marketing in

Chunghwa Telecom. His research interests cover Data Warehouse and

Data Mining, Strategic Planning and Customer Profitability Management.