ijbm in search of the internet-banking...

20
In search of the Internet-banking customer Exploring the use of decision styles Mark Durkin School of Marketing, Entrepreneurship and Strategy, University of Ulster, Belfast, Northern Ireland, UK Keywords Banking, Internet, Decision making, Consumer behaviour, Internet marketing Abstract The focus on new technologies in service situations is growing and is of particular importance in financial-services contexts. It is argued that there is mutuality of benefit for both bank and customer through the adoption of self-service technologies (SSTs), of which e-banking is but one example. Having established problems in the use of conventional segmentation methods, this paper reports on a study into Internet banking that focuses on the extent to which 480 retail-bank customers can be clustered according to an adapted decision-making framework. How such clusters can help influence the adoption of the Internet-banking interface is explored. Findings show an encouraging match between the four sample clusters identified from the case bank and the a priori classification of decision styles. This raises opportunities for the case bank’s marketing strategy in terms of offering greater insight into the motivations for the adoption of e-banking solutions within the customer base. High levels of Internet use at work are seen to positively influence e-banking registration. Introduction Technology continues to make a dramatic and profound impact in service industries and radically shapes how services are delivered (Bitner et al., 2000). The primary motivation for the increasing role of technology in service organisations has been to reduce costs and eliminate uncertainties (Kelly, 1989) as well as being used to standardise services by reducing the heterogeneity prevalent in the typical employee/customer encounter (Quinn, 1996). Such technological change raises the important question of the extent to which different customers may prefer face-to-face interaction in preference to new technology-enabled remote channels and what the influencers and inhibitors on the balance between these two interaction modes might be for different customer groups. To what extent the personalised interface could, or should, be removed from the front line in banking and for what customers is a key question for banks today (Joseph et al., 1999). This general service issue forms part of the challenge facing financial service providers and involves managing the balance between staffing levels, new technological delivery platforms, branch networks and customer preferences (Hewer et al., 2003). The recent academic focus on customer self-service technologies (SSTs) highlights the importance of exploring research issues where technology acts as a service enabler for the customer (see Gwinner et al., 1998; Bitner et al., 2000; Selnes and Hansen, 2001). The benefits of such technologies are argued to be centred around the fact that The Emerald Research Register for this journal is available at The current issue and full text archive of this journal is available at www.emeraldinsight.com/researchregister www.emeraldinsight.com/0265-2323.htm IJBM 22,7 484 Received February 2004 Revised July 2004 Accepted August 2004 The International Journal of Bank Marketing Vol. 22 No. 7, 2004 pp. 484-503 q Emerald Group Publishing Limited 0265-2323 DOI 10.1108/02652320410567917

Upload: others

Post on 10-May-2020

5 views

Category:

Documents


0 download

TRANSCRIPT

In search of the Internet-bankingcustomer

Exploring the use of decision styles

Mark DurkinSchool of Marketing, Entrepreneurship and Strategy, University of Ulster,

Belfast, Northern Ireland, UK

Keywords Banking, Internet, Decision making, Consumer behaviour, Internet marketing

Abstract The focus on new technologies in service situations is growing and is of particularimportance in financial-services contexts. It is argued that there is mutuality of benefit for bothbank and customer through the adoption of self-service technologies (SSTs), of which e-banking isbut one example. Having established problems in the use of conventional segmentation methods,this paper reports on a study into Internet banking that focuses on the extent to which 480retail-bank customers can be clustered according to an adapted decision-making framework. Howsuch clusters can help influence the adoption of the Internet-banking interface is explored. Findingsshow an encouraging match between the four sample clusters identified from the case bank and thea priori classification of decision styles. This raises opportunities for the case bank’s marketingstrategy in terms of offering greater insight into the motivations for the adoption of e-bankingsolutions within the customer base. High levels of Internet use at work are seen to positivelyinfluence e-banking registration.

IntroductionTechnology continues to make a dramatic and profound impact in service industriesand radically shapes how services are delivered (Bitner et al., 2000). The primarymotivation for the increasing role of technology in service organisations has been toreduce costs and eliminate uncertainties (Kelly, 1989) as well as being used tostandardise services by reducing the heterogeneity prevalent in the typicalemployee/customer encounter (Quinn, 1996).

Such technological change raises the important question of the extent to whichdifferent customers may prefer face-to-face interaction in preference to newtechnology-enabled remote channels and what the influencers and inhibitors on thebalance between these two interaction modes might be for different customer groups.To what extent the personalised interface could, or should, be removed from the frontline in banking and for what customers is a key question for banks today (Joseph et al.,1999). This general service issue forms part of the challenge facing financial serviceproviders and involves managing the balance between staffing levels, newtechnological delivery platforms, branch networks and customer preferences (Heweret al., 2003).

The recent academic focus on customer self-service technologies (SSTs) highlightsthe importance of exploring research issues where technology acts as a service enablerfor the customer (see Gwinner et al., 1998; Bitner et al., 2000; Selnes and Hansen, 2001).The benefits of such technologies are argued to be centred around the fact that

The Emerald Research Register for this journal is available at The current issue and full text archive of this journal is available at

www.emeraldinsight.com/researchregister www.emeraldinsight.com/0265-2323.htm

IJBM22,7

484

Received February 2004Revised July 2004Accepted August 2004

The International Journal of BankMarketingVol. 22 No. 7, 2004pp. 484-503q Emerald Group Publishing Limited0265-2323DOI 10.1108/02652320410567917

“customers can access services when and where they want without some of thecomplications of inter-personal exchanges” (Bitner et al., 2000).

Internet banking is one such SST and represents the research focus in this paper.This paper reports on a two-stage research study. The study examined the

perceptions of bankers and customers regarding the adoption of Internet banking.Stage 1 of the research was a qualitative study comprising depth interviews withsenior bank executives in Sweden, UK/Ireland and USA (see Durkin and Howcroft,2003). Stage 2 followed sequentially and involved the issue of a survey questionnaire to5,000 UK bank customers that sought to identify key influencers and inhibitors onadoption of the Internet for banking.

Prompted by the key stage 1 finding that many international bank respondentsinterviewed were unclear as to the segment profile of their Internet banking users,especially in behavioural terms, a key part of the stage 2 study focused on attemptingto identify such groups. The difficulties of effectively managing segmentation infinancial services has been reported in the literature (Machauer and Morgner, 2001;Stone et al., 2002)

A central aspect of the stage 2 study reported here was to more usefully identifycustomer groups through how they made decisions about the adoption of a newinnovation rather than by use of conventional demographic segmentation criteria(Machauer and Morgner, 2001; Smith, 2004). A review of extant research in decisionmaking based on the work of Driver et al. (1998) led to the development of an adapteddecision styles framework deemed more suitable to bank customers. Factor analysisand a cluster analysis methodology was used.

The research objective in this research is therefore to examine the extent to whichthe sample of case bank[1] customers can be classified into clusters according to atypology developed and adapted from the extant literature from Driver’s work ondecision making and decision styles. It is the findings of the factor analysis and clusteranalysis part of the research study that are reported here and which raise key issues formanagement of the case bank as they attempt to more clearly identify customermotivations for Internet banking adoption and use.

Retail banking and segmentationIn recent years banks have moved towards a marketing orientation and the adoption ofrelationship banking principles (Axson, 1992; Hollander, 1985; Berry, 1995). The keymotivators for embracing marketing principles were the competitive pressures thatarose from deregulation of the financial services market. This essentially exposedclearing banks and the retail banking market to increased competition and led to ablurring of boundaries in many traditional product markets (Howcroft, 1998).

Segmentation is a key method employed by banks to better understand and servicetheir customers in this increasingly competitive environment (Meadows and Dibb,1998a). Characteristics of the financial services sector indicate it is a suitable area formarket segmentation, most notably a diverse customer base with wide ranging needsand buyer behaviour manifest through various channel interfaces (Meadows and Dibb,1998b). While conceptually segmentation is straightforward, financial institutionshave been slower to capitalise on its potential than some other industries (McKechnieand Harrison, 1995). Speed and Smith (1992) highlight a research gap in understandinghow segmentation strategies could be implemented more effectively.

TheInternet-banking

customer

485

In their critical review of financial services segmentation, Speed and Smith (1992)suggest that a priori segmentation, which charges the researcher with determining thesize and character of segments, is the most widely used. This approach involves theuse of demographic information. In contrast, post hocmethods are less widely used andentail the grouping of respondents according to their responses to particular variables.Multivariate techniques such as cluster analysis and factor analysis can then beapplied to these responses (Meadows and Dibb, 1998b).

Increasingly there is a focus on behavioural segmentation (Elliott and Glynn, 1998;Soper, 2002). The behavioural approach contrasts with the process of segmentation basedon customer characteristics in that the focus is more driven by customer “needs”. It isargued that such a need identification approach is more robust than a classification ofcharacteristics and that it is more probable that the segments that are consequentlyidentified will be ultimately more predictive of purchase behaviour (Elliott and Glynn,1998). Nunes and Cespedes (2003, p. 99) argue that while “demographic segmentation canstill tell you what people buy, demographics no longer tell you how people shop . . . it’s apoor basis for channel design. The only rational basis is to integrate buyer behaviour”. Insupport of this approach, Smith (2004, p. 27) argues that any approach to segmentationthat is not focused on clustering customers according to their motivations “is simply anapproximation based on the assumption that descriptors (i.e. characteristics) andmotivations (i.e. needs/behaviour) are closely aligned – usually they are not”.

Machauer and Morgner (2001) propose that the a priori segmentation method(Green, 1977) and post hoc segmentation methods (Gwinn and Lindgren, 1982)currently employed reveal little of predictive use to bank marketers. In support of this,qualitative research conducted with an international sample of senior bank executivesfound a lack of clarity in terms of what market segments these bank executives feltwere best served by their Internet banking proposition and what motivations ofdifferent customer groups were in adoption (Durkin and Howcroft, 2003).

Research by Barczak et al. (1997) argues that adopters of consumer innovations differ intheir usage of technological products because of their behaviour and motivations forusage. These researchers provided the example of an on-line grocery shopping service thatmay be used by one individual through choice but another through necessity perhapsbecause of a physical disability. Thus similar usage patterns may actually hide differentmotives for use. Their study based on financial service consumers identified distinctmotivational clusters that were independent of the more established demographicsegmentation variables banks used in targeting and communicating. The researchsuggests that customer motivations may be useful in predicting their response to newproducts as well as persuading them to use existing services for the specific benefits theyvalue. The researchers conclude that all of the clusters identified needed to be informedand educated about the benefits given their own specific personal motivations formanaging money as many of the generic advertising and merchandising messagesundertaken by the bank were not picked up on by these distinct clusters. Similarly,Machauer and Morgner (2001) adopted a cluster analysis methodology as they attemptedto better understand customer perceived benefits of the bank relationship and in particularthe e-banking channel.

In summary therefore, it is evident that there is a general lack of clarity as to whatsegments of banks’ customer bases are adopting the Internet innovation for theirbanking and that the important issues of motivation and how decisions regarding

IJBM22,7

486

adoption of the e-banking platform are not well served through traditionalsegmentation and profiling tools.

Technology in bankingTechnology is making a dramatic impact upon service industries generally and thefinancial services sector is no exception. Indeed, commentators believe, that with thepossible exception of deregulation, technological change is likely to have the greatestimpact on the banking sector over the next decade (McCartan-Quinn et al., 2004;Jayawardhena and Foley, 2000). While “the infusion of new technologies in the servicessector is ubiquitous” (Lee and Allaway, 2002) there remains limited literaturedescribing studies that have been conducted with regard to the propensity and motivesof customers to use technology when interacting with their banks (Zeithaml and Gilly,1987; Moutinho and Meidan, 1989; Leblanc, 1990). This research attempts to focus onthis area of omission.

Historically research in technology adoption in banking focused on the automatedteller machine technology (ATMs) (see Marr and Prendergast, 1991, 1993). More recentresearch specific to the growth of self-service technologies (SSTs) has been conductedby Lee and Allaway (2002) who propose that a successful SST improves service firmsresource management by lowering delivery costs and by releasing service personnel toprovide better and more varied service. This point is supported by Ricard et al. (2001, p.300) who claim that SSTs can “ensure a customised service offering, help companiesrecover from service failure and are often perceived by customers as a delightfulexperience”. Lang and Colgate (2003, p. 30) argue however that technology may notalways have a positive impact on the relationship between supplier and customer andhighlight that “few authors have investigated whether the presence of IT-mediatedchannels have a detrimental effect on firms’ relationships with their customers’.

Lee (2002) highlights a need for the availability of both a “high-tech” and a“high-touch” approach where the human service dimensions are considered asimportant as the technology enabled remote service interactions. Several authors havecautioned against any attempt to replace the availability of human service interactionwith technology in various industry contexts (Pine et al., 1995; Chase, 1978; Kelly,1989). Ricard et al. (2001) highlight the difficulty in proposing a broader perspectivetowards a reconciliation of the extremes of dominantly personal or dominantly remoteinteractions. They highlight two “diametrically opposed schools of thought; onesuggests that the use of technology has a positive impact on the relationship approach. . . the other predicts a negative impact because the technology can diminish thecustomer’s interest in a relationship approach” (Ricard et al., 2001, p. 301).

The conflicting views cited in this literature point up a lack of clarity in issues ofeffectively determining interaction preferences for both banks and varying groups ofcustomers. The propensity and tolerances of the varying customer segments to embraceSSTs and for what purposes, products and services seems a key consideration in thisdebate and adds to its complexity. This, combined with the contention that “a correlationbetween conventional demographic segmentation and clients’ needs cannot be assumed”(Machauer and Morgner, 2001) points up the need for an alternative approach.

It is argued in this research that key to understanding customer motivations inembracing SSTs, specifically Internet banking in this case, is an appreciation of howcustomers make decisions in the adoption of new innovations such as the Internet.

TheInternet-banking

customer

487

Accordingly, the issue of how customers make decisions, what will motivatecustomers to make the decision to embrace the Internet banking platform, and howsuch motivations can be identified, understood and influenced, is deemed key to bankmarketers.

Decision-making processes and decision stylesAn example of decision making proposed by Driver (1979) is that of themaximiser-satisficer model – a contingency model relating to information search.Psychological research conducted over many decades by Driver indicates there aresatisficers and maximisers and that various decision styles exist within these broadcategories (Schroder et al., 1967; Driver and Streufert, 1969; Driver and Mock, 1975;Driver, 1979; Driver et al., 1996, 1998).

Decision styles refer to learned habits or patterns of decision making that resultfrom fundamental differences in information gathering and information use tendenciesamong individuals (Driver et al., 1996). Driver argues that two factors account fordecision styles:

(1) Information use: the amount of information actually considered when making adecision. In satisficing mode the minimum amount of information needed isused upon which to base a decision. In maximising mode all relevant data areexamined.

(2) Focus: the number of solutions considered. In uni-focus mode, information isused to determine only one course of action. In contrast, using information tocome up with many alternatives is the multi-focus pattern. Uni-focused peopleare usually those who have very definite ideas about how things ought to bedone. Multi-focused people tend to see more pros and cons in any course ofaction or state of affairs.

In the maximiser-satisficer model Driver combines aspects of “information use” and“focus” to propose a framework for defining five basic decision styles (Driver et al.,1996, 1998). These are illustrated in Figure 1.

The satisficer is willing to abandon the search for further information to get thedecision made. The trade off is often stated as time vs quality. The maximiser on theother hand opts for quality and will seek information until it is of no further value. As

Figure 1.Driver’s decision styles

IJBM22,7

488

indicated above “focus” relates to the number of solutions considered. Characteristicsof the five styles are now outlined:

(1) The decisive. Here use is made of the minimum amount of information in orderthat a solution can be more rapidly proposed. Decisives prize action, speed,efficiency and consistency. Once they decide on a course of action theirtendency is to stick with it. In dealings with people the hallmarks of the decisivestyle are honesty and loyalty.

(2) The flexible. Like the decisive, the flexible moves fast but the emphasis here ison adaptability. Any piece of information is seen as having several meanings orimplications. Faced with a problem requiring action, flexibles rapidly identify aline of attack; if it appears not to be working they quickly shift to a secondcourse of action. A key issue for people with this style is to keep options openand never get trapped by over-committing to any one course of action.

(3) The hierarchic. These individuals use much information to evaluate a problemand then to carefully construct a very detailed and specific plan for handling theproblem. They prize thorough analysis and quality of outcome. Hierarchicsform relationships based on mutual trust and respect and prefer deep long termfriendships to acquaintances. The relational orientation of the hierarchic maylead to a greater comfort with face-to-face interactions.

(4) The integrative. These also use much information to evaluate situations.However rather than zeroing in on a single solution their tendency is to explorea problem from many perspectives to come up with a variety of alternatives.Creativity and exploration are highly important. Methods and plans are neverfixed or final. Integratives are usually thinking on several trackssimultaneously and are particularly suited to working in groups. They thrivebest in an atmosphere of co-operation and trust.

(5) The systemic. This category is a recent addition to the Driver model. Somedecision makers make frequent use of both the integrative and hierarchic stylesand there is a two-stage decision process in evidence. Stage 1 involves thesystemic approaching a problem as would an integrative – using lots ofinformation, sizing up the situation, laying out alternatives. In stage 2, thesystemic shifts into a more hierarchic mode and orders or evaluates thealternatives according to one or more criteria. The final result is a prioritised setof strategies for dealing with the situation.

Style dominanceDriver et al. (1998) argue that until a style is used in a situation, strengths andweaknesses are merely potential strengths and weaknesses. If a particular style doesn’tfit the demands of a job, a task or a decision situation, its potential strengths don’treally matter nearly as much as its weaknesses which are no longer potential.

Key influences upon style dominance are:. environmental load;. influence of role style; and. influence of operating style.

TheInternet-banking

customer

489

Environmental load is considered to be “anything in the environment that increases aperson’s sense of pressure”. Situations having factors like time pressure, uncertainty,complexity and the potential for important consequences will create a circumstance ofhigh environmental load. When environmental load is very high or very low manypeople will use a style of satisficing or uni-focus. When load is moderate conditions areright for using one of the maximising or multi-focus styles.

Role style is heavily influenced by both national and organisational cultures.Culture determines role style because it strongly establishes value systems as to whatyou think “right” behaviour should be. These cultural forces are strongest whencircumstances force you to become aware of how you should act.

Operating style most often reflects the task demands of a particular job. The drivingforces behind operating style are subtle. Thought processes are constantly changingand being shaped by the tasks performed. Operating style can be seen as reflective ofthe cumulative effects of the task history. Doing varied or complex work seems tocreate greater complexity of thought process. Operating style is affected by education,which positively induces multi-focus thinking

In general, Driver argues that people tend to use one of these styles most frequently,but we can see a bit of each style in behaviour from time to time. People will vary inhow strongly the rely on a given style. There are no socio-economic or demographicattitudes assigned to his style classifications by Driver.

MethodologyThe research objective in this research is to examine the extent to which groups ofcustomers can be classified into clusters according to Driver’s decision-makingtypology. Respondents are bank customers and the clustering framework applied isthat of Driver as detailed above. The extent to which the case bank data set “fits” theDriver model is of interest as this will present the case bank with an alternativeframework through which to segment customers on the basis of motivation and needas regards the Internet banking proposition.

The stage 1 qualitative findings identified that a common feature of the feedbackfrom bankers interviewed in Sweden, America and the UK/Ireland was that trying topredict customer behaviour in the area of Internet banking adoption was fraught withproblems[2]. That it is difficult to identify a “typical” Internet banking user throughconventional segmentation approaches is largely unsurprising as bankers interviewedadmitted actively encouraging all customers, irrespective of relationship worth orprofile, to adopt the Internet banking proposition. The lack of clarity as to whichcustomers were adopting the e-banking platform and what their adoption decision wasmotivated by is at the heart of this research paper.

Given this it is appropriate to examine the decision-making characteristics ofrespondents to see the extent to which the extant classifications of decision-makingstyle by Driver are consistent with the sample of UK retail case bank customers. Suchan analysis may be helpful in allowing banks to better target their communication andeducation efforts more appropriately as it will overcome a reliance on conventionalsegmentation criteria and focus more on behavioural/motivational aspects of customerinteraction. It is hoped that should clusters emerge through Driver’smaximiser-satisficer decision-style model this would give the case bank anopportunity to better segment and profile customers based on their decision styles

IJBM22,7

490

and that this could in turn lead to more effective marketing for the case bank’se-banking platform.

Accordingly, an adapted version of Driver’s scale which features 26 questions thathave been derived from both the Driver-Streufert complexity index and more generalindicators of decision style characteristics as discussed in Driver et al. (1998) wasincluded in a dedicated section of the survey instrument. This comprised 26 attitudestatements with responses captured through a five-item Likert scale ranging from “Notvery characteristic of me” through to “Very characteristic of me” (see the Appendix).The main four types of decision style that were used in this research were decisive,flexible, hierarchic and integrative; the recent addition of systemic was not included asit represents a compromise state between integrative and hierarchic and it was felt ofgreater value to pursue the original distinct style clusters first determined by Driver.

The questionnaire within which the decision-style attitude statement questionswere a constituent part was issued to 5,000 UK retail bank customers. A 9.6 per centresponse rate was achieved for the overall survey, represented by 480 usableresponses.

Reliability analysis was conducted on the grouped attitude statements in each offour decision style categories. The Cronbach’s alpha scores indicate adequatereliability levels in all four style groupings with the lowest score being 0.646 and thehighest 0.711.

Having established the reliability of the decision style groupings, a decision-style scorewas allocated to each respondent. Following initial analysis however unacceptable levelsof multi-collinearity were identified between styles and this made this original approachunusable. The original attitude statements adapted from Driver were again refined andthe last four of the statements deleted as it was felt they added little to the analysis andmay have led to biased responses. A factor analysis was conducted on the remaining 22Driver attitude statements (see the Appendix for these statements). To establish thevalidity of employing factor analysis for the 22-item attitude scale the following tests wereconducted. The Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy shows afigure of 0.782 and therefore lies between “middling“ and “meritorious” as defined Hairet al. (2002). Bartlett’s test of sphericity is a statistical test for the presence of correlationsamong the variables. It provides the statistical probability that the correlation matrix hassignificant correlations among at least some of the variables. The figure was significant(chi-square ¼ 2; 380:5; p ¼ 0:000) indicating that there are correlations between thevariables and the scale is appropriate for factor analysis. Additionally, inspection of thereproduced correlation matrix shows that there are 85 (36 per cent) of non-redundantresiduals with absolute values . 0.

The SPSS output indicated that seven factors were identified and these accountedfor 65.2 per cent of the total variance. These factors are identified below with adescriptive label attributed by the researcher and the attendant attitude statementsfrom the survey.

(1) Factor 1 – face-to-face oriented:. I value long term and personally held relationships.. I am loyal and value relationships highly.. I consider myself trusting and am loyal and honest.. I don’t buy the latest technology in order to be seen to have it.

TheInternet-banking

customer

491

(2) Factor 2 – information-searching oriented:. I like to keep my options open and not risk over-committing.. I refer to others before taking any decision.. If faced with too much information I seek advice from a third party.. I value many sources of information and would analyse all before making a

decision.

(3) Factor 3 – convenience-searching oriented:. Convenience is highly important to me.. I would be willing to pay for added convenience.. I consider myself extremely short of time.. I would be willing to try out ways if I thought it would save me time.

(4) Factor 4 – creatively oriented:. I would describe myself as creative.. I would describe myself as venturesome/enthusiastic.. I explore problems from many perspectives.

(5) Factor 5 – brand oriented:. Brand is important to me.. I prefer to buy brands I know.

(6) Factor 6 – technology oriented:. I buy the latest technology when prices begin to fall.. I buy the latest technology after I can see real benefits.

(7) Factor 7 – speed of decision oriented:. I make decisions quickly but may change my mind if an alternative seems

better.. I feel pressure to make decisions quickly.

These seven new factors replace the imperfect variables derived from the originalanalysis on Driver’s decision-making variables described above and the seven newvariables were used in subsequent analysis. It must be noted that at this stage theseseven new factors represent groups of “questions” rather than groups of “customers”.To develop the customer groups a cluster analysis on the seven new factors wasundertaken.

Cluster analysisThe general objective of cluster analysis is to partition or sub-divide a set of objectsinto homogeneous sub-groups or into a hierarchical arrangement of homogeneoussub-groups. The appropriateness of cluster analysis in a financial services context hasbeen clearly established in the work of Machauer and Morgner (2001) who state that“customer segmentation by banks is limited” and that this limitation can be overcomethrough the “reference to different attitudinal dimensions concerning thecustomer-bank relationship” (Machauer and Morgner, 2001, p. 8).

IJBM22,7

492

FindingsIn the context of this research, both a hierarchical (using Ward’s method) and k-meansclustering approach were undertaken through SPSS. In using the hierarchicalapproach four main clusters were identified in the dendogram produced.

This was helpful since Driver’s decision-making classification offers four maingroupings. However, Ward’s hierarchical method represents just a single pass throughthe data and there exists the possibility that points may be joined together that withhindsight would be allocated to different clusters. With the k-means clustering approachthe test does multiple passes (iterative) through the data and so allows this concept of“hindsight” to happen throughout the process and therefore results in a fine-tuningaround the edges of the clusters. With the k-means approach the number of clusters mustbe indicated so in this case, given the Driver classifications and the preliminaryhierarchical findings, four clusters were chosen as shown in Tables I and II.

In total, 445 cases were included in the analysis (93 per cent of respondentsclassified). It was encouraging to see a good spread between the four clusters and whenthe final cluster centres were established Table III emerged.

Cluster1 2 3 4

Desire for face-to-face 5.00 4.00 1.00 5.00Desire for information 1.00 5.00 1.75 3.00Desire for convenience 4.00 5.00 1.00 3.25Creativity 1.00 5.00 2.33 4.67Brand important 5.00 5.00 5.00 1.00Desire for technology benefits 1.00 5.00 4.00 3.00Speed 5.00 5.00 2.50 1.00

Table I.Initial cluster centres

Cluster 1 129Cluster 2 174Cluster 3 39Cluster 4 103Valid 445Missing 35

Table II.Number of cases in each

cluster

Cluster1 2 3 4

Desire for face-to-face 4.61 4.57 2.27 4.57Desire for information 3.51 4.00 2.25 3.56Desire for convenience 2.93 3.68 2.56 3.12Creativity 3.33 3.93 2.31 3.75Brand important 4.23 4.20 2.29 2.30Desire for technology benefits 2.31 3.79 2.56 2.79Speed 2.58 3.47 2.68 2.67

Table III.Final cluster centres

TheInternet-banking

customer

493

The key elements from each cluster are summarised below:

(1) Cluster 1:. face-to-face oriented;. desire for information;. brand important; and. not convenience or technology oriented.

(2) Cluster 2:. face-to-face oriented;. brand important;. information desire relatively low; and. doesn’t seek speed.

(3) Cluster 3:. desire for speed;. not face-to-face oriented; and. desire for information very low.

(4) Cluster 4:. face-to-face oriented;. brand not important;. doesn’t seek speed; and. middling desire for information.

All mean scores fall between 1 and 5 as this was the range from which the questions inseven factors clustered here were originally derived (i.e. Driver’s attitude statements asdetailed in the Appendix).

On analysing the components of the four different clusters the following wasfound:

. Cluster 1. The seven factors are evenly spread throughout the 1-5 range in thiscluster. Desire for face to face important, as is brand importance and desire forinformation. Less important to this cluster is a desire for technology benefit,speed and convenience.

. Cluster 2. Unlike the case of cluster 1, here the seven factors are bunched towardsthe “very characteristic of me” end of the continuum. Again most important isdesire for face to face, followed by brand importance. While desire for speed fallslast in this cluster it is important to highlight that the score is still higher thanthat for any other cluster.

. Cluster 3. Similar to cluster 2 the seven factors are again bunched together in thiscluster but here it is towards the “not very characteristic of me” end of thecontinuum. Desire for speed was most important here while desire for face to faceand desire for information scored lowest.

. Cluster 4. As was the case with cluster 1, the seven dimensions are again spreadthroughout the 1-5 scale in this cluster. As is the case in clusters 1 and 2, desire

IJBM22,7

494

for face to face scored highest followed by creativity. The perception that brandis important is seen to be least characteristic here and desire for speed is secondlast.

These four clusters are illustrated in Figure 2.

Discussion: comparison with Driver’s classificationAs was established above Driver’s four key decision-making styles were:

(1) Decisive (unifocus, satisficer). Characterised by using a minimum of informationto make decisions, and speed being a factor in decision making. Honest andloyal; time-poor, convenience-oriented.

Figure 2.Graphic showing four new

clusters

TheInternet-banking

customer

495

(2) Hierarchic (unifocus, maximiser). Characterised by using a lot of information tomake decisions; slow to decide as they plan the implementation of the onedecision they have reached on the basis of the information.Relationship-oriented, enjoy face-to-face relationships built on mutual trust.

(3) Flexible(multi-focus, satisficer). Characterised by using a minimum ofinformation to make a decision but considering that different information hasdifferent meanings or implications. This over-consideration can slow decisionsbeing made and they may refer to a third party for help in reaching a finaldecision rather than search out more information on their own.

(4) Integrative (multi-focus, maximiser). Characterised by the use of a lot ofinformation to evaluate situations with many perspectives being consideredrather than one solution being focused upon. Creativity and exploration are keyin this process. Very adaptable. Thrive in atmospheres of co-operation andtrust.

The research objective in this research is to examine the extent to which groups ofcustomers can be classified into clusters according to Driver’s decision-makingtypology.

Figure 3 demonstrates a “threads of commonality” approach that can be drawnbetween the characteristics of Driver’s four key decision styles and the clustersidentified in this research.

While it was not expected that the clusters would exactly replicate the componentsof Driver’s decision-making model there are clear commonalities in evidence.Replicating Driver’s work was not the purpose of this study – rather it was designed tosee the extent to which an a priori model of decision making might lend itself toadoption when adjusted for bank customers and their interactions on the net.

Indeed, as Driver highlights, the manifestation of decision style can be situationspecific and when environmental load changes (e.g. additional stress, time pressures)subject style behaviour can change. Each proposed match between Driver and thecluster analysis output will now be discussed:

. Cluster 1 – proposed integratives. This cluster had 129 respondents classified.Desire for information is a common feature as is a lack of focus on technologybenefit or convenience. Decision processes are slow and all options considered.Face-to-face relationships and decisions by consensus are desired.

. Cluster 2 – proposed flexibles. This cluster had 174 respondents classified.Commonalities can be found in the relatively low level of informationrequired/desired but combined with no need to achieve a quick decision. Thedesire for face-to-face interaction and propensity to refer to a third party to helpmake the final decision are also mutually supporting.

. Cluster 3 – proposed decisives. This cluster has the least number of respondentsclassified (39) but it is here that of all the cluster comparisons the greatest degreeof commonality exists. The decisive profile of one who prizes action, makes quickdecisions and uses minimum information is reflected in the desire for speed, lowdesire for information and no need for face-to-face orientation identified in thecluster.

IJBM22,7

496

. Cluster 4 – proposed hierarchics. This cluster had 103 respondents classified.Commonalities can be found in the relatively high level of informationrequired/desired, no focus on speed as being important and an enthusiasm forface-to-face relationship building.

Each proposed match between Driver and the cluster analysis output from thisresearch study is mapped in Figure 4.

Exploring findings from cluster profilingIn an attempt to assess the extent to which each of the new clusters had a distinctdemographic profile logistic regression analysis was undertaken.

The demographic data capture questions and net usage questions in the customersurvey instrument were used as categorical independent variables as the goal of the

Figure 3.Thread of commonality

between driver andresearch clusters

TheInternet-banking

customer

497

logistic regression analysis was to identify the independent demographic variablesthat would act as predictors for membership of various clusters.

At the level of individual clusters each one was coded 1 and compared against theremaining three clusters, which were collectively grouped and coded as 2. The findingsof the logistic regression now follow:

. Cluster 1 (integrative). The key predictor which emerged was that of maritalstatus and the nature of the relationship (ExpB , 1) shows that constituents ofcluster 1 were less likely to be “single” although nothing more specific than thiscould be determined. The second predictor was “Work involves a high level ofInternet use” and the nature of the relationship indicated that this was true forclusters other than cluster 1.

. Cluster 2 (flexible). The key predictor was “Work involves a high level ofcomputer usage” and the nature of the relationship indicated that this wasincreasingly true in cluster 2 rather than the other clusters.

. Cluster 3 (decisive). The key predictor was “gender” and the nature of therelationship indicated female respondents were more likely to be classified inother clusters.

. Cluster 4 (hierarchic). The key predictor was chief income earner (CIE) and thenature of the relationship indicated that the membership of cluster 4 wasincreasingly unlikely to be the CIE.

InterpretationThe logistic regression analysis undertaken in an attempt to establish demographicand net usage predictors for cluster membership had limited success with only cluster2 demonstrating the influence of Internet use in work on membership of this cluster.

As previously identified cluster 2 had 174 respondents classified. Commonalitiescan be found in the relatively low level of information required/desired but combinedwith no need to achieve a quick decision. The desire for face-to-face interaction andpropensity to refer to a third party to help make the final decision are also mutuallysupporting. The logistic regression analysis on demographic predictors formembership of this cluster did result in “Work involves a high level of Internet use”being identified as key.

Figure 4.Research clusters mappedwith driver clusters

IJBM22,7

498

The logistic regression findings show consistently that in cluster 2 there is apositive influence in respondents’ high levels of use of the Internet at work leading totheir being registered for Internet banking. There was no influence played bydemographic profile on this. It is interesting though that the satisficing, multi-focuscharacter of the Flexible customer should demonstrate this “Internet use at work”characteristic. It would seem logical for this cluster that given the inherent multi-focusdimension that the Internet would not be the sole means used for gathering informationin making the decision quickly. Indeed, the use of the Internet in the decision-makingprocess is highly consistent with this cluster’s trait not to rely or over-commit on anyone information source, a desire for speed and the retrieval of limited information fromvarious sources relatively quickly.

Management implications and concluding remarksThe cost-effective implementation of Internet banking is of paramount importancenot only to the case bank in this study but also to the industry generally. Asestablished in the literature review, given the limitations of conventional a priorisegmentation strategies in predicting Internet banking adoption patterns thisresearch adopted a post hoc cluster analysis methodology. Customers wereclustered according to an established decision-making classification system in anattempt to identify customer groups in a more meaningful and actionable manner. Itis hoped that the case bank could then target benefits-led communicationsmessages tailored at the various cluster memberships and that this would prove aneffective approach in encouraging the adoption of the case bank’s Internet bankingproposition.

The identification of the clusters found in the sample is encouraging in thisregard. The results represent preliminary findings of a possible match betweenelements of Driver’s decision styles and the clusters identified in the case bank’ssample. While not a replication of the Driver methodology the adapted set of attitudestatements has allowed for sample case bank customers to be grouped into fourdistinct clusters each of which displays different aspects of satisficing andmaximising behaviour.

The identification of four customer clusters based on Driver’smaximiser-satisficer decision style model is therefore both interesting andstrategically important for the case bank. Key strategic issues centre on thecharacteristics of each cluster and how such characteristics can be used by the casebank to identify four key decision-style segments in the broader customer base.Given the lack of clarity in the case bank as to what the typology of customersembracing the e-banking platform actually is, the clusters identified here may be ofhelp to marketing staff in identifying and more meaningfully assessing themotivations of customers in making decisions about e-banking adoption andongoing use. Preliminary regression analysis did not uncover any key predictors forInternet banking registration in the demographic profile of respondents but thepositive influence of respondents using the Internet at work on e-bankingregistration was identified, especially among the satisficing, multi-focus cluster 2membership. Such a finding could be useful to the case bank especially in theircommunication strategy development.

TheInternet-banking

customer

499

Further research needs to be completed in order to better test the efficacy of the fourclusters identified. This could be completed by an extension of the questionnaire to awider population of bank customers.

Should the findings then be generalisable to the larger bank customer base this mayform a mechanism through which the bank can better target e-bankingcommunications towards the specific clusters of customers. An obvious examplewould be to communicate the time-saving and convenience aspects of the Internetbanking proposition to the satisficing, convenience-oriented clusters eg decisives orflexibles.

A limitation of this research is that there were insufficient responses to differentiatebetween those customers who were relationship-managed (i.e. higher net worth) andnon relationship-managed (i.e. lower net worth). This would be an important point ofdifferentiation since the bank may wish for the lower net worth base to see e-bankingas a replacement for face-to-face interaction but may wish the higher net worth clientsto see e-banking as a complement to the personalised relationship managementstrategy employed.

Further research will examine the differences between relationship managed clientsand those not relationship managed as regards their decision styles and assess theattendant impacts on Internet banking adoption behaviour.

Notes

1. The case bank in this research wishes to remain anonymous.

2. Further detail of the stage 1 methodology employed and the findings from this stage can befound in Durkin and Howcroft (2003).

References

Axson, D. (1992), “A return to managing customer relationships”, International Journal of BankMarketing, Vol. 10 No. 1, pp. 30-5.

Barczak, G., Ellen, P.S. and Pilling, B. (1997), “Developing typologies of consumer motives for useof technologically based banking services”, Journal of Business Research, Vol. 38 No. 4,pp. 131-9.

Berry, L. (1995), “Relationship marketing of services – growing interest, emerging perspectives”,Journal of the Academy of Marketing Science, Vol. 23 No. 4, pp. 236-45.

Bitner, M.J., Brown, S.W. and Meuter, M.L. (2000), “Technology infusion in service encounters”,Journal of the Academy of Marketing Science, Vol. 28 No. 1, pp. 138-49.

Chase, R.B. (1978), “Where does the customer fit in a service operation?”, Harvard BusinessReview, November/December, pp. 137-42.

Driver, M.J. (1979), “Individual decision making and creativity”, in Kerr, S. (Ed.), OrganisationalBehavior, Grid Publishing, Columbus, OH, pp. 59-94.

Driver, M.J. and Mock, T.J. (1975), “Human information processing, decision style theory andaccounting information systems”, Accounting Review, Vol. 50, pp. 490-508.

Driver, M.J. and Streufert, S. (1969), “Integrative complexity: an approach to individuals andgroups as information processing systems”, Administration Science Quarterly, Vol. 14,pp. 272-85.

Driver, M.J., Brousseau, K.R. and Hunsaker, P. (1998), The Dynamic Decision-Maker, Jossey-Bass,San Francisco, CA.

IJBM22,7

500

Driver, M.J., Svensson, K., Amato, R.P. and Pate, L.E. (1996), “A human-information processingapproach to strategic change”, International Studies of Management and Organisation,Vol. 26 No. 1, pp. 41-58.

Durkin, M. and Howcroft, J.B. (2003), “Relationship marketing in the banking sector: the impactof new technologies”, Marketing Intelligence & Planning, Vol. 21 No. 1, pp. 64-71.

Elliott, G. and Glynn, W. (1998), “Segmenting financial services markets for customerrelationships: a portfolio-based approach”, The Service Industries Journal, Vol. 18 No. 3,pp. 38-55.

Green, P.E. (1977), “A new approach to market segmentation”, Business Horizons, February,pp. 61-6.

Gwinn, J.M. and Lindgren, J.H. (1982), “Bank market segmentation: methods and strategies”,Journal of Retail Banking, Vol. 4 No. 4, pp. 8-13.

Gwinner, K.P., Grender, D.D. and Bitner, M.J. (1998), “Relational benefits in service industries: thecustomer’s perspective”, Journal of the Academy of Marketing Science, Vol. 26 No. 2,pp. 101-14.

Hair, J.F., Anderson, R., Tatham, R. and Black, W.C. (2002), Multivariate Data Analysis,Prentice-Hall, London.

Hewer, P., Howcroft, J.B. and Durkin, M. (2003), “Banker-customer interactions in financialservices”, Journal of Marketing Management, Vol. 19, pp. 1001-20.

Hollander, S. (1985), “A historical perspective on the service encounter”, in Czepiel, J.A.,Solomon, M.R. and Surprenant, C.F. (Eds), The Service Encounter: ManagingEmployee/Customer Interaction in Service Businesses, Lexington Books, Lexington, MA,pp. 49-64.

Howcroft, J. (1998), “The new retail banking revolution”, Journal of the Association ofProfessional Bankers, Vol. 8 No. 4, pp. 45-55.

Jayawardhena, C. and Foley, P. (2000), “Changes in the banking sector – the case of Internetbanking in the UK”, Internet Research: Electronic Networking Applications and Policy,Vol. 10 No. 1, pp. 19-30.

Joseph, M., McClure, C. and Joseph, B. (1999), “Service quality in the banking sector: the impact oftechnology on service delivery”, International Journal of Bank Management, Vol. 17 No. 4,pp. 54-71.

Kelly, S.W. (1989), “Efficiency in service delivery: technological or humanistic approaches?”,Journal of Services Marketing, Vol. 3 No. 3, pp. 43-50.

Lang, B. and Colgate, M. (2003), “Relationship quality, online banking and the informationtechnology gap”, International Journal of Bank Marketing, Vol. 21 No. 1, pp. 29-37.

Leblanc, G. (1990), “Customer motivations: use and non-use of automated banking”, InternationalJournal of Bank Marketing, Vol. 18 No. 4, pp. 36-40.

Lee, J. (2002), “A key to marketing financial services: the right mix of products, services, channelsand customers”, Journal of Services Marketing, Vol. 16 No. 3, pp. 238-58.

Lee, J. and Allaway, A. (2002), “Effects of personal control on adoption of self-service technologyinnovations”, Journal of Services Marketing, Vol. 16 No. 6, pp. 553-72.

McCartan-Quinn, D., Durkin, M. and O’Donnell, A. (2004), “Explaining the application of IVR:lessons from retail banking”, The Service Industries Journal, Vol. 25 No. 2.

Machauer, A. and Morgner, S. (2001), “Segmentation of bank customers by expected benefits andattitudes”, International Journal of Bank Marketing, Vol. 19 No. 1, pp. 6-17.

TheInternet-banking

customer

501

McKechnie, S. and Harrison, T. (1995), “Understanding consumers and markets”, in Ennew, C.,Watkins, T. and Wright, M. (Eds), Marketing Financial Services, Butterworth-Heinemann,Oxford.

Marr, N.E. and Prendergast, G.P. (1991), “Strategies for retailing technologies at maturity: a retailbanking case study”, Journal of International ConsumerMarketing, Vol. 3 No. 3, pp. 99-125.

Marr, N.E. and Prendergast, G.P. (1993), “Consumer adoption of self-service technologies in retailbanking”, International Journal of Bank Marketing, Vol. 11 No. 1, pp. 3-10.

Meadows, M. and Dibb, S. (1998a), “Implementing market segmentation strategies in UKpersonal financial services: problems and progress”, The Service Industries Journal, Vol. 18No. 2, pp. 45-64.

Meadows, M. and Dibb, S. (1998b), “Assessing the implementation of market segmentation inretail financial services”, International Journal of Service Quality Management, Vol. 9 No. 3,pp. 266-79.

Moutinho, L. and Meidan, A. (1989), “Bank customers’ perceptions, innovations and newtechnology”, International Journal of Bank Marketing, Vol. 7 No. 2.

Nunes, P.F. and Cespedes, F.V. (2003), “The customer has escaped”, Harvard Business Review,November, pp. 96-105.

Pine, B.J. II, Peppers, D. and Rogers, M. (1995), “Do you want to keep your customers forever?”,Harvard Business Review, Vol. 73 No. 2, pp. 103-14.

Quinn, J.B. (1996), “The productivity paradox is false: information technology improves serviceperformance”, in Swartz, A.T, Bowen, D.E. and Brown, S.W. (Eds), Advances in ServicesMarketing and Management, Vol. 5, JAI Press, Greenwich, CT.

Ricard, L., Prefontaine, L. and Sioufi, M. (2001), “New technologies and their impact on Frenchconsumer behaviour: an investigation in the banking sector”, International Journal of BankMarketing, Vol. 19 No. 7, pp. 299-311.

Schroder, H.M., Driver, M.J. and Streufert, S. (1967), Human Information Processing, Holt,Reinehart and Winston, New York, NY.

Selnes, F. and Hansen, H. (2001), “The potential hazard of self-service in developing customerloyalty”, Journal of Service Research, Vol. 4 No. 2, pp. 79-90.

Smith, B. (2004), “Getting motivated”, Marketing Business, January, pp. 26-7.

Soper, S. (2002), “The evolution of segmentation methods in financial services: where next?”,Journal of Financial Services Marketing, Vol. 7 No. 1, pp. 67-75.

Speed, R. and Smith, G. (1992), “Retail financial services segmentation”, The Service IndustriesJournal, Vol. 12, July, pp. 368-83.

Stone, M., Kiran, C., Brew, T. and Selby, D. (2002), “Managing customers in retail bankbranches”, in Foss, B. and Stone, M. (Eds), CRM in Financial Services: A Practical Guide,Kogan Page, London.

Zeithaml, V.A. and Gilly, M.C. (1987), “Characteristics affecting the acceptance of retailingtechnologies: a comparison of elderly and nonelderly consumers”, Journal of Retailing,Vol. 63 No. 1, pp. 49-68.

IJBM22,7

502

Appendix

Figure A1.

TheInternet-banking

customer

503