Page 1 of 9 ANZMAC 2009

A Logit Model of Customer Churn as a Way to Improve the

Customer
Retention Strategy in the Fixed Telecommunications Industry in
Portugal

Rui Menezes a and Sofia Portela b

a
Department of Quantitative Methods, ISCTE Business School
Av. Forças Armadas, 1649-026 Lisboa, Portugal
Phone No.: 00 351 21 793 50 00
rui.menezes@iscte.pt

b
Department of Quantitative Methods, ISCTE Business School
Av. Forças Armadas, 1649-026 Lisboa, Portugal
Phone No.: 00 351 21 793 50 00
slportela@iscte.pt

Abstract

The considerable increase of business competition in the Portuguese fixed

telecommunications industry over the last decade has given rise to a phenomenon of customer
defection, which has serious consequences for the business performance. Consequently,
researchers have recognised the importance of a deeply study of customer defection in
different industries and geographical locations. This study examines the determinants of
customer churn in this industry in Portugal by using a logit model. We found evidence that
the variables that most influence churn are related to the customer spending with the provider.
The results allow managers to have a priori knowledge about what customers are probably
going to defect, and, thus, they can take preventive strategies to improve the retention of
profitable customers.

Keywords
Customer management, customer retention, customer churn, logit.

1
 ANZMAC 2009 Page 2 of 9

A Logit Model of Customer Churn as a Way to Improve the

Customer
Retention Strategy in the Fixed Telecommunications Industry in
Portugal

Abstract

The considerable increase of business competition in the Portuguese fixed

telecommunications industry over the last decade has given rise to a phenomenon of customer
defection, which has serious consequences for the business performance. Consequently,
researchers have recognised the importance of a deeply study of customer defection in
different industries and geographical locations. This study examines the determinants of
customer churn in this industry in Portugal by using a logit model. We found evidence that
the variables that most influence churn are related to the customer spending with the provider.
The results allow managers to have a priori knowledge about what customers are probably
going to defect, and, thus, they can take preventive strategies to improve the retention of
profitable customers.

Introduction

Although the customer-centric paradigm is more than 50 years (see Shah et al., 2006),
the customer revolution just happened in the 80s (Boyce, 2000), when firms have been
encouraged to focus on customers rather than on products. In this way, customers became the
“centre” of the organizations (Boyce, 2000). For a long time, researchers suggested that
“customer is always right” and thus managers should focus in satisfying the customer needs
and improving the customer satisfaction. This period was designed as “traditional marketing
strategy” by Gupta and Lehmann (2005). These authors argue that a new paradigm has
emerged, which they denominate “customer-base strategy”, and which is also designated
“customer equity management” by Blattberg, Getz, and Thomas (2001) and Hogan, Lemon,
and Rust (2002). The main difference between these paradigms is that the traditional
marketing strategy was only concerned with the value that a firm provides to a customer, and
the customer-base strategy is also concerned with the value of a customer to the firm. Thus,
this approach emphasises the two sides of customer value. According to the customer-base
approach, the firm should invest to provide value to the customer and, conversely the
customer should provide returns to the firm and its shareholders (Bolton and Tarasi, 2006).
Boyce (2000) and Johnson and Selnes (2005) do suggest that the marketing thinking reveals
an evolution from the “customer is king” to “customer is cash”.
On the other hand, for a long time, managers focused on firm growth, and, thus, on
customer acquisition (Rosenberg and Czepiel, 1983). More recently, a new paradigm has been
suggested by researchers, which is based on customer retention. Even though Reichheld and
Sasser (1990) were not pioneers in pointing out the advantages of customer retention (East et
al., 2006), they firstly provided evidence about those advantages, which are based on a strong
relationship between customer retention and profitability. They found that long-time
customers (i) spend more over time, (ii) the operating costs to serve them decline over time,
(iii) become more loyal and then promote the word-of-mouth, and (iv) are less price-sensitive.
Additionally, Reichheld (1996) argues that customer defection has severe effects on firms’
profitability because firms have to incur in heavy costs to acquire new customers and older
customers usually generates greater cash flow and profits than newer ones. It is popularly
believed that to get a new customer cost at least five times more than keep an existing one.
These conclusions caused a change in the marketing theory, since researchers started

1
Page 3 of 9 ANZMAC 2009

definitely arguing that enterprises should focus on customer retention rather than on customer
acquisition (e.g., Reichheld, 1996; Thomas et al., 2004; Trubik and Smith, 2000).
Reichheld and Sasser (1990) defended that firms should develop a customer
management toward zero defections. Nevertheless, Blattberg et al. (2001) classify this idea as a
myth seeing that there are some exogenous and uncontrollable factors that affect the
customer retention potential. They also emphasise that retention is not free. Van den Poel and
Larivière (2004) point out that a retention rate of 100 percent is utopian, due to uncontrollable
reasons of defection, like natural death or moving to a foreign country.
Simultaneously, several researchers argue that the enterprises should neither focus on nor
try to retain all of their current customers, because they are probably investing in
unprofitable customers (Gupta and Lehmann, 2003; Jain and Singh, 2002; Malthouse and
Blattberg, 2004; Thomas et al., 2004), and, in this way, they are destroying value (Gupta and
Lehmann, 2005; Jain and Singh, 2002) because (i) the retention of unprofitable customer is
damaging to the firm, and (ii) the money wasted on the retention of unprofitable is not used on
the retention of profitable ones, who are harder to get (Thomas et al., 2004). In conclusion, the
retention strategy must be strongly linked with the customer value.
We argue that quantitative models that detect probable customer churners (i.e., the
customer’s decision to terminate the relationship with the service provider) in the next few days
are very useful for managers. Actually, if managers have a priori knowledge about what
customers are probably going to cancel the relationship with the service provider, they can take
preventive strategies to avoid the defection of potentially profitable customers. Thus, correct
decisions about the retention strategy of each specific potentially churner should be
complemented with his/her expected customer lifetime value (CLV).
This study intends to examine the determinants of the residential churn in the fixed
telecommunications industry in Portugal and detect the probable customer churners in the
next few days by using a logit model. The study of this industry in Portugal is of special
importance given the dynamism and competitiveness of this market. Furthermore, firms
present significant customer churn rates, which is very damaging to their financial
performance. We are convinced that a better knowledge of the customer churn patterns will
support managers on customer retention and, thus, to improve their competitiveness.
Despite the large amount of research done on customer churn, there are few studies
applied to the fixed telecommunications industry. Furthermore, to the best of our knowledge, none
is applied to firms that simultaneously offer ADSL, fixed line telephone, pay-TV, and home
video services (i.e. bundled offers). The majority of published research about customer churn
prediction in the telecommunications industry analyse the mobile telecommunications. This issue
has never been studied in Portugal.
Customer churn has been studied in different industries ( e.g., banking, insurance,
telecommunications) and using different statistical techniques. Several researchers have used
logit models (e.g., Ahn, Han, and Lee, 2006; Burez and Van den Poel, 2007; Kim and Yoon,
2004; Lemmens and Croux, 2006; Mozer et al., 2000a; Mozer et al., 2000b; Neslin et al.,
2006; Seo, Ranganathan, and Babad, 2007). Other researchers have analysed the customer
churn in contractual settings and continuous time by using duration models (e.g.,
Bolton,1998; Burez and Van den Poel, 2007; Burez and Van den Poel, 2008; Drew et al.,
2001; Jamal and Bucklin, 2006; Mani et al., 1999; Schweidel, Fader, and Bradlow, 2008).
Lastly, several techniques of data mining have been used to model customer churn (e.g., Bin,
Peiji, and Juan, 2007; Burez and Van den Poel, 2007; Burez and Van den Poel, 2008; Drew et
al., 2001; Ferreira et al., 2004; Hung, Yen, and Wang, 2006; Lemmens and Croux, 2006; Lu,
2002; Lu, 2003; Mani et al., 1999 ; Mozer et al., 2000a; Mozer et al., 2000b; Neslin et al.,
2006; Wei and Chiu, 2002; Yan et al., 2001; Zhang et al., 2006).

2
 ANZMAC 2009 Page 4 of 9

Buckinx and Van den Poel (2005), Hadden et al. (2005), Reinartz and Kumar (2003),
Song et al. (2004), and Van den Poel and Larivière (2004) present literature reviews of
customer churn studies. Ahn, Han, and Lee (2006) point out that the reasons of customer
churn and the customer behaviour towards churn need to be more studied.

Methodology

This study uses data from a Portuguese fixed telecommunications firm which presents
bundled offers of ADSL, fixed line telephone, pay-TV and home-video. The data
encompasses a large number of covariates, which include customer’s basic information,
demographics, churn flag, customer historical information about usage, billing, subscription,
credit, and other, for 830 customers who completed a questionnaire about satisfaction with the
service provider. Logit models will be used. The dependent variable of the logit models is
customer status (0 if the customer is active and 1 if the customer defected). The significant
variables of these models are those that discriminate between churners and non-churners.

Results

Two logit models were estimated to examine the determinants of customer churn. The
significant covariates that discriminate churners from non-churners are presented in table 1 and
the final models are presented in table 2.
Our results show that the determinants of customer churn are: gender, total number of
overdue bills since the beginning of the contract, value of current debt, monthly average of
value of calls made by the customer, total revenues from the customer since the beginning of
the contract, monthly average of the revenues from the fixed-telephone service, and duration
of the relationship with the service provider. The model 1 does not include the duration of the
relationship. On the other hand, model 2 includes this duration but not the overall and
telephone revenues, because duration and these revenues are correlated. Both models classify
the customers very well (the percentage of correct assignments are 95.3% and 95.4% for the
model 1 and 2, respectively). Comparing the models 1 and 2, it can be concluded that the
effect of the common covariates are similar. Table 3 compares both groups of customers
(active and inactive) using the means/ proportions of the significant covariates of the models.
It should be noted that there is a relationship between the probability of churn and the
customer lifetime, because the customer lifetime is shorter for customers that have a larger
probability of churn. The results are discussed taking this idea into consideration.

Table 1 - Definition of the significant covariates

Covariates Description
Gender Gender of the customer (0 - female; 1 - male)
Total dunning Total number of overdue bills since the beginning of the contract
Debts Value of current debt (€)
Monthly average of value of calls made by the customer between
Value of calls December 2007 and November 2008 (€)
Overall revenues Total revenues from the customer since the beginning of the contract (€)
Telephone Monthly average of the revenues from the fixed-telephone service (calls
revenues and fee) between December 2007 and November 2008 (€)
Duration Duration of the relationship with the service provider (days)

3
Page 5 of 9 ANZMAC 2009

Table 2 - Determinants of customer churn

Variable Model 1a,b Model 2a,b

Gender -2.536 -2.054
(0.603) (0.593)
Total dunning 4.586 4.660
(0.704) (0.689)
Debts -0.069 -0.068
(0.010) (0.010)
Gender × Overall revenues 0.002 0.001
(0.001) (0.000)
Value of calls -0.457 -0.352 **
(0.166) (0.168)
Overall revenues -0.001 -
(0.000)
Telephone revenues 0.080 ** -
(0.031)
Duration - -0.001 **
(0.000)
McFadden's R2 0.278 0.264
Log-L(max) -113.539 -115.760
Log-L(min) -157.326 -157.326
AIC 0.291 0.294
SBC 0.331 0.328
% correct assignments 95.3% 95.4%
Notes: a Dependent variable is Customer Status. Estimated by ML binary logit (quadratic hill climbing). b 829
observations. Standard errors in parentheses. *** significant at 1%. ** significant at 5%. * significant at 10%.

Table 3 - Description of failed and active customers

Failed ( X / p) Active ( X / p)
Gender 0.49 0.71
Total dunning 0.28 0.05
Debts 24.93 38.66
Value of calls 0.80 1.22
Overall revenues 1145.95 1146.03
Telephone revenues 6.92 5.76
Duration 754.44 799.07

The results indicate that females are more prone to churn than males, which is
consistent with Ahn, Han, and Lee (2006) and Seo, Ranganathan, and Babad (2007) who
found evidence that the customer lifetime for males is larger than for females. Nevertheless,
this result contradicts Kim and Yoon (2004) and Madden, Savage, and Coble-Neal (1999).
It seems that the probability of a customer to churn increases as the total number of
overdue bills and the monthly average of customer spending on fixed-telephone increase.
Some authors have found similar evidence about the monthly spending with the provider
(e.g., Ahn, Han, and Lee, 2006; Bolton, 1998; and Madden, Savage, and Coble-Neal, 1999),
but other found the opposite (e.g., Kim and Yoon, 2004).

4
 ANZMAC 2009 Page 6 of 9

Furthermore, it seems that customers with smaller current debts are more susceptible to
churn than customers with larger current debts, which contradicts our prior expectations. This
can be due to the fact that, until recently, the firm’s policy was not stopping the service to
customers with debts. Ahn, Han, and Lee (2006) did not find any relationship between the value
of current debts and customer lifetime.
Our results also show that the probability of customer churn is larger for customers with
smaller amounts of overall revenues and monthly average of value of calls. This is consistent
with the results of Jamal and Bucklin (2006). Zhang et al. (2006) also found that the overall
revenues from the last 6 months affects the customer lifetime.
It seems that customers with longer relationships with the service provider have a
lower probability of churn, which is in conformity with the customer retention theory of
Reichheld and Sasser (1990).
The number of products that each customer has purchased and the number of products that
each customer has presently are not significant variables, which mean that these variables do not
discriminate between churners and non-churners.
Lastly, contrary to our prior expectations, customer satisfaction is not a significant
variable, which suggests that customer satisfaction in this context is not a reason for customer
churn. A possible explanation of this finding is that even though the customer is not satisfied,
he/she may do not switch to other operator due to inertia or habit. This contradicts the
majority of literature about satisfaction ( e.g., Bolton, 1998; Eshghi, Haughton, and Topi,
2007). Kim and Yoon (2004) found that whereas some types of satisfaction positively affects
the survival time, other do not have any influence. Van den Poel and Larivière (2004) show
that some studies did not find any influence of satisfaction on survival time.

Conclusions

The analysis of the customer churn in the Portuguese fixed telecommunications

industry suggests very interesting conclusions, which can support the marketing managers.
We identify some variables that discriminate between churners and non-churners customers.
Based on the results of this study, it can be concluded that, among the profitable customers,
managers should develop special retention strategies for females, customers with more
overdue bills, larger monthly average spending on fixed-telephone, smaller current debts,
smaller amounts of overall revenues and monthly average of value of calls, and more recent
customers, because these customers are more prone to churn. In this way, firms can become
more profitable, because the retention strategy will be more efficient. It appears that firms
should concentrate less on customer satisfaction because it does not seem to be an important
reason of customer churn. Furthermore, these models provide knowledge about the
probability of each specific customer to churn in the next few days, and managers can develop
retention strategies to specific customers and for the most urgent cases.

Acknowledgments

We wish to thank the firm for providing its internal customer database, without which this
study would not be possible, and those who strongly believe on the importance of the
expected results to the improvement of its customer portfolio management. We also thank to FCT
- Fundaç ‫م‬o para a Ciência e Tecnologia for its financial support under the grant
PTDC/GES/73418/2006.

5
Page 7 of 9 ANZMAC 2009

References

Ahn, J.-H., Han, S.-P., Lee, Y.-S., 2006. Customer churn analysis: churn determinants and
mediation effects of partial defection in the Korean mobile telecommunications service
industry. Telecommunications Policy 30 (10-11), 552-568.
Berger, P.D., Bolton, R.N., Bowman, D., Briggs, E., Kumar, V., Parasuraman, A., Terry, C.,
2002. Marketing actions and the value of customer assets: a framework for customer
asset management. Journal of Service Research 5 (1), 39-54.
Bin, L., Peiji, S., Juan, L., 2007. Customer churn prediction based on the decision tree in
personal handyphone system service. Proceedings of the International
Conference on
Service Systems and Service Management.
Blattberg, R.C., Getz, G., Thomas, J.S., 2001. Customer Equity - Building and Managing
Relationships as Valuable Assets, Harvard Business School Press.
Bolton, R.N., 1998. A dynamic model of the duration of the customer’s relationship with a
continuous service provider: the role of satisfaction. Marketing Science 17 (1), 45-65.
Bolton, R.N., Tarasi, C.O., 2006. Managing customer relationships. Review of Marketing
Research 3, 3-38.
Boyce, G., 2000. Valuing customers and loyalty: the rhetoric of customer focus versus the
reality of alienation and exclusion of (devalued) customers. Critical Perspectives on
Accounting 11 (6), 649-689.
Buckinx, W., Van den Poel, D., 2005. Customer base analysis: partial defection of
behaviourally loyal clients in a non contractual FMCG retail setting. European Journal
of Operational Research 164 (1), 252-268.
Burez, J., Van den Poel, D. 2008. Separating financial from commercial customer churn: a
modeling step towards resolving the conflit between the sales and credit department.
Expert Systems with Applications 35, 497-514.
Burez, J., Van den Poel, D., 2007. CRM At a pay-TV company: using analytical models to
reduce customer attrition by targeted marketing for subscription services. Expert Systems
with Applications 32 (2), 277-288.
Dhar, R., Glazer, R., 2003. Hedging customers. Harvard Business Review 81 (5), 86-92.
Drew, J.H., Mani, D.R., Betz, A.L., Datta, P., 2001. Targeting customers with statistical and
data-mining techniques. Journal of Service Research 3 (3), 205-219.
East, R., Hammond, K., Gendall, P., 2006. Fact and fallacy in retention marketing. Journal of
Marketing Management 22 (1), 5-23.
Eshghi, A., Haughton, D., Topi, H., 2007. Determinants of customer loyalty in the wireless
telecommunications industry. Telecommunications Policy 31 (2), 93-106.
Ferreira, J.B., Vellasco, M., Pacheco, M.A., and Barbosa, C.H., 2004. Data mining techniques
on the evaluation of wireless churn. Proceedings of the European
Sysmposium on
Artificial Neural Networks, 483-488.
Gupta, S., Lehmann, D.R., 2003. Customers as assets. Journal of Interactive Marketing 17 (1),
9-24.
Gupta, S., Lehmann, D.R., 2005. Managing Customers as Investments: The Strategic Value of
Customers in the Long Run, Wharton School Publishing.
Hadden, J., Tiwari, A., Roy, R., and Ruta, D., 2005. Computer assisted customer churn
management: state-of-the-art and future trends. Computers & Operations Research
34
(10), 2902-2917.
Hogan, J.E., Lemon, K.N., Rust, R.T., 2002. Customer equity management: charting new
directions for the future marketing. Journal of Service Research 5 (1), 4-12.
6
 ANZMAC 2009 Page 8 of 9

Hung, S.-Y., Yen, D.C., Wang, H.-Y., 2006. Applying data mining to telecom churn
management. Expert Systems with Applications 31 (3), 515-524.
Jain, D., Singh, S.S., 2002. Customer lifetime value research in marketing: a review and
future directions. Journal of Interactive Marketing 16 (2), 34-46.
Jamal, Z., and Bucklin, R.E., 2006. Improving the diagnosis and prediction of customer
churn: a heterogeneous hazard modeling approach. Journal of Interactive Marketing
20
(3-4), 16-29.
Johnson, M.D., Selnes, F., 2005. Diversifying your customer portfolio. MIT Sloan
Management Review 46 (3), 10-14.
Kim, H.-S., Yoon, C.-H., 2004. Determinants of subscriber churn and customer loyalty in the
Korean mobile telephony market. Telecommunications Policy 28 (9-10), 751-765.
Lemmens, A., Croux, C., 2006. Bagging and boosting classification trees to predict churn.
Journal of Marketing Research 43 (2), 276-286.
Lu, J., 2002. Predicting customer churn in the telecommunications industry - an application
of survival analysis modelling using SAS, Proceedings of the SAS Conference.
Lu, J., 2003. Modeling customer lifetime value using survival analysis - an application on the
telecommunications industry. Proceedings of the SAS Conference.
Madden, G., Savage, S.J., Coble-Neal, G., 1999. Subscriber churn in the Australian ISP
market. Information Economics and Policy 11 (2), 195-207.
Malthouse, E.C., Blattberg, R.C., 2004. Can we predict customer lifetime value?. Journal of
Interactive Marketing 19 (1), 2-16.
Mani, D.R., Drew, J., Betz, A., Datta, P., 1999. Statistics and data mining techniques for
lifetime value modelling. Proceedings of the fifth ACM SIGKDD
International
Conference on Knowledge Discovery and Data Mining, 94-103.
Mozer, M.C., Wolniewicz, R., Grimes, D.B., Johnson, E., Kaushansky, H., 2000a. Churn
reduction in the wireless industry. Advances in Neural Information Processing
Systems
12, MIT Press, Cambridge, MA., 935-941.
Mozer, M.C., Wolniewicz, R., Grimes, D.B., Johnson, E., Kaushansky, H., 2000b. Predicting
subscriber dissastisfaction and improving retention in the wireless telecommunications
industry. IEEE Transactions on Neural Networks - Special Issue on Data
Mining and Knowledge Representation 11 (3), 690-696.
Neslin, S.A., Gupta, S., Kamakura, W., Lu, J., and Mason, C.H., 2006. Defection detection:
measuring and understanding the predictive accuracy of customer churn models. Journal
of Marketing Research 43 (2), 204-211.
Reichheld, F.F., 1996. Learning from Customer Defections. Harvard Business Review 74 (2), 56-69.

Reichheld, F.F., Sasser, W.E., 1990. Zero defections: quality comes to services. Harvard
Business Review 68 (5), 105-111.
Reinartz, W., Kumar, V., 2003. The impact of customer relationship characteristics on
profitable lifetime duration. Journal of Marketing 67 (1), 77-99.
Rosenberg, L.J., and Czepiel, J.A., 1983. A Marketing Approach for Customer Retention. The
Journal of Consumer Marketing 1 (2), 45-51.
Schweidel, D.A., Fader, P.S., Bradlow, E.T., 2008. Modeling retention within and across
cohorts. Journal of Marketing 72 (1), 82-94.
Seo, D.B., Ranganathan, C., Babad, Y., 2007. Two-level model of customer retention in the
US mobile telecommunications service market. Telecommunications Policy, 1-15.
Shah, D., Rust, R.T., Parasuraman, A., Staelin, R., Day, G.S., 2006. The path to customer
centricity. Journal of Service Research 9 (2), 113-124.
7
Page 9 of 9 ANZMAC 2009

Song, H.S., Kim, J.K., Cho, Y.B., and Kim, S.H., 2004. A personalized defection detection
and prevention procedure based on the self-organizing map and association rule mining:
applied to online game site. Artificial Intelligence Review 21 (2), 161-184.
Thomas, J.S., Reinartz, W., Kumar, V., 2004. Getting the most out of all your customers.
Harvard Business Review 82 (7/8), 116-123.
Trubik, E., and Smith M., 2000. Developing a Model of Customer Defection in the Australian Banking

Industry. Managerial Auditing Journal 15 (5), 199-215.

Van den Poel, D., Larivière, B., 2004. Customer attrition analysis for financial services using
proportional hazard models. European Journal of Operational Research 157 (1),
196-
217.
Wei, C.-P., Chiu, I.T., 2002. Turning telecommunications call details to churn prediction: a
data mining approach. Expert Systems with Applications 23 (2), 103-112.
Yan, L., Millar, D.J., Mozer, M.C., Wolniewicz, R., 2001. Improving prediction of customer
behaviour in nonstationary environments. Proceedings of the International
Joint
Conference on Neural Networks (vol 3), 2258-2263.
Zhang, Y., Qi, J., Shu, H., Li, Y., 2006. Case study on CRM: detecting likely churners with
limited information of fixed-line subscriber. Proceedings of the International
Conference
on Service Systems and Service Management (vol 2), 1495-1500.
8