Customer Satisfaction Measurement in the Private Bank Sector

Customer satisfaction measurement in the private bank sector

G. Mihelis a,*, E. Grigoroudis b, Y. Siskos b, Y. Politis b, Y. Malandrakis b

a Commercial Bank of Greece, Sofokleous 11, Athens 10235, Greeceb Decision Support Systems Laboratory, Technical University of Crete, University Campus, Chania 73100, Greece

Received 7 January 1999; accepted 15 March 1999

Abstract

Customer satisfaction represents a modern approach for quality in enterprises and organisations and serves the

development of a truly customer-focused management and culture. Measuring customer satisfaction o�ers an imme-

diate, meaningful and objective feedback about clientsÕ preferences and expectations. In this way, companyÕs perfor-

mance may be evaluated in relation to a set of satisfaction dimensions that indicate the strong and the weak points of a

business organisation. This paper presents an original customer satisfaction survey in the private bank sector. The

implemented methodology is based on the principles of multicriteria analysis and preference disaggregation modelling.

The most important results are focused on the determination of the critical service dimensions and the segmentation to

customer clusters with distinctive preferences and expectations. Ó 2001 Elsevier Science B.V. All rights reserved.

Keywords: Multicriteria analysis; Preference disaggregation; Ordinal regression; Customer satisfaction analysis

1. Introduction

Modern management scienceÕs philosophyconsiders customer satisfaction as a baselinestandard of performance and a possible standardof excellence for any business organisation (Ger-son, 1993). Moreover, customer satisfaction mea-surement provides a sense of achievement andaccomplishment for all employees involved in anystage of the customer service process. In this way,satisfaction measurement motivates people toperform and achieve higher levels of productivity(Hill, 1996; Wild, 1977, 1980).

To reinforce customer orientation on a day-to-day basis, a growing number of companieschoose customer satisfaction as their main per-formance indicator. It is almost impossible,however, to keep an entire company permanentlymotivated by a notion as abstract and intangibleas customer satisfaction. Therefore, customersatisfaction must be translated into a number ofmeasurable parameters directly linked to peopleÕsjob-in other words factors that people can un-derstand and in¯uence (Deschamps and Nayak,1995).

The aim of this paper is to present an originalcustomer satisfaction survey conducted in theGreek private bank sector. The methodologicalapproach is based on the principles of multicriteria

European Journal of Operational Research 130 (2001) 347±360www.elsevier.com/locate/dsw

* Corresponding author.

0377-2217/01/$ - see front matter Ó 2001 Elsevier Science B.V. All rights reserved.

PII: S 0 3 7 7 - 2 2 1 7 ( 0 0 ) 0 0 0 3 6 - 9

modelling, while a preference disaggregationmethod is used for data analysis and interpreta-tion. The survey took place in di�erent branches ofthe Commercial Bank of Greece, in order to il-lustrate benchmarking capabilities within a busi-ness organisation.

The objectives of the customer satisfactionsurvey are focused on the assessment of the criticalsatisfaction dimensions and the determination ofcustomer groups with distinctive preferences andexpectations. In particular, the purpose of thisapplication consists of a set of the most importantqueries expressed by the managers of the Com-mercial Bank of Greece:· Which are the satisfaction parameters and

which is their impact to customer behaviour?· Which is important of these factors?· How many customers are satis®ed or dissatis-

®ed?· Which is the satisfaction level according to the

characteristics of provided service?· Which is the average global satisfaction level ac-

cording to customersÕ preference and expecta-tions?

· Which are the weak and the strong points of thebank?

· Which are the satisfaction dimensions thatshould be improved and how this improvementcan be achieved?This paper is organised into 5 sections. Section

2 presents brie¯y the basic principles of the mul-ticriteria preference disaggregation approach, aswell as the implemented methodological frame.Preliminary client behavioural analysis and cus-tomer satisfaction survey designs are described in

Section 3, while the main results of the applicationare presented in Section 4. Section 5 summarisessome concluding remarks, as well as the basicadvantages of a permanent customer satisfactionmeasurement approach. Finally, a detailed pre-sentation of the used preference disaggregationmulticriteria satisfaction MUSA model (MUlti-criteria Satisfaction Analysis) is given in AppendixA, while a form of the questionnaire used in thesurvey is presented in Appendix B.

2. Preference disaggregation approach

2.1. Basic principles

The main objective of the MUSA model is theaggregation of individual judgements into a col-lective value function assuming that clientÕs globalsatisfaction depends on a set of criteria or vari-ables representing service characteristic dimen-sions (Fig. 1).

The preference disaggregation methodology isan ordinal regression based approach (Jacquet-Lagr�eze and Siskos, 1982; Siskos, 1985; Siskosand Yannacopoulos, 1985) in the ®eld of multi-criteria analysis used for the assessment of a setof marginal satisfaction functions in such a waythat the global satisfaction criterion becomes asconsistent as possible with customerÕs judge-ments.

According to the model, each customer is askedto express his/her judgements, namely his/herglobal satisfaction and his/her satisfaction withregard to the set of discrete criteria (see Siskos et al.

Fig. 1. Aggregation of customerÕs preferences.

348 G. Mihelis et al. / European Journal of Operational Research 130 (2001) 347±360

(1998) and Grigoroudis et al. (2000) for furtherdetails).

A more detailed presentation of the mathe-matical development of the model is presented inAppendix A.

2.2. Satisfaction analysis results

The main results from the aforementionedpreference disaggregation approach are focused onglobal and partial explanatory analysis. Globalexplanatory analysis lays emphasis on customersÕglobal satisfaction and its primary dimensions, asindicated in Fig. 1, while partial explanatoryanalysis focuses on each criterion and their rele-vant parameters separately.

Satisfaction analysis results, in more detail,consist of:1. Global satisfaction index: This average index

shows in a range 0±100% the level of global sat-isfaction of the customers; it may be consideredas the basic average performance indicator forthe business organisation.

2. Added value curve: This curve shows the realvalue (0±100) that customers give for each levelof the global ordinal satisfaction scale; the form

of the curve indicates if customers are demand-ing.

3. ``Fragile'' customers: The % of customers re-ceiving satisfaction value less than a particularlevel can be calculated, using the global addedvalue; this curve represents the probability dis-tribution function of the added value curve.In this way, if a particular level of the addedvalue curve is believed to be critical, the per-centage of fragile customers can be calculated.

4. Criteria/subcriteria satisfaction indices: Theseindices show in a range 0±100% the level of par-tial satisfaction of the customers according tothe speci®c criterion/subcriterion, similarly tothe global satisfaction index.

5. Weights of criteria/subcriteria: They show therelative importance within a set of criteria orsubcriteria.Combining weights and satisfaction indices, a

series of ``Performance/Importance'' diagrams canbe developed (Fig. 2). These diagrams are alsomentioned as action, decision, and strategic orperceptual maps (Customers Satisfaction Council,1995; Dutka, 1994; Naumann and Giel, 1995). Eachof these maps is divided into quadrants according toperformance (high/low), and importance (high/low), that may be used to classify actions:

Fig. 2. Performance/Importance diagram.

G. Mihelis et al. / European Journal of Operational Research 130 (2001) 347±360 349

· Status quo (low performance/low importance):Generally, no action is required.

· Leverage opportunity (high performance/highimportance): These areas can be used as advan-tage against competition.

· Transfer resources (high performance/low im-portance): CompanyÕs resources may be betterused elsewhere.

· Action opportunity (low performance/high im-portance): These are the criteria/subcriteria thatneed attention.This grid can be used in order to identify

priorities for improvement. The bottom rightquadrant is obviously the ®rst priority, for theattributes are important to customers but com-panyÕs performance is rated moderately low. Thesecond priority may be given to the satisfactioncriteria/subcriteria in the top right quadrant, es-pecially if there is room for improvement. Thethird priority issues are indicated in the bottomleft quadrant; although these issues are not ter-ribly pertinent at the time of the analysis, theymay be more important in the future, and com-panyÕs performance is certainly not good. Finally,last priority for improvement can be given to thecriteria/subcriteria in the top left quadrant be-cause this category is the least important andcompanyÕs performance is relatively good. Ap-parently, priorities for improvement can varyamong di�erent companies, depending on thepotential capabilities of improving the particularcategory.

2.3. Methodological frame

The main stages of the customer satisfactionsurvey were based on the methodology presentedin Fig. 3. This research process consists of the stepsbelow (Hayes, 1992):1. Preliminary analysis: Customer satisfaction re-

search objectives should be speci®ed in thisstage; preliminary market and customerbehavioural analysis should be conducted in or-der to assess satisfaction dimensions (custom-ersÕ consistent family of criteria).

2. Questionnaire design and conducting survey: Us-ing results from the previous step, this stage re-

fers to the development of the questionnaire,the determination of survey parameters (samplesize, collection data form, etc.) and the surveyconduction.

3. Analysis: The implementation of the model isincluded in this stage providing several resultsas described in the previous paragraph. Analy-sis is performed into the total set of customers,as well as into distinctive customer segments.Provided results involve basic descriptive statis-tical models, as well as the multicriteria prefer-ence disaggregation MUSA model.

4. Results: Using the results from the analysisstage, ®nal proposals for companyÕs improve-ment strategy can be formulated; a reliabilitytesting process for the results of the model isalso included in this stage.

3. Customer satisfaction survey

3.1. Satisfaction criteria

The assessment of a consistent family of criteriarepresenting customersÕ satisfaction dimensions isone of most important stages of the implementedmethodology, as mentioned in the previous sec-tion. This assessment can be achieved through anextensive interactive procedure between the ana-lyst and the decision-maker (company). In anycase, the reliability of the set of criteria/subcriteriahas to be tested in a small indicative set of cus-tomers.

The hierarchical structure of customersÕ satis-faction dimensions is presented in Fig. 4 and itindicates the set of criteria and subcriteria used inthis survey.

The main satisfaction criteria consist of:· Personnel of the bank: This criterion includes all

the characteristics concerning personnel (skillsand knowledge, responsiveness, communicationand collaboration with customers, friendliness,etc.).

· Products: This criterion refers mainly to the of-fered products and service (variety, refund, cost,special services, etc.).

· Image of the bank: Credibility of the bank(name, reputation), technological excellence, as


well as ability to satisfy future customersÕ needsare included in this criterion.

· Service: This criterion refers to the service of-fered to the customers; it includes the ap-pearance of the stores, the waiting time(queue, telephone, etc.), the complexity ofservice processes and the information provid-ed (informing customers in an understand-able way, explaining the service and otherrelevant factors, informing for new products,etc.).

· Access: Network expansion of the bank, branch-es location, as well as observed troubles in theservice system (strikes, damaged ATMs, etc.)are included in this criterion.

3.2. Survey conduction

The presented customer satisfaction surveytook place in two di�erent branches of the Com-mercial Bank of Greece in the area of the city ofChania. The survey was conducted within the pe-riod 25 July±30 September, 1998.

Final input data consist of 303 questionnaires:122 in store A and 181 in store B. Moreover, 160private customers and 95 companies have beenparticipated in the survey (Fig. 5).

A more detailed presentation of the generalpro®le of the sample is presented in Figs. 6±8:· Fig. 6 presents the profession of the private cus-

tomersÕ cluster, while Fig. 7 shows the activity

Fig. 3. Customer satisfaction survey process.


sector of the segment of companies. The ob-served distributions show a well-balanced sam-ple.

· Fig. 8 presents products usage per store; itshows that customers prefer to be served forloans and similar products by store A.

4. Results

4.1. Global satisfaction analysis

Customers seem to be quite satis®ed from theprovided service, given that the average global

Fig. 5. Customer sample per store.

Fig. 6. Profession of private customersÕ sample.

Fig. 7. Activity sector of companiesÕ sample.

Fig. 4. Hierarchical structure of customersÕ satisfaction dimensions.


satisfaction index has a very high value (90.1%).Moreover, criteria satisfaction analysis shows thatcustomers are quite satis®ed according to the cri-teria of ``Access'' and ``Personnel'', while lowersatisfaction indices appear for the rest of the cri-teria (65±76%), as Fig. 9 displays.

The most important criterion, with a signi®cantimportance level, seems to be Access (Fig. 10).This can justify the high value of the global satis-faction index. Customers are more satis®ed ac-cording to the most important criterion and lesssatis®ed on the dimensions that seem to play a lessimportant role to their preferences.

The added value curve for the global set ofcustomers is presented in Fig. 11. It shows thatthey do not seem to demand according to theirpreferences. The majority of customers have anadded value greater than 87%. This added valuelevel seems to be the most critical satisfaction in-dex (Fig. 12).

The analytical results for the total set of sat-isfaction dimensions (criteria and subcriteria) arepresented in Fig. 13. In this ®gure, the satisfac-tion indices represent the performance of thebank according to each satisfaction dimensions,and the weight values indicate their importancelevel.

4.2. Segmentation satisfaction analysis

The main objective of segmentation analysis isto identify particular customersÕ clusters with dis-tinctive preferences and expectations. This type ofanalysis is considered necessary, given that theimplemented preference disaggregation methodol-ogy is based on a multicriteria collective model.The discriminate variables that have been used foridentifying special groups of customers are:· the type of customer (private or company),· the visiting branch of the bank, andFig. 9. Average satisfaction indices.

Fig. 10. Weights of the criteria.

Fig. 11. Global satisfaction function (added value curve).

Fig. 8. Products usage per store.


· the usage of the o�ered banking products(cards, loans, bank-assurance, etc.).The most important distinctive results relate to

the segmentation according to the type of cus-tomers. The other groups of customers have notshown a signi®cant variation.

Analytical results for private customers and forcompanies are presented in Figs. 14 and 15, wherethe satisfaction indices with a signi®cant low valueare specially marked. These ®gures show that:· Both types of customers are not satis®ed regard-

ing the cost of the o�ered products and services.

Fig. 12. ``Fragile'' customers curve.

Fig. 13. Analytical results for the global set of customers.


Fig. 14. Analytical results for private customers.

Fig. 15. Analytical results for companies.


· Waiting time has a low satisfaction index for allcustomers.

· Special services (leasing, factoring, bank-assur-ance, etc.) and information o�ered to privatecustomers should be improved.Performance/Importance diagrams indicate

current and potentially critical satisfaction di-mensions (Figs. 16 and 17):1. Private customers are not satis®ed by the of-

fered special services, although this subcriterionis very important to them. Taken into accountthat provided information has not a high satis-faction value, it may be concluded that the cus-tomers should be informed in a moreunderstandable way. This conclusion poses amore general problem for the bank sector.

2. On the other hand, companies are usually wellinformed for the bankÕs products and services,

and this is the reason why urgently critical sat-isfaction dimensions are not observed. Never-theless, waiting time could be a potentialcritical factor in the future, given that compa-nies are quite demanding in this particularissue.

5. Concluding remarks

The original application presented in this paperillustrates the implementation of a preference di-saggregation methodology in the private banksector. The most important results include:· the determination of the weak and the strong

points of the bank,· the performance evaluation of the bank (global-

ly and per criteria/subcriteria), and

Fig. 16. Performance/Importance diagrams for private customers.

Fig. 17. Performance/Importance diagrams for companies.


· the identi®cation of distinctive critical groups ofcustomers.It should be noted that customer satisfaction is

a dynamic parameter of the business organisation.Changes in the current market can a�ect custom-ersÕ preferences and expectations. For example,some satisfaction dimensions may become criticalin the near future if customers give more impor-tance to them. For this reason, the installation of apermanent customer satisfaction barometer isconsidered necessary, given that this particularapplication was basically a pilot survey. The mainadvantages of a permanent customer satisfactionmeasurement system are:· The Commercial Bank of Greece will have the

ability to analyse customersÕ behaviour for dif-ferent regions in the country, taking into ac-count special regional characteristics.

· An interior benchmarking system can be estab-lished, based on customer satisfaction evalua-tion in each branch. In this way, the most``weak'' stores of the bank may be identi®edand improved.

· Competition analysis will be performed for dif-ferent regions of the country.

· The e�ectiveness of marketing plans will be eval-uated through customer satisfaction measure-ment.

· The establishment of a motivating system foremployees may be directly related to customersatisfaction measurement. In this way, pro-ductivity may be improved and e�cientlymeasured.A permanent customer satisfaction barometer

can assist Total Quality Management and Con-tinuous Improvement concepts in every businessorganisation (Edosomwan, 1993). Moreover, thefocus on total customer satisfaction should be in-tegrated into the accepted management processand the enterpriseÕs culture.

Acknowledgements

The authors are thankful to all the employees ofthe Commercial Bank of Greece who helped tocustomer satisfaction survey conduction. Espe-cially, the authors would like to thank the man-

agers of the bankÕs branches in Chania and theemployees of the New Product DevelopmentDepartment.

Appendix A. The MUSA method

The preference disaggregation model MUSAassesses global and partial satisfaction functionsY � and X �i , respectively given customersÕ judge-ments Y and Xi. Following the principles of ordi-nal regression analysis under constraints usinglinear programming techniques, the ordinal re-gression analysis equation is as follows:

~Y � �Pni�1

biX �i ÿ r� � rÿ;Pni�1

bi � 1;

8>><>>: �A:1�

where ~Y � is the estimation of the global satisfac-tion function, Y �, X �i are the partial satisfactionfunctions, r� and rÿ are the overestimation andthe underestimation error, respectively, and bi isthe weight of the ith criterion.

It should be noted that Y � and X �i are mono-tonic functions normalised between 0 and 100 (seeFig. 18). Also, in order to reduce the number of themathematical constraints the following transfor-mation equations are used:

Fig. 18. Satisfaction function and error variables for the jth

customer.


Fig

.19.

Qu

esti

on

nair

eo

fth

esu

rvey

.


zm � ym�1 ÿ ym for m � 1; 2; . . . ; aÿ 1;wik � bix�k�1

i ÿ bix�ki for k � 1; 2; . . . ; ai ÿ 1and i � 1; 2; . . . ; n:

8<:�A:2�

According to the aforementioned de®nitions andthe assumptions mentioned in Section 2.1, thebasic estimation model can be written in a linearprogram formulation, as follows:

�min�F �Pmj�1

r�j � rÿj ;

under the constraints

Pni�1

Pxjiÿ1

k�1

wik ÿPyjÿ1

m�1

zm ÿ r�j � rÿj � 0

for j � 1; 2; . . . ;M ;

Paÿ1

m�1

zm � 100;

Pni�1

Paiÿ1

k�1

wik � 100;

zm P 0; wik P 0 8 m; i; and k;r�j P 0; rÿj P 0 for j � 1; 2; . . . ;M ;

8>>>>>>>>>>>>>>>>>>>>>>>>>><>>>>>>>>>>>>>>>>>>>>>>>>>>:

�A:3�

where M is the number of customers, n the numberof criteria, and x�ji , y�j are the jth level on whichvariables Xi and Y are estimated.

The preference disaggregation methodologyalso consists of a post optimality analysis stage inorder to face the problem of model stability. The®nal solution is obtained by exploring the poly-hedron of near optimal solutions, which is gener-ated by the constraints of the above linearprogram. This solution is calculated by n linearprograms (equal to the number of criteria) of thefollowing form:

max� �F 0 � Paiÿ1

k�1

wik for i � 1; 2; . . . ; n;

under the constraints

F 6 F � � e;

all the constraints of LP �A:3�;

8>>>>>>>>><>>>>>>>>>:�A:4�

where e is a small percentage of F�. The average ofthe solutions given by the n LPs (3) may be takenas the ®nal solution. In case of non-stability thisaverage solution is less representative.

The assessment of a performance norm may bevery useful in customer satisfaction analysis. Theaverage global and partial satisfaction indices areused for this purpose and can be assessed ac-cording to the following equations:

S �Xa

m�1

pmy�m and si �Xai

k�1

pki x�ki ; �A:5�

where S and si are the average global and partialsatisfaction indices, and pm and pk

i are the fre-quencies of customers belonging to the ym and xk

isatisfaction level, respectively.

Appendix B. Questionnaire of the survey

Fig. 19 presents the questionnaire of the survey.

References

Customers Satisfaction Council, 1995. Customer Satisfaction

Assessment Guide, Motorola University Press, Schaum-

burg, IL.

Deschamps, J.P., Nayak, P.R., 1995. Product Juggernauts:

How Companies Mobilize to Generate a Stream of Market

Winners. Harvard Business School Press, Boston.

Dutka, A., 1994. AMA Handbook of Customer Satisfaction: A

Complete Guide to Research Planning and Implementation.

NTC Business Books, Lincolnwood, IL.

Edosomwan, J.A., 1993. Customer and Market-Driven Quality

Management. ASQC Quality Press, Milwaukee, WI.

Gerson, R.F., 1993. Measuring Customer Satisfaction. Menlo

Park, CA.

Grigoroudis, E., Siskos, Y., Saurais, O., 2000. TELOS: A

customer satisfaction evaluation software. Computers and

Operations Research 27, 799±817.

Hayes, B.E., 1992. Measuring Customer Satisfaction: Develop-

ment and Use of Questionnaire. ASQC Quality Press,

Milwaukee, WI.

Hill, N., 1996. Handbook of Customer Satisfaction Measure-

ment. Gower Press, Aldershot, UK.

Jacquet-Lagr�eze, E., Siskos, J., 1995. Assessing a set of additive

utility functions for multicriteria decision-making: The

UTA method. European Journal of Operational Research

10 (2), 151±164.


Naumann, E., Giel, K., 1995. Customer satisfaction measure-

ment and management: using the voice of the customer.

Thomson Executive Press, Cincinnati, OH.

Siskos, J., 1985. Analyse de regression et programmation

lin�eaire. Revue de Statistique Appliqu�ee 23 (2), 41±55.

Siskos, J., Yannacopoulos, D., 1985. Utastar: An ordinal

regression method for building additive value functions.

Investigacßao Operacional 5 (1), 39±53.

Siskos, Y., Grigoroudis, E., Zopounidis, C., Saurais, O., 1998.

Measuring customer satisfaction using a collective prefer-

ence disaggregation model. Journal of Global Optimization

12, 175±195.

Wild, R., 1977. Concepts for Operations Management. Wiley,

New York.

Wild, R., 1980. Operations Management: A Policy Framework.

Pergamon Press, New York.


Customer Satisfaction Measurement in the Private Bank Sector

Documents

Transcript of Customer Satisfaction Measurement in the Private Bank Sector