Market segmentation - Wharton Marketing

CHAPTER 11

Market segmentationYORAM (JERRY) WIND and DAVID R. BELL

All markets are heterogeneous. This is evident fromobservation and from the proliferation of popularbooks describing the heterogeneity of local andglobal markets. Consider, for example, The NineNations of North America (Garreau, 1982), Latitudesand Attitudes: An Atlas of American Tastes, Trends,Politics and Passions (Weiss, 1994) and MasteringGlobal Markets: Strategies for Today’s Trade Globalist(Czinkota et al., 2003). When reflecting on the natureof markets, consumer behaviour and competitiveactivities, it is obvious that no product or serviceappeals to all consumers and even those who pur-chase the same product may do so for diverse rea-sons. The Coca Cola Company, for example, varieslevels of sweetness, effervescence and package sizeaccording to local tastes and conditions. Effectivemarketing and business strategy therefore requiresa segmentation of the market into homogeneoussegments, an understanding of the needs and wantsof these segments, the design of products and serv-ices that meet those needs and development of marketing strategies, to effectively reach the targetsegments. Thus focusing on segments is at the core of organizations’ efforts to become customerdriven; it is also the key to effective resource alloca-tion and deployment. The level of segment aggre-gation is an increasingly important issue. In today’sglobal economy, the ability to customize productsand services often calls for the most micro of seg-ments: the segment of one. Following and imple-menting a market segmentation strategy allows thefirm to increase its profitability, as suggested by theclassic price discrimination model which providesthe theoretical rationale for segmentation.

Since the early 1960s, segmentation has beenviewed as a key marketing concept and has been

the focus of a significant part of the marketingresearch literature. The basic concept of segmenta-tion (as articulated, e.g. in Frank et al., 1972) hasnot been greatly altered. And many of the funda-mental approaches to segmentation research arestill valid today, albeit implemented with greatervolumes of data and some increased sophistica-tion in the modelling method. To see this, considerthe most compelling and widely used approach toproduct design and market segmentation – con-joint analysis. The essence of the approach out-lined in Wind (1978) is still evident in recent workby Toubia et al. (2007) that uses sophisticated geo-metric arguments and algorithms to improve theefficiency of the method. Other advances use for-mal economic theory to specify optimal consumertrade-offs – see Iyengar et al. (2007) for an applica-tion to non-linear pricing.

Despite the underlying stability of the basicconcept, recent advances in information technologyand the trend towards globalization are introducinga discontinuous change to the adoption and imple-mentation of segmentation strategies. The revolu-tion in information technology and strategy makespossible the creation of databases on the entire uni-verse and enormous advances in database market-ing and innovative distribution approaches. It hasalso facilitated much of the development in flexiblemanufacturing with the consequent emergence ofmass customization. In addition, the Internet hasexpanded not only the ability to implement marketsegmentation research more effectively, but alsoexpanded the portfolio of segmentation methodsavailable for use (see, e.g. Dahan and Srinivasan,2000). These changes are leading to the creation of‘one-on-one marketing’ or segments of one. The

Ch11-H8566.qxd 8/8/07 2:04 PM Page 222

Market segmentation 223

globalization of business expands the scope ofoperations and requires a new approach to local,regional and global segments. Moreover, busi-nesses that have not traditionally embraced mar-keting in general or segmentation in particular, seeit as imperative for success and even survival.Consider the current enormous effort by the lead-ing financial services firms to understand how tosegment the global at-retirement market fuelled bythe baby boomers in North America and WesternEurope. Finally, it is important to recognize a subtlebut pervasive shift in the bases for segmentation.Historically, marketers segment the market accord-ing to characteristics (e.g. demographics), prefer-ences, usage rates, etc. Increasingly, it is difficult tofully articulate a segmentation strategy without anaccompanying discussion of customer lifetimevalue (CLV) and a thought process that makes theCLV calculation explicit (see Gupta and Lehmann,2003).

These changes require not only an appraisalof what we know about segmentation, and whatworks and does not work, but also a review of thesegmentation area as part of an entirely new mar-keting and management paradigm.

Therefore, the purpose of this chapter is tointroduce the reader to both the ‘best practice’ inthe segmentation area and the likely new develop-ments. These observations are based on advancesin marketing concepts, marketing science researchand modelling tools, generalizations from empir-ical studies, successful practices of leading firmsand the conceptual implications of operating inthe global information age. This discussion of thebest segmentation practices and likely advancesencompasses five areas:

1 Use of segmentation in marketing and businessstrategy.

2 Decisions required for the implementation of asegmentation strategy.

3 Advances in segmentation research.4 Impact of operating in the global information age

on segmentation theory, practice and research.5 Expansion of segmentation to other stakeholders.

Thus, this chapter is based on the premise thatsegmentation is the firm’s response to a funda-mental market feature – heterogeneity. The likelysuccess (or otherwise) of the firm’s segmentationstrategy is assessed through a segmentation auditdiscussed next. The firm enacts the segmentationstrategy through: (1) data collection, (2) applica-tion of models and frameworks and (3) resource

allocation and differential action based on seg-ment (customer) value. The chapter concludeswith a set of critical issues that provide the guide-lines for research agenda in this area.

Use of segmentation in marketingand business strategy

Conceptually any business strategy should bebased on understanding, meeting and evenexceeding the needs of target segments. Figure 11.1illustrates the centrality of segmentation and theprogression of fundamental questions to address.At the core is the identification of the existing andpotential customer base, an understanding ofunderlying heterogeneity and the evolving needsand wants of target segments. Next is the responseto segmentation, namely guidelines for the devel-opment of products and services, and their associ-ated positioning to meet the evolving needs of thetarget segments. Finally, the product positioningprovides the foundation for the rest of the market-ing strategy and the processes, resource allocationdecisions and other activities of the firm.

Numerous published and unpublished casestudies attest to the value of segmentation. Forexample, Bell et al. (1998) show how segmentationof store choice decisions of supermarket shoppersreveals fundamental differences in store attractive-ness, conditional on a shoppers preferred shoppingstyle. The model illustrates how one store formatcan capture market share from another. It is import-ant to recognize that applications of segmentationcover a diversity of business contexts. In an indus-trial buying setting, Gensch et al. (1990) providecompelling evidence of the positive consequencesof segmentation of electrical equipment buyers. Ina 1-year test segmentation applied in two of threegeographic districts, sales increased 18 and 12 percent – while sales declined 10 per cent in the districtin which model-based segmentation was notapplied and 15 per cent for the industry. The firmreports continuous market share increases from theapplication of the segmentation approach. This isnot an isolated case. The popular business pressand the conference circuit are full of anecdotal casesin which creative segmentation has paid off. In facta growing number of firms do use segmentation asthe basis of their marketing strategy.

Yet despite the general acceptability of seg-mentation and its value, too many firms are notsegmenting their markets effectively and are not

Ch11-H8566.qxd 8/8/07 2:04 PM Page 223

224 The Marketing Book

basing their strategies on the evolving needs of tar-get segments. The experience of the more success-ful firms in consumer and industrial markets alikesuggests, however, that effective segmentation is amust. The likelihood of a positive response to thefirm’s offerings is increased, the cost of reachingcustomers and chances of new product and servicefailures are reduced. The need to ‘rediscover’ thecentrality of segmentation is made forcefully in arecent article by Yankelovich and Meer (2006).

A segmentation audit can help a firm make an initial determination as to whether it uses aneffective segmentation strategy. We propose threeinterrelated approaches in decreasing order of com-plexity and time commitment. Table 11.1 presents afairly comprehensive audit template (based on anactual audit for a large computer manufacturer). Inscoring this particular audit it is important to notethat effective segmentation requires a positiveanswer to each question. Any lower score on any ofthese dimensions requires correction.

A second approach is the ‘Five Question’ (5Q)method which requires the firm to articulate ananswer to each of the following:

1 Who are my market segments? This descriptiveapproach forces management to try and first

identity the observable characteristics ofindividuals or firms thought to be in the targetmarket segments.A firm unable to effectivelyanswer this question is likely to have considerabletrouble not only locating existing segments, butalso predicting the evolution of new marketsegments.

2 What do my segments think and feel? Anattitudinal evaluation of segments focuses onpyschographics and underlying segmentpreferences.A firm that cannot develop ananswer to this question is unlikely to be able tocommunicate effectively with target segments.Anunderstanding of perceptions, preferences andattitudes towards the firm and its competitorsare necessary conditions for being able to tailormarketing messages and talk in the appropriate‘language’.

3 How do my segments behave? This questionforces the firm to think about usage, demand andconsumption patterns, and the reaction tochanges in the marketing mix (product, price,promotion and distribution).

4 Where are my segments going? Here, the firmmust attempt to map out the trajectory ofsegment growth. All segments follow a life cycleand this question forces the firm to address

2. What product/service offerings willmeet the target segments’ needs

and offer us a sustainablecompetitive advantage?

3. What strategies and programmes, resources,capabilities, and processes are required to

develop and effectively implement theproduct/service solutions?

1. Who are thecustomers andwhat are their

needs and wants?

Figure 11.1 Focus on market-driven strategy

Ch11-H8566.qxd 8/8/07 2:04 PM Page 224


Table 11.1 A segmentation audit

Practice Completely Somewhat Does not Do not describes describes describe knowus us us at all

1 Our business strategies recognize the need to prioritize target segments

2 Our marketing plans include specific plans for each of the selected segments

3 We have specific product offerings for each target segment

4 We have a process for updating the information on our segments on an ongoing basis

5 Our segments balance the unique country needs with potential synergies across countries

6 We have an effective process for implementing segmentation research

7 We have an effective process for implementing segmentation strategies

8 We have P&L reports and accountability by segment

9 We have detailed information about segments,including:� Current size of segment� Potential size of segment� Key business needs of the segment� Information systems needs of the segment� Their prioritized needs/benefits sought� Their prioritized preference for product and

service features� Demographic characteristics of the

segments� Product/system ownership and usage� Competition’s strength in each segment� Perceived positioning of each competitor

by the members of the segment10 Information about the target market

segments are incorporated effectively into the following strategies:� Positioning� Product and service offerings� Pricing� Promotion� Advertising� Distribution� Sales force

Ch11-H8566.qxd 8/8/07 2:04 PM Page 225

dynamics, and thereby understand long-termviability of segments.

5 What are my individual customers (and segments)worth? This final question stipulates that the firmmust attempt to place a financial value on marketsegments.The inability to answer this question islikely to hamper firm efforts to effectively allocatemarketing resources and maximize marketingreturn on investment (ROI).

A third exercise which is less time intensive, butcan nevertheless reveal fundamental strengthsand weaknesses in the current segmentationstrategy is the ‘Product by Segment Matrix’ (PSM).To create the PSM, the firm simply lists as rows ofa matrix, the distinct products and services cur-rently on offer. The columns of the matrix are alist of the segments thought to be addressed bythe firm. At the very least, this exercise will pro-vide some insight into the firm’s view of marketstructure. For example, if all the matrix entries fallon the diagonal, each product is addressing a dif-ferent market segment. A ‘full row’ implies thatone of the firm’s products is serving the needs ofmultiple segments simultaneously. Such a prod-uct is clearly of key strategic importance.Similarly, a ‘full column’ implies the presence of asegment with a deep level of attachment to thefirm’s product line.

These three different but complementaryapproaches to the segmentation audit illustrateimportant principles to keep in mind. First, effect-ive segmentation requires a good deal of effort andattention. Cognitive effort, financial resources andtime commitment from top management are pre-requisites to the development of a viable segmen-tation strategy. A firm with limited commitment isunlikely to simply happen upon an effective strat-egy. Second, all good segmentation approachesseek to ‘triangulate’ – or converge on understand-ing segments through a different lens (description,attitude, behaviour, movement, value). Third, seg-mentation and the product/service portfolio mustreflect and reinforce each other.

Decisions required in implementing a segmentation strategy

While poor segmentation can result from flawedthinking or conceptions of segmentation, it is ourexperience that many segmentation efforts fall flatbecause of poor execution and implementation.

Effective segmentation strategy requires detailedanswers to the following sets of questions:

1 How to segment the market?2 What research procedure to use to develop a

segmentation strategy?3 What segment(s) to target?4 How to allocate resources among the segments?5 How to implement the segmentation strategy?

Segment identification decisionThe determination of which set of variables – basis –to use for segmentation of the market is critical.Conceptually, the guiding principle is fairly obvi-ous. A good segmentation variable is one thatexplains variation in use of the firm’s products andservices. If a proposed segmentation variable has nocorrelation to choice or other important behaviours,it is clearly of little value. Practically, the approach isquite involved and requires consideration of the fol-lowing issues. Should we segment on productusage patterns (e.g. users versus non-users or heavyversus light users)? Should we segment based onbenefits sought (e.g. product performance versusconvenience versus price sensitivity)? Should weuse some other measure of consumer response tomarketing variables (e.g. likelihood of buying anew product concept, response to price promotion,participation in a loyalty programme)? The ‘bestpractice’ in this area suggests three propositions:

1 An effective basis for segmentation should allowone to differentiate among segments based ontheir response to marketing variables; thus buyersversus non-buyers or price sensitive versus non-price sensitive are possible bases for segmentation.Age, sex, marital status, psychographiccharacteristics or other general characteristics ofthe consumer may not be good bases forsegmentation since they do not assure differentialresponse to marketing variables.

2 The selected basis for segmentation should bedirectly related to the strategic purpose of thesegmentation effort. In general, there are twotypes of segmentation with two differentunderlying rationales:� A general segmentation of the market which

allows the organization to organize itselfaround the selected target segments. As anincreasing number of companies shift from aproduct management organization to amarket-driven organization or a matrixorganization of product by market (as might


Ch11-H8566.qxd 8/8/07 2:04 PM Page 226

be implied by the PSM analysis), it is critical toidentify relatively stable and large segmentswhich could serve as strategic business units.Examples of such segments, in the case of afinancial service firm, are the ‘Delegators’ –individuals who prefer to have a moneymanager take complete control over themanagement of their financial affairs, and the‘Electronic DIY’ – who prefer to do itthemselves using direct computer trading. Toreach these two segments effectively the firmneeds distinctly different strategies. Membersof each of these strategic segments sharesome common financial/investment needs,yet each of these segments may still be quiteheterogeneous with respect to other needsand, thus, could benefit from further sub-segmentation into more homogeneousgroups.

� Specific segments for specific marketing andbusiness decisions. For example, in theintroduction of a new product, this may meana focus on the segment that has the highestlikelihood of buying the product. In the launchof a new online electronic shopping service, itcould involve a focus on time-constrainedindividuals with high-speed Internet access athome. Other specific segments and theirassociated characteristics can be developedfor each of the marketing mix decisions.Notice that in the following examples theinformational requirements (and thereforebasis for segmentation) differs systematicallyfrom issue to issue:� For positioning: product usage, product

preference, benefits sought or a hybrid ofthe variables above.

� For new product concepts (and new productintroduction): intention to buy, preferenceover current brand, benefits sought.

� For pricing decisions: price sensitivity, dealproneness, price sensitivity bypurchase/usage patterns.

� For advertising decisions: benefits sought,media usage, psychographic/lifestyle. Ahybrid of the variables above with orwithout purchase/usage patterns.

� For distribution decisions: store loyalty andpatronage, benefits sought in storeselection.

� For general understanding of a market:benefits sought, or in industrial markets,the criterion used is purchase decision,product purchase and usage patterns,

needs, brand loyalty and switching patternsor a hybrid of the above variables.

3 To gain a better understanding of the varioussegments and their characteristics, it is critical toprofile the segment’s key discriminatingcharacteristics.These include the completesegment profile on demographic, psychographic,product usage, perceptions and preferences,attitudes and the like. In the case of industrial(business-to-business) segmentation, these variablesshould include both the characteristics of theorganization and each of the key members of therelevant buying centre.Table 11.2 identifies a list ofvariables commonly used as a basis forsegmentation and as segment descriptors. It isimportant to note that variables not used as a basisfor segmentation can become descriptors ofsegments that are developed based on other bases.

Selecting a research programmeThe quality of a segmentation programme dependslargely on the quality of information used indeveloping the scheme. Segmenting any marketrequires information on the characteristics of themarket including:

� Attractiveness:The size and growth of the marketsegment(s).

� Decision making:The evaluation and choice criteriaused by individuals (or buying centres in business-to-business marketing contexts), including thebenefits they seek and problems they try to solve.

� Perceptions and preferences:The perceptions ofpreference for attitudes towards and usage of theproducts and services of the firm and itscompetitors.

� Characteristics:The demographic characteristics ofthe segment members or other relevant stablecharacteristics of the buying centre in the case ofbusiness-to-business markets.

� Attitudes/feelings:The psychographic profile(lifestyle, personality and other psychological andattitudinal characteristics).

� Responsiveness: Reaction and sensitivity to themarketing actions of the firm and its competitors.

To collect data on these variables and to analyseand interpret them requires a systematic researchprogramme. Historically, segmentation researchcentred on customer surveys. Yet there are a num-ber of additional approaches that should be con-sidered. Figure 11.2 outlines the range of options.Formal primary research and especially surveys on


Ch11-H8566.qxd 8/8/07 2:04 PM Page 227


a sample of customers and prospects have been themost common approach to segmentation research.Recent developments of databases on the entireuniverse have changed the nature of segmentationresearch. Consider, for example, the pharmaceuticalindustry. Here a number of firms are developingdatabases on the entire industry. These databasesinclude hundreds of thousands of physicians,thousands of hospitals and thousands of managedcare organizations.

A second type of formal research includes sec-ondary data analysis using geo-demographic datawhich can be used both for segmentation researchas well as for experimentation. Adaptive experi-mentation (Wind, 2007) is another approach which

could yield insights into the characteristics of vari-ous market segments and their response to mar-keting activities. One of the most importantdevelopments of recent years is the adaptation oftraditional marketing research methods for appli-cation on the Internet. Dahan and Srinivasan(2000), for example, show which conjoint analysisreplicates on the Internet. Of equal importance isthe work by Dahan and Hauser (2002) that docu-ments new research methods which are facilitatedby the unique communication capabilities affordedby the Internet.

The third approach is input from ongoingbusiness activities. In this case, we include bothdata on the market response to various strategies

Table 11.2 Variables commonly used as basis for segmentation andas descriptors of segments

Basis for segmentation Descriptors of segments

Organizational� Share� Trial� Purchase/adoption� Source loyalty� Price sensitivity� Customers of key competitors� Etc.

Buying Centre� Buying process� Informational search� Criteria/benefits sought� Negotiation style� Application� Decision� Post-purchase evaluation� Etc.

Individual� Awareness� Knowledge� Perceptions, preferences, attitudes towards

the brand� Preference� Recommendation� Purchase� Usage� Loyalty� Etc.

External� Socioeconomic, political environment (culture,

technology, economic, political, regulatory, legal)� Behaviour towards competitors� Etc.

Organizational� Industry type (e.g. SIC)� Size� Degree of centralization� Capabilities (technical, financial, etc.)� Geographic location (country, region, city, etc.)� Etc.

Buying centre� Size� Composition� Buying situation� Influence� Consensus among the members� Buying process� Buying organization and policies� Relations with suppliers� Etc.

Individual� Demographic (age, sex, family life cycle,

income, education, social class, etc.)� Psychographics (personality, lifestyle, activities,

interest opinions, etc.)� Etc.

Ch11-H8566.qxd 8/8/07 2:04 PM Page 228


and insights gained from strategic alliances withcustomers. This latter approach is especially criticalin the new product development area when thefirm works closely with its customers on the devel-opment of new products and service. “Best researchpractices” would include using a number of theseapproaches, integrating the resulting data in a data-base, continuously updating the database and hav-ing it as part of a decision-support system (DSS).

Selecting target segmentsHaving segmented a market it is often not desirableor possible to pursue all segments. Thus, a criticaldecision is the selection of target market segments.If more than one segment is to be addressed, thenconsideration needs to be given to the order inwhich each will be targeted. Moreover, all segmentshave to meet four conditions: measurability (abilityto measure the size and characteristics of the seg-ment); substantiality (having a minimum profitablesize); accessibility (ability to reach and serve thesegments) and actionability (ability to implementstrategies to serve the segments) (Kotler and Keller,2005). Beyond these fundamental requirements, theselection of target market segments requiresanswers to the questions given below. (Note therelationship between these questions for selectingsegments, and those for segmenting the marketitself, given previously.)

� Relative Attractiveness:What is the size of thesegment in terms of the revenues and profits it islikely to generate?

� Responsiveness:What is the likely response to thefirm’s offering, including response to thepositioning of the firm’s products and services?

� Competencies: Do we have the required offeringsand competencies to effectively develop, reachand serve the selected segments?

� Exogenous factors:What is the likely impact ofchanges in relevant environmental conditions (e.g.economic conditions, lifestyle, regulations, etc.) onthe potential response of the target marketsegment?

� Accessibility: Can the segment be reached (viacontrolled media and distribution or via a self-selection strategy)?

� Cost:What are the likely costs of effectivelyreaching the target segment?

� Competition:What are the current and likelycompetitive activities directed at the targetsegment?

� Portfolio management: How many segments can bemanaged effectively? What synergies orrelationships exist among the segments?

These and related criteria often require informationthat is not always available from market research.In addition, it is important that any informationcollected through marketing research or other

Database, analysis, DSS, planning simulation, resource allocation modeland/or a management subjective model using approaches such asthe analytical hierarchy process (AHP).

From strategicalliances with

end users

From theregular business

activities

Input from ongoingbusiness activities

Adaptive experimentation

Secondarydata sources

Sample Universe

Primary research-original data

sources

Formal research

Problem definition and segmentation model

Figure 11.2 Selecting a segmentation research programme

Ch11-H8566.qxd 8/8/07 2:04 PM Page 229


sources is structured in a way that is amenable toaction. Consequently, a decision-support frame-work is needed to aid managerial decision making.This framework should utilize both managerialexperience and available data. At the minimum itshould include: identifying relevant criteria, andevaluating the segments on the selected criteria.

Following the logic of product and businessportfolio analysis, a portfolio of current and poten-tial market segments can be constructed (Wind andRobertson, 1983). Figure 11.3 illustrates a portfolioof segments in which each segment is evaluated onits attractiveness, on the firm’s expected strength inthe segment and on synergy. The first two criteriaare the same dimensions used in the GE/McKinseyportfolio matrix. These dimensions can be based ona single criterion or represent a composite of mult-iple criteria. For example, segment attractivenesscould include such factors as the segment’s size andgrowth, risk factors and the cost of reaching the seg-ment. Segment strength could include such factorsas current and expected share in the segment andexpected profitability. The specific criteria and theirrelative weights can be determined by managementjudgement and marketing research input.

The segment selection procedure illustrated inFigure 11.3 utilized the analytic hierarchy process(AHP). The AHP (Wind and Saaty, 1980; Saaty,1992) is a measurement approach and process thathelps quantify management subjective judge-ments. The essential steps in implementing anAHP include: (1) setting up the decision problemby the relevant management group as a hierarchyof interrelated decision elements, (2) evaluating thevarious elements of the hierarchy by pair-wisecomparisons and (3) using a mathematical methodto estimate the relative weights of decision elem-ents. The Decision Lens (http://www.decisionlens.com) process and software greatly facilitates theimplementation of an AHP and its extension to theAnalytic Network Process (ANP).

The application of AHP or ANP to segmentselection involves bringing together the key execu-tives and presenting to them all available market-ing research information. This provides input tothe deliberation and evaluation of the varioussegments, in a pair-wise comparison, againsteach of the chosen criteria. The results for the pri-oritization of the segments from our illustrativeexample are given in Figure 11.3. An examination

Well-beingof thefirm

Long term

.20

Sales growth

.20

Reducedownside risk

.07

Profit growth

.34

Segment attractiveness

.20

Synergy

.21

Strength in segment

.51

A.10

B.21

C.11

D.02

E.03

F.18

G.02

L*.02

H*.10

I*.17

J*.01

K*.03

Short term

.80

Profit

.39

Planning horizon:

Objectives:

Criteria:

Marketsegments:

Current segments New segments

Note: The numbers are the composite priorities of each item.

Figure 11.3 An illustrative output of an AHP designed to select a portfolio of market segments

Ch11-H8566.qxd 8/8/07 2:04 PM Page 230


of this illustrative hierarchy suggests a number ofsegmentation-related conclusions:

� Management established three criteria (segmentattractiveness, strength in segment and synergy)which vary in their importance with respect tothe firm’s ability to achieve their four objectives(profit, profit growth, sales growth and downsiderisk).The objectives, in turn, vary in theirimportance under short- or long-term conditions(not shown in the figure).The overall importanceof the four objectives, assuming an 80–20 weight for short versus long term is presented inFigure 11.3.The seven current segments whenevaluated against the three criteria (segmentattractiveness, strength and synergy), which inturn are weighted by their importance to theaccomplishment of the four objectives (weightedby their importance for the short- and long-termwell-being of the firm), suggest that three of thesegments – D, E and G – are not very attractiveand should be considered as candidates fordeletion, or at least destined to receive noincremental resources.

� Five new segments were identified.Whenevaluated on the three criteria, two of thesegments – H and I – were viewed as candidatesfor inclusion in the portfolio and three – j, K andL – as candidates for exclusion.

� As a result of the process, a new portfolio ofsegments was established with segments A, B, C, F,H and I.

� The outcome also suggests how resources shouldbe allocated to each segment. Since thedimensions included in the hierarchy encompassboth the expected benefits from each segment aswell as the cost of reaching them and risk, thepriorities can be used as a rough guide forresource allocation.This would lead to thefollowing allocation:A � 11 per cent, B � 24 percent, C � 13 per cent, F � 21 per cent, H � 11per cent and I � 20 per cent.

This example reinforces the idea that the identifi-cation and selection of market segments not onlyrequires explicit structure, but also quantificationand prioritization wherever possible.

Allocating resources among segmentsThe selection of target segments and the allocationof resources among segments are interrelated anditerative decisions. For example, in the preceding

section, AHP was utilized to select a portfolio of seg-ments. It also provided guidelines for resource allo-cation among the segments. Allocation of resourcestypically involves not only the allocation amongsegments, but also the allocation of resources to themarketing mix variables – product, price, distribu-tion, promotion and advertising.

The basic problem of resource allocation is todecide on the mix of resources that generates opti-mal response (sales, profitability, etc.). Estimatingand modelling the sales response of each segment tothe various marketing resources is critical. How, forexample, should sales persons currently covering amarket be allocated across segments to optimizetheir return? A comprehensive set of analytical tools for how to answer such questions have beendeveloped in Lilien and Rangaswamy (2002). Othertraditional approaches, typically based on conjointanalysis studies among current and potential cus-tomers, have been applied to a wide range of situ-ations including computers, telecommunicationsproducts and services, pharmaceuticals, etc. (Greenand Krieger, 1985; Krieger et al., 2004). In the cases in which empirically based market response func-tions are not available, management’s subjectivejudgement, using either decision calculus methods(Little, 1979) or the AHP (Wind and Saaty, 1980), can be employed. Above all else, these approachesstress the importance of structure and managerialinput. The principles underlying successful adop-tion and execution of marketing science models areclearly articulated by Lodish (2001) aptly titledBuilding Marketing Models that Make Money. Thecomplexity and importance of the resource allocation decision suggests the advisability ofemploying a methodology such as the AHP thatincorporates managerial judgement, empirical dataderived from econometric market response modelsand experiments, consumer studies and decisionsmodels.

Implementing a segmentation strategyThe most difficult aspect of any segmentationproject is the translation of the study results intomarketing strategy and programmes. No rulescan be offered to assure a successful translationand, in fact, little is known (in the published litera-ture) on how this translation occurs. However,informal discussions with executives and obser-vations from our own experience of ‘successful’and ‘unsuccessful’ translations suggest a fewgeneral conclusions on the conditions likely toincrease successful implementation.

Ch11-H8566.qxd 8/8/07 2:04 PM Page 231

1 Involvement:All relevant decision makers (e.g.product managers, new product developers,advertising agency, sales management, etc.) mustbe involved in the problem definition, researchdesign, strategy generation and evaluation.

2 Texture:The segment characteristics used in theanalysis should be rich enough to provide thebasis for innovative product positioning andproduct, pricing, communication and distributiondecisions.

3 Direction:The segment characteristics should alsoprovide guidelines for the generation of creativeexecutions of the selected strategy.

4 Ownership: Given the oftentimes non-marketingorientation of product management, it can bedesirable to shift the responsibilities forsegmentation strategy to segment-drivenmanagers and teams.

5 Science: Since vital information on the cause andeffect relationship between marketing resourceallocation and response is often absent, it is vitalthat firms undertake to explicitly measure andmonitor outcomes. For example, linking marketresponse to media and distribution systems.

Advances in segmentation research

Many of the papers published by academics andpractitioners in Marketing Science, ManagementScience and the Journal of Marketing Research offerspecific advances in segmentation research andmodelling. Moreover, Interfaces offers in-depth casestudies of marketing science tools that demon-strably improve the bottom line for the client firm.While the approaches outlined in these articles arenecessarily diverse and sometimes focused in theirapplication, they nevertheless are centred aroundsix sets of key segmentation ‘tools’. We list the toolsbelow in approximate increasing order

Classification methodsClassification methods for segmenting markets areespecially critical for clustering-based segmentationapproaches and for hybrid designs. A clustering-based segmentation design involves determining thenumber, size and characteristics of the segmentsbased on the results of clustering of respondents ona set of ‘relevant’ variables. Benefit, need and atti-tude segmentation are examples of this type ofapproach. Significant advances have been made inthe clustering area. Hybrid segmentation design

includes those cases in which a priori segmentationdesign such as product purchase or usage, loyaltyor customer type is augmented by some cluster-based segmentation on variables such as benefitssought. Hybrid designs are especially common inbusiness-to-business segmentation.

Approaches such as macro and micro segmen-tation (Wind and Cardozo, 1974) and sequentialclustering (Peterson, 1974) are often used in prac-tice. Sequential clustering, for example, clusters onsome market-based demographics followed by atti-tude (or benefit) clustering within each demo-graphic segment. In both cluster-based design andhybrid design, the size and other characteristics(demographics, socioeconomic, purchase, etc.) ofthe segments are estimated. Classification proced-ures based on cluster analysis and variants thereofare especially useful in providing managementwith a data-driven view of segmentation. A ‘classic’illustration of this value is given in Moriarty andReibstein (1986) who use clustering methods toreveal a view of the market segments of the per-sonal computer market that challenged manage-ment intuition about the way the market wasthought to be structured. In particular, they are ableto show how benefit segmentation relates to firmdescriptors such as size and industry classification.

Discrimination methodsDiscrimination methods employed to establish theprofile of the segments commonly use multiplediscriminant analyses or regression analysis. Suchstatistical methods are often augmented withgraphic packages that graphically display the pro-file of the segments. These methods are especiallyuseful when the goal is to assign new individuals(e.g. ‘Prospects’) into groups based on their likelydisposition to use the firm’s products and services.In recent years, discriminant-based techniqueshave fallen out of favour due to an absence ofeffective solutions to implementation and inter-pretation problems such as those documented inEisenbeis (1977).

Simultaneous evaluationOne area receiving more attention from academicresearchers involves application of sophisticatedmultidimensional scaling and multivariate statis-tical analyses in order to jointly locate segments andproducts in a consistent market map. One earlydevelopment was the componential segmentationapproach proposed by Green and DeSarbo (1979).


Ch11-H8566.qxd 8/8/07 2:04 PM Page 232

This method shifts the emphasis of the segmenta-tion model from the partitioning of a market to aprediction of which person type (described by aparticular set of demographic and other psycho-graphic attribute levels) will be most responsive towhat type of product feature. Componential seg-mentation is a logical extension of conjoint analy-sis and orthogonal arrays to cover not onlyproduct features but also person characteristics. Incomponential segmentation the researcher is inter-ested in parameter values for various respondentcharacteristics (demographic, product usage, etc.)in addition to those for the product stimuli.

In a typical conjoint analysis approach to mar-ket segmentation, a matrix of subjects by utilities isdeveloped. This matrix can serve as the input tothe determination of the profile of some a priorisegments (e.g. product users versus non-users) oralternatively as the input to a clustering pro-gramme which would result in a number of bene-fit segments. In componential segmentation, thesame design principles which guide the selectionof (product) stimuli are applied also to the selec-tion of respondents. For example, in a study for anew health insurance product, four sets of respond-ent characteristics were identified on the basis ofprevious experience and management judgement:age, sex, marital status and current insurance sta-tus. Employing an orthogonal array design, onescreens respondents to select those who meet thespecific profiles specified by the design. Eachrespondent is then interviewed and administeredthe conjoint analysis task for the evaluation of a setof hypothetical health insurance products (alsoselected following an orthogonal array design).

Having completed the data collection phase,the researcher would have a matrix of averagedprofile evaluations of the product stimuli by thegroups of respondents. This data matrix is thensubmitted to any number of componential seg-mentation models, which decompose the matrixinto separate parameter values (utilities) for eachof the levels of the product feature factors (com-prising the stimulus cards) and separate param-eter values (saliences) for each of the levels of thecustomer profile characteristics (describing therespondents) which indicate how much each pro-file characteristic contributes to variation in theevaluative responses.

Given these two sets of parameters, theresearcher can make predictions about the relativeevaluation of any of the possible product featuresby any of the respondent types. The results areused with a simulation to estimate: (1) for each

respondent segment the frequency of first choicesfor each of the new product combinations con-sidered, and (2) for each new product combination, the frequency of first choices across segments.Componential segmentation offers a new conceptu-alization of market segmentation in that it focuseson the building blocks of segments and offerssimultaneously an analysis of the market segmentfor any given product offering and an evaluation ofthe most desirable product offering (or positioning)for any given segment. The concept and algorithmof componential segmentation can be extended tocover not only two data sets (product feature andrespondent characteristics) but three or more datasets by adding, for example, the components ofusage situations and distribution options.

Other studies have used the output of logitand nested logit choice models to the simultan-eous estimation of segment size, characteristics andchoice probabilities to uncover market structure.Examples of this work can be seen in Bucklin andGupta (1992) and other approaches documentedin Grover and Srinivasan (1987, 1992). Elrod and Keane (1995) use panel data to simultan-eously uncover latent attributes and consumerpreferences. More recently, Moe and Fader (2001)use a stochastic modelling approach which allowsclusters of products to draw from a fixed popula-tion of underlying segments. One advantage ofthis approach is that it simultaneously considersrates of purchase within segments and the mix ofsegments that interact with different clusters ofproducts.

A conceptually related, but methodologicallydifferent stream of research has sought under thedecision rules used by members of different mar-ket segments. Kamakura et al. (1996) illustrate thatsegments are not only heterogeneous in prefer-ences, but also in information processing and con-sideration of product attributes. In that study,members of the ‘Brand-type’ segment focus firston brand, then on product form; ‘Form-type’ seg-ment members reverse this process. Gilbride andAllenby (2004) show how to uncover and test fornon-compensatory and compensatory decisionrules using only choice data.

Simulation and optimizationSimulation and optimization are at the core of models for selection of a target portfolio of seg-ments. These models are typically based on conjointanalysis studies. Among the most powerful of theseapproaches is flexible segmentation. In contrast to


Ch11-H8566.qxd 8/8/07 2:04 PM Page 233

a priori segmentation in which the segments aredetermined at the outset of the study and cluster-ing-based segmentation in which the selected seg-ments are based on the results of the clusteringanalysis, the flexible segmentation model offers adynamic approach to the problem. Using thisapproach, one can develop and examine a largenumber of alternative segments, each composed ofthose consumers or organizations exhibiting a simi-lar pattern of responses to new ‘test’ products(defined as a specific product feature configur-ation). The flexible segmentation approach is basedon the integration of the results of a conjoint analy-sis study and a computer simulation of consumerchoice behaviour.

The simulation model in a flexible segmenta-tion approach uses three data sets:

1 Utilities for the various factors and levels for eachrespondent.

2 Perceptual ranking or rating of the currentbrands on the same set of attributes.

3 A set of demographics and other backgroundcharacteristics.

The active participation of management is alsorequired to design a set of ‘new product offerings’(each defined as a unique combination of productfeatures – specific levels on each factor included in the conjoint analysis study). Management par-ticipation can be on a real-time basis in which managers interact directly with the computer simu-lation. Research suggests that cross-functionalteams should also have input into the conjoint simu-lator approach. For example, Srinivasan et al. (1997) leverage inputs from marketing, design andmanufacturing in order to offer ‘customer-ready’prototypes to respondents.

The choice simulator is based on the assump-tion that consumers choose the offering with thehighest utility and is designed to establish the con-sumer’s share of choices among the existing brands,which can be validated against current marketshare data if available; and the consumer’s likelyswitching behaviour upon the introduction of anynew product. This phase provides a series of brand-switching matrices. Within each matrix manage-ment can select any cell or combination of cells as apossible market segment (e.g. those consumersremaining with brand i versus those who switchedto new brand, etc.). Once the desired segments (cellor cell combination) have been selected, the demo-graphic, lifestyle, product purchase and usage, and other relevant segment characteristics can be

determined by a series of multivariate statisticalanalyses which can be incorporated into the simula-tion. Some of the more significant developments inthe segmentation area in the last two decades beganwith advances in simulation and optimization pro-cedures and associated user-friendly software.Early developments by Green and Krieger (1991,1994) greatly facilitate the task of selecting an optimal (or close to optimal) set of segments, and have now been incorporated into AdaptiveConjoint Analysis (ACA) routinely implemented bySawtooth Software, one of the leading industry sup-pliers in the US (see e.g. Bryan Orme’s Introductionto Market Simulators for Conjoint Analysis). A rigor-ous practitioner-oriented account of recent develop-ments and applications can also be found in Kriegeret al. (2004) and Green et al. (2003).

One of the more promising lines of researchin simulation and optimization is emerging fromthe joint efforts of researchers in operationsresearch and marketing. Camm et al. (2006) providean exact algorithm to identify the new productconcept that will maximize the number of respon-dents for whom that product exceeds a particularutility threshold. This is an important develop-ment as prior research in marketing was only ableto identify heuristics for this task and thereforecould not guarantee global optimality.

Bayesian methodsOne important development in segmentationresearch is the application of Bayesian methods tosegmentation problems. Bayesian methods allowresearchers to estimate individual-level responseparameters and therefore connect the modellingapproach to the conceptual notion of ‘one-to-one’or ‘customized’ marketing. These methods havebecome increasingly prevalent in a host of segmen-tation-related studies that cover issues rangingfrom coupon response (Rossi et al., 1996), advertis-ing effectiveness (Schweidel et al., 2006), priceresponse and new product design (Sandor andWedel, 2005; Evgenio et al., 2005). Review articleson the impact of Bayesian methods and the implications for market segmentation are given inAllenby and Rossi (1999) and Rossi and Allenby(2003).

Customer-to-customer interactionOne fundamental shift in the practice of segmen-tation involves the harnessing of ‘customer-to-customer’ interactions for segment involvement


Ch11-H8566.qxd 8/8/07 2:04 PM Page 234

and development. Customer-to-customer intera-ctions facilitate the evolution of new segments andthe transmission of product-relevant informationand preference data. These interactions are explicitin evolving social networks (e.g., MySpace,YouTube, LinkedIn) and will be increasinglyimportant to firms of all types. Consider, for example, 1800diapers.com which has grown froma start-up with no customers to over 100,000 regis-tered customers in the space of a few months. A particularly important segment for this businessis the cluster of customers who respond to theincentive to recruit other customers. This notion ofemploying existing customers and segments tocreate new ones is analysed formally using data-mining methods (see subsequent section in thischapter) by Hill et al. (2006) in their application totelecommunications data. Among other things, theauthors find that customers linked to a prior cus-tomer adopt new services of the firm at a rate thatis three to five times higher than baseline groupsthat were selected according to the firm’s existingbest practices. The relationship between physicaland virtual proximity is also likely to influencecustomer acquisition and segmentation strategiesfor firms that have spatially dispersed clienteles. Inexamining trials of a new online grocery service,Bell and Song (2007) find that ‘neighbourhoodeffects’ play an important role in generating newcustomers, even after controlling for typical demo-graphic and other location specific variables.

Linking segmentation findings withmanagement subjective judgementsGiven the complexity of deciding on the portfolio ofsegments to target, it is helpful to use a frameworkand methodology that captures managements sub-jective judgement while allowing the incorporationof findings from various segmentation studies. TheAHP, as illustrated in Figure 11.3, and the more gen-eral ANP is ideally suited for this task.

Addressing the problemsIn addition to methodological advances in theseseven areas, some of the more interesting advancesin segmentation research are those developed toaddress three of the criticisms of segmentationresearch. Namely, that it: (1) has too narrow a focus,(2) is static and deterministic and (3) is poorly inte-grated with strategy. The advances in addressingthese problems are briefly discussed in the follow-ing section.

Too narrow a focusThis criticism encompasses five areas that can beaddressed using specific advances in modellingtechnologies:

1 The traditional focus on ‘one segment per customer’(the assignment of individuals to mutuallyexclusive and collectively exhaustive segments) istoo narrow.This problem can be overcome by theuse of overlapping clustering using a clusteringprocedure that allows for an individual to belongto more than one segment. For a discussion ofthis method, see Arabie et al. (1981) andChaturvedi et al. (1994). One could also argue thatit is more appropriate to jointly model product-segment relationships as Moe and Fader (2001).Moreover, the interrelationships among productpurchases across multiple product categories alsoprovide insight into which customers are mostworth pursuing. For an application of this methodsee Kamakura et al. (2004).

2 One segmentation scheme fits all.This problem oftrying to fit one segmentation scheme for allmarketing decisions can be solved by employingthe flexible segmentation approaches or bydeveloping a number of segmentation schemesand linking them. Moreover, the results frommultiple schemes should be complementary andconvergent.While traditional segmentationapproaches focus on uncovering differences inbehaviours and preferences, it may also beimportant to consider other possiblesegmentation approaches.This includes variationin decision-making rules or heuristics adopted bycustomers (see Gilbride and Allenby, 2006).

3 Neglect of sub-segments.With the exception ofsegments of one, most segments areheterogeneous. It is important to recognize thisand augment the basic segmentation withadditional sub-segmentation.This is the conceptunderlying hybrid segmentation models and isincreasing in its popularity. In this context, it isalso helpful to develop a hierarchy of segments.A byproduct of this research is the developmentof measures on the segmentability of each market(and the degree of homogeneity of selectedsegments). Several methods have been developedto deal with within-segment heterogeneity.Theseinclude the multi-mode Bayesian methods ofAllenby et al. (1998) and the empirical Bayesapproach of Kamakura and Wedel (2004).

4 Individual as the unit of analysis. Most segmentationstudies use data on individuals. However, few


Ch11-H8566.qxd 8/8/07 2:04 PM Page 235


decisions are made by a single individual. Mosthouseholds and industrial (business-to-business)decisions are made by a buying centre.Animportant advance in segmentation research is theshift in the unit of analysis from individuals tobuying centres, which may be a pair of individualsin a family unit, or a more complex set ofrelationships within an organization. Recognizingthe heterogeneity within a buying centre, the useof key informants as representative of the buyingcentre is often not appropriate.A better approachwould be to identify all the members of thebuying centre and collect the information on asubset of the buying centre members.This allowsan assessment of the level of consensus amongmembers of the buying centres.The level ofconsensus as well as the composition of thebuying centres can be used as a basis for, ordescriptors of, segments. Choffray and Lilien(1978) demonstrate critical differences in decisioncriteria among members of buying centres forindustrial heating and cooling equipment. Inaddition,Wilson et al. (1991) describe and test thebest ways to combine the preferences ofindividuals and buying centre members whentrying to determine how the buying centre ismost likely to act. Individual preference formationand the integration of preferences in a groupprocess are analysed in Arora and Allenby (1999).The authors show that the method also facilitatesrevelation of ‘high influence individuals’ as part ofthe estimation and segmentation process.A morerecent paper by Aribarg et al. (2002) examines theseparate processes of preference formation andconsensus building and shows how the twointerplay to produce final decisions. Finally, state ofthe art work by Arora (2006) shows that one mayin fact impute joint preference using smaller andmore cost effective datasets using ‘sub-sampling’.A formal treatment of sub-sampling – or therepeated sampling from observations which exhibitdependence – is given in Politis et al. (1999).

5 The segmentation of the month. Segmentation areahas not escaped other management fads and hashad its share of ‘segment of the month’ promiseand advocates.To avoid this trap, one mustattempt theory-driven segmentation. It also helpsto keep the focus on market response variation asbases for segmentation and to include all othervariables as segment descriptors.A recentconceptual article on ‘marketing malpractice’illustrates that segmentation schemes can becomecumbersome, a-theoretic and consequently,ineffectual (see Christensen et al., 2005).

Static and deterministic perspectiveA major limitation of many of the segmentationstudies is their neglect of the dynamic aspects of segmentation. Static/deterministic segmenta-tion tends to ignore segment change and marketdynamics; ignoring such changes often has severalconsequences. Johnson and Lilien (1994) provide aconceptualization and model-based approach todeal with segment dynamics. The Internet is oneenvironment in particular that requires specialattention to dynamics and several authors haveaddressed this issue. A recent example is given inReutterer et al. (2006) in which the authors developa dynamic model built from purchase historyinformation provided by a loyalty programme. Ingeneral, to address the issue of dynamics one cando the following:

� Define bases for segmentation that focus onchange.

� Monitor changes in segment composition over time.� Focus on strategies that will change segment

membership (from non-users to users, light toheavy users).

� Incorporate competitive actions and reactions, sincethe desirability of a segment depends not only onthe segment’s characteristics and our own strategiesbut also on competitive actions and reactions.

A second major weakness of much of the segmenta-tion research is the missing stochastic component.While early and seminal methods of segmentation(e.g. Kamakura and Russell, 1989) explicitly recog-nize that classification into segments is a probabilis-tic outcome, many applications seem to gloss overthis important fact. One new direction in this areaaddresses the ‘stochastic’ component of the seg-mentation problem directly by explicit introductionof estimates of segment size and characteristics. Theacademic research world has also examined fuzzyclustering (see, e.g. Wedel and Steenkamp, 1991;Rayward-Smith, 2002), although this method hasyet to be widely embraced and applied in the mar-keting community.

Poor integration with strategyTo address the problem that segmentation studiesare often not reflected in the resulting strategy, anumber of actions can be taken. These include:

1 Analysis. Carefully map the results of all studiessuch as copy, concept, product, distribution andother marketing studies at the segment level.To

Ch11-H8566.qxd 8/8/07 2:04 PM Page 236

the extend possible, develop empiricalgeneralizations regarding your target segments.

2 Targeting.Avoid infrequent and expensive largebase-line segmentation studies and insteadinclude in all marketing and business studies asegmentation analysis.

3 Linking. Link the segmentation to positioning andits associated marketing mix strategies. Specificallyrecognize the interdependence between the two.Given a positioning, what is the best segment(s)?Given a segment, what is the best positioning(s)?Having fixed on a segment/positioning, it is criticalto develop a marketing strategy that will meet theneeds of the selected segment and reflect thetarget positioning.

4 Implementation. Carry the segmentation efforts tothe sales force level by encouraging each salesperson to segment his/her market. In addition,segmentation methods can also be very helpful insegmenting the sales force (or other aspects ofthe distribution channel).

5 Selection. Carry through the segmentationstrategies to the business and corporate level byfocusing on a portfolio of segments and by usingthe portfolio of segments as the core of thebusiness and corporate strategy.

Segmentation in the globalinformation age

The information revolution has been the subject ofan increasing number of scholarly studies that hascaptivated the imagination of scholars, managersand the population at large. This revolution isgreatly affecting the ways in which firms are man-aging their operations and is likely to change dra-matically the way in which business is conducted.Information strategy is at the heart of most recentefforts to reengineer and reinvent the corporationand is leading to the creation of a new managementand associated marketing paradigm.

In the new management and marketing para-digm (Wind, 2005) information is having a pro-found effect on: (a) the nature and quality ofmanagement decisions; (b) the nature of businessstrategies and (c) the creation of innovative com-munication and distribution systems.

Management decisionsManagement decisions are affected by the availabil-ity and use of databases on the entire population

and their linkage to decision-(and executive-) support systems (DSS/ESS) and dashboards, whichin turn can include expert/knowledge-based sys-tems. In this context, many advances in segmenta-tion research and modelling can be incorporated inthe DSS; moreover, much of our knowledge of seg-mentation can be developed as rules for a know-ledge-based system that could help management toselect target segments.

Some of the advances in this area include theability to have ‘live’ databases in which one canupdate on an ongoing basis the customer data-base. Consider, for example, Citibank’s interactiveintelligent DSS which guides all interactions withthe customer. These interactions include the deliv-ery of a direct mail or telephone sales message tar-geted by the system which also coordinates anumber of customer interactions. The coordinatedinteractions create a ‘dialogue’ to follow the con-sumer with the appropriate intervention at eachtouch-point of contact of the consumer withCitibank. The touch points include subsequenttelephone enquiries, ATM use, bank teller inter-actions and receipt of statements. This system isbased on a new relationship model with house-hold and not only on the traditional banking focuson accounts.

Data miningData mining (e.g. Lewinson, 1994) offers enormouspotential for empirically driven insights. ‘Datamining’ refers to a number of pattern recognitionmodels that may use neural nets or fractal technol-ogy to discover patterns in the data. These methodshave been employed commercially to identify seg-ments of customers most likely to buy a givenproduct (in banking, e.g.) and to the identity of cus-tomers most likely to leave (cellular telephones,e.g.). These approaches, although intriguing, arestill in their infancy and require further validation.Some marketing researchers have combined data-mining methodologies with more traditional statis-tical models. Cooper and Giuffrida (2000), forexample, utilize data mine the residuals of a trad-itional segment-based model of response to promo-tions and use this combined model to improvesales forecasts. A notable exception in this area isthe work of Levin and Zahavi (2001) who compareseveral segmentation methods used in a predictivemodelling context – several supervised-learningsegmentation methods involving decision trees,and a couple non-supervised methods involvingjudgemental FRM and FRAT methods. This line of


Ch11-H8566.qxd 8/8/07 2:04 PM Page 237


work is promising and adds power to segmenta-tion research and modelling.

Whereas the Citibank example illustratesfuture development, much of the work todayrelies more on the traditional geo-demographicsegmentation based on consensus and other data.An example of this type of effort is the ClaritasPrizm lifestyle segmentation. This segmentationdivides the USA into 62 clusters. These clustersare further divided into 15 groups that vary withrespect to the type of location – rural, town, suburb or urban – and with respect to level ofaffluence. The Prizm lifestyle clusters can berelated to any target group of interest. Geo-demographic information continues to generatenew applications and has been shown to be espe-cially useful in customer targeting (see e.g. Sleight,2004). Much of this work also signals a new inter-face between geography and marketing and howdata sources and concepts can be combined formore effective segmentation and targeting (Harriset al., 2005).

Information strategiesInformation strategies are relatively recent add-itions to the traditional set of marketing strategies.One of the early and most effective information-based strategies was the ‘capture the client’ strat-egy of America Hospital Supply’s direct linkbetween hospital computers and AHS computers,eliminating the need for human interaction in thestraight rebuy case. ‘Capture the clients’ strategies,such as the direct computer link between P&G andWalmart, are increasing in popularity.

New communication and distributionoptionsInformation technology is also dramatically chang-ing the nature of the communication and distribu-tion options. Electronic shopping developed at amuch faster pace than was ever expected (see, e.g.Blattberg et al., 1994; Rangaswamy and Wind, 1994)and online business-to-consumer (B2C) sales in2006 in the US alone were estimated at $212 billion(US Department of Commerce).

These changes affect all aspects of our livesand are altering the concept of segmentation. Inthe new marketing paradigm the traditional massmarket is being replaced with segments includ-ing, at the extreme, segments of one. In a break-through book, The One to One Future, Peppers and

Rogers (1993) presented a vision of their one-to-one paradigm which includes and focuses on:

� share of customers not share of market;� collaborating with customers to create products

and relationships;� customizing products, services and promotional

efforts for each customer;� economics of scope;� engaging the customers in dialogues – the

interactive individualization of media is here.

The shift to segments of one requires a rethinkingof the segmentation concept as well as the develop-ment and utilization of sophisticated databases,marketing analysis, modelling and strategy. Thetrend towards such developments is inevitable andis accompanied by another discontinuous trend –the globalization of business. The early visionexemplified in the work of Peppers and Rogers hasrecently been augmented by the notion of the‘Long Tail’ – a phrase coined by Chris Andersonand popularized in his 2006 book. In the long tail amultitude of ‘low demand’ products are targeted atsmall segments of customers – such that their col-lective share outweighs that of traditional ‘block-buster’ products. Effective segmentation is at theheart of delivering in long tail markets.

GlobalizationIncreasing numbers of industries are global. To suc-ceed in this environment, firms have to shift from adomestic perspective to considering the world asthe arena of operations both with respect to the con-sumer markets for products and services as well asfor the resources markets for raw material, R&D,manufacturing, human and capital resources.

The globalization of industries is also accom-panied by trends towards regional economic inte-gration – the European Union, NAFTA and thevarious other efforts for regional integration in Asiaand Latin America. The implication of these devel-opments for segmentation is that management hasto consider portfolios of segments that include:

� global segments;� regional segments;� segments within specific countries.

Added to this complexity is the need to consider asthe unit of analysis not just countries but countriesby mode of entry, since the risk and attractivenessof a country depend on the mode of entry. The

Ch11-H8566.qxd 8/8/07 2:04 PM Page 238

selection and implementation of a portfolio of seg-ments which includes global segments, regionalsegments and segments within countries (by modeof entry) requires a significant amount of informa-tion on all relevant markets around the world. Thecreation and maintenance of such a data/know-ledge base is not a trivial undertaking and is one ofthe major obstacles to the development of globalsegmentation strategies.

Creation of processes for the development andmaintenance of country, regional and world data-bases is a high priority undertaking for all globalfirms. Yet the development of effective segmenta-tion can take place even without such databases if the firm will proceed in an iterative bottom–up and top–down segmentation. This processinvolves three bottom–up steps:

1 Segmentation of the market in each country (bymode of entry).

2 Examination of the resulting segments in all theselected countries to identify common segmentsacross countries – clustering of countrysegments.

3 The creation of a global portfolio based onvarious clusters of segments.

The resulting portfolio of segments should be com-pared to a desired (top–down) conceptual portfolioof segments. The comparison and contrast of thetwo portfolios should be driven by the concept ofglobal operation which balances the need todevelop strategies that best meet the needs of thelocal markets (given the idiosyncratic market, com-petitive and environmental conditions), while atthe same time trying to achieve economies of scaleand scope by focusing on cross-country segmentsin a number of markets. The AHP framework andmethodology can and has been used in this contextto help make such decisions, even in the absence ofthe needed ‘hard’ market data. As data becomeavailable, both from the firm’s own experience andfrom other sources, the data can be integrated in adatabase and used as input to the AHP process.

The segmentation of global markets offersenormous opportunities but is still one of theneglected areas of segmentation. It does offerintellectual and methodological challenges and iscritical from a management point of view. The lit-erature presents scattered examples of global seg-mentation. Helsen et al. (1993) offer a proposedsegmentation of multinational diffusion patterns,while Ter Hofstede et al. (2002) extend this notionto include explicit modelling of spatial aspects of

cross-national segmentation schemes. A conceptualoverview of how to think about cross-national seg-mentation is given in Steenkamp and Ter Hofstede(2002). Interestingly, very recent research by TerHofstede and Park (2007) suggests that spatial orgeographic differences across countries provideconsiderable explanatory power for segmenta-tion schemes. This result holds in an analysis ofEuropean Union consumers that controls for differ-ences in other critical variables including cultureand socioeconomic status. As global firms continueto struggle with the appropriate mix of ‘customiza-tion’ and ‘standardization’ there is enormous poten-tial for important conceptual and methodologicalbreakthroughs in this area.

Extending the segmentation concept

In the marketing literature, in practice and in thediscussion so far, segmentation has been limitedto ‘customer’ markets. Yet the concept applies toall heterogeneous populations and can andshould be extended to the other stakeholders inthe firm – all those who have a ‘stake’ in its sur-vival and growth.

Consider, for example, the firm’s own salesforce. Most large firms employ thousands of salespeople who vary considerably in their perform-ance. The 20/80 rule often applies to them as itdoes to the customers (i.e. 20 per cent of the salesforce often accounts for 80 per cent of the profits).In multi-product firms, different sales people oftentend to sell different mixes of products. They differin their family life cycle stages and hence have dif-ferent financial and time needs (some are still wor-ried about college education for their kids whileothers are single, etc.). These and other differencesamong the sales persons of any firm suggest thatthe traditional approach, in which a single salesstrategy and compensation is employed, is sub-optimal. To benefit fully from one’s sales force, it iscritical to segment it.

The segmentation of the sales force based onneeds, benefits sought, expertise, perceptions andpreference or any other relevant characteristicscould lead to the identification of homogeneoussegments and the design of separate strategiestowards them. In fact in any situation where man-agement relies heavily on a sales force a dual mar-keting strategy should be developed – one for the(target segments of) customers and a correspondingone for the (segments of the) sales force. Obviously,


Ch11-H8566.qxd 8/8/07 2:04 PM Page 239

these two strategies should be coordinated andintegrated. Furthermore, a segmented strategytowards compensation is also desirable. Toimplement it while avoiding discriminatory prac-tices requires the use of a compensation systemwith a number of options relying on a self-selec-tion strategy in which the various sales peoplecould select the option most appropriate for theirneeds. While the segmentation of the sales forceand the resulting segmented strategies are likelyto meet with considerable resistance, futureresearch needs to address whether the benefitsoutweigh the difficulties and cost involved.

Similarly, a segmentation strategy can bene-fit the firm’s dealings with its other stakeholders.Wind (1978) described a segmentation of securityanalysts and portfolio managers that led to betterunderstanding by a firm of the criteria used in theevaluation of firms in their industry and the per-ceptions of the given firm and its competitors.Following a segmentation/positioning study, astrategy was developed to meet the needs of atarget segment of security analysts that resultedin a spectacular increase in the P/E ratio of thegiven firm.

Other stakeholders such as suppliers, cus-tomer service personnel, competitors, governmentagencies and the firm’s own shareholders are oftenheterogeneous. In all of these cases, understandingthe key segments and selecting desired target seg-ment(s) can greatly enhance the firm’s effective-ness. In fact, as the cost of doing business in today’senvironment increases, a segmented strategy maybe essential for any organization concerned withthe return of their marketing investments.

Issues and associated research agenda

In one of the author’s introduction to the Journalof Marketing Research special issue – segmentationresearch (Wind, 1978), the following conclusionswere presented:

‘Market segmentation has served as the focal pointfor many of the major marketing research devel-opments and the marketing activities of mostfirms. Yet, too many segmentation researchers havesettled on a fixed way of conducting segmenta-tion studies. This tendency for standardization ofprocedures is premature and undesirable. Giventhe current state of the art, we offer the following12-question areas as particularly ripe for newundertakings’.

Of particular importance seems to be researchon the following areas:

1 Conceptualization. New conceptualizations of thesegmentation problem.

2 Theory. Re-evaluation and operationalization ofthe normative theory of segmentation withspecial emphasis on the question of how toallocate resources among markets and productsover time.

3 New variables.The discovery and implementationof new variables for use as bases forsegmentation (i.e. new attitudinal and behaviouralconstructs such as consumption-basedpersonality inventories and variables which focuson likely change in attitude and behaviouralresponses to the marketing variables) of themarkets for products, services and concepts.

4 Research design.The development of new researchdesigns and parallel data collection and analysistechniques which place fewer demands on therespondents (i.e. data collection which is simplerfor the respondent and data analysis procedurescapable of handling missing data and incompleteblock designs).

5 Analytic methods.The development of simple andflexible analytical approaches to data analysiscapable of handling discrete and continuousvariables and selected interaction at a point intime and over time.

6 Boundary conditions. Evaluation of the conditionsunder which various data analytical techniquesare most appropriate.

7 Generalizations.The accumulation of knowledgeon successful bases for segmentation acrossstudies (product, situations and markets).Thiscould entail meta-analytic work to identifysuccess drivers for different bases andmethodologies.

8 Validation. Undertaking external validation studiesto determine the performance of segmentationstrategies which were based on findings ofsegmentation studies.

9 Data generation techniques. Designing andimplementing multi-trait, multi-methodapproaches to segmentation research aimed atboth the generation of more generalizable(reliable) and valid data.

10 Cross-functional applications. Integration ofsegmentation research with the marketinginformation system of the firm.

11 Translation. Exploring alternative approaches tothe translation of segmentation findings intomarket strategies.


Ch11-H8566.qxd 8/8/07 2:04 PM Page 240

12 Implementation. Studies of the organizationaldesign of firms which were successful andunsuccessful in implementing segmentationstrategies.

As noted in this chapter, numerous innovativeapproaches to segmentation have evolved over thepast 20 years. The centrality of the segmentationconcept within the marketing field is still para-mount, yet further work on the new conceptualand methodological aspects of segmentation shouldbe undertaken. Review of some of the issues andcurrent advances in segmentation research indi-cates that despite the great advances in the man-agement of and research practice of segmentationnumerous frontiers still require creative and sys-tematic study. Furthermore, despite the advancesin academic research during the past 30 years, westill do not have satisfactory solutions for many ofthe issues raised in the 1978 research agenda(Wind, 1978); yet new issues have emerged as well. The changes in the business environment andespecially the implications of operating in a globalinformation age, the emergence of empoweredconsumers and social networks do suggest, how-ever, the need to challenge our mental models of segmentation (Wind et al., 2004) and add a fewadditional items to the research agenda. Theseinclude:

1 Problem orientation. Reconceptualization ofsegmentation problems in light of the impact ofoperating in the global information age and theemergence of segments of one seems a usefulfirst step. Many of the issues raised in this reviewfrom trends in customer-to-customer interactionto the application of sophisticated Bayesian anddata-mining methodologies suggest that strongconceptualization is still a necessary condition foreffective segmentation.

2 Decision support. Development of expertsystem/knowledge-based systems to helpmanagement to select and manage the portfolioof segments seems vital. Such systems wouldideally be incorporated in an effective DSS.Thekey to this is the development of a set of rulessummarizing our current understanding ofmarket segmentation.These rules can reflect theempirical generalizations in this area and can beaided by appropriate meta-analyses. In addition,more thought needs to be given to thepsychological and other impediments to use andimplementation (see e.g. Hoch and Schkade,1996).

3 Management and implementation. Development ofthe processes and organizational architecturerequired to assure effective adaptation andmanagement of segmentation.This includesadoption of the segmentation concept as thefoundation of all marketing and business decisions(as outlined in Figure 11.1) as well as thedevelopment of guidelines for effectivemanagement of segmentation.

4 Implicit segmentation.The effect of the Internet onwithin and across customer informationtransmission and demand aggregation isenormous (as acknowledged earlier in thischapter). It is therefore critical that firmsunderstand and utilize customer-to-customerinteractions as part of the segmentation processand allow customers to self-select and engage inmutually beneficial patterns of influence (considerthe earlier example of 1800diapers.com). Suchapproaches may be more effective and leveragedthan attempts to explicitly segment the market,regardless of the quality of the underlying dataand statistical methods.

Addressing these newer challenges and continu-ing to build our knowledge concerning the itemsidentified in the 1978 research agenda is critical inorder to increase the value of segmentation.Continuous innovation and improvement in seg-mentation research and modelling for generatingand evaluating segmentation strategies is neces-sary, but not sufficient. Real progress in this arearequires rethinking the role of segmentation in theglobal information age and concentrated efforts bymanagement to develop and implement innova-tive and effective segmentation strategies.

The obstacles to effective segmentation are notmethodological, nor even the lack of data, butrather the ability and willingness of management toundertake a segmentation strategy and establishthe processes and resources required for successfulimplementation. The concept of segmentation,once adjusted to reflect the impact of the informa-tion revolution and the globalization of business, issound and valid. It is the practice of segmentationthat is fraught with problems. These problems aresolvable but the solutions require revision andalteration to most of the current approaches used tosegment markets. If we have the conviction andcourage to re-examine the traditional segmentationconcept and approaches, we shall be able to signifi-cantly increase the value of our segmentationefforts in creating value to our customers and otherstakeholders.


Ch11-H8566.qxd 8/8/07 2:04 PM Page 241


References

Allenby, G. M., Arora, N. and Ginter, J. L. (1998)On the heterogeneity of demand, Journal ofMarketing Research, 35(3), 384–389.

Allenby, G. M. and Rossi, P. E. (1999) Marketingmodels of consumer heterogeneity, Journal ofEconometrics, 89, 57–78.

Anderson, C. (2006) The Long Tail: Why the Futureof Business is Selling Less of More, Hyperion.

Arabie, P., Douglas, C. J., DeSarbo, W. S. andWind, J. (1981) Overlapping clustering: a newmethod for product positioning, Journal of Mar-keting Research, 18, 310–317.

Aribarg, A., Arora, N. and Onur Bodur, H. (2002)Understanding the role of preference revisionand concession in group decisions, Journal ofMarketing Research, 39(3), 336–349.

Arora, N. (2006) Estimating joint preference: asubsampling approach, International Journal ofResearch in Marketing, 23(4), 409–421.

Arora, N. and Allenby, G. M. (1999) Measuringinfluence of individual preference structures ingroup decision making, Journal of MarketingResearch, 36(4), 476–487.

Blattberg, R., Glazer, R. and Little, J. (1994) TheMarketing Information Revolution, Harvard Busi-ness School Press, Harvard.

Bell, D. R., Ho, T.-H. and Tang, C. S. (1998) Deter-mining where to shop: fixed and variable costsof shopping, Journal of Marketing Research, 35(3),352–369.

Bell, D. R. and Song, S. (2007) Neighborhoodeffects and trial on the internet: evidence fromonline grocery retailing, Quantitative Marketingand Economics, forthcoming.

Bucklin, R. E. and Gupta, S. (1992) Brand choice,purchase incidence, and segmentation: an inte-grated modeling approach, Journal of MarketingResearch, XXIX, May, 201–215.

Camm, J. D., Cochran, J. J., Curry, D. J. and Kannan, S. (2006) Conjoint optimization: anexact branch-and-bound algorithm for theshare-of-choice problem, Management Science,52(3), 435–447.

Chaturvedi, A., Douglas, C. J. and Green, P. E.(1994) Market Segmentation via Overlapping K-centroids Clustering, Working Paper, WhartonSchool, University of Pennsylvania, Pennsylva-nia, May, 1–21.

Choffray, J.-M. and Lilien, G. L. (1978) A newapproach to industrial market segmentation,Sloan Management Review, 19(3), Spring, 17–30.

Christensen, C., Cook, S. and Hall, T. (2005) Mar-keting malpractice: the cause and cure, HarvardBusiness Review, HBR Reprint 2386.

Cooper, L. G. and Giuffrida, G. (2000) Turningdatamining into a management science tool,Management Science, 47(2), 249–264.

Czinkota, M. R., Ronkainen, I. A. and Donath, B.(2004) Mastering Global Markets: Strategies forToday’s Trade Globalist, South-Western, Mason,OH.

Dahan, E. and Hauser, J. R. (2002) The virtual cus-tomer, Journal of Product Innovation Manage-ment, 19(5), 332–347.

Dahan, E. and Srinivasan, V. (2000) The predictivepower of internet-based product concept test-ing using visual depiction and animation, Jour-nal of Product Innovation Management, 17,99–109.

Eisenbeis, R. A. (1977) Pitfalls in the application ofdiscriminant analysis in business, finance, andeconomics, Journal of Finance, 33(3), 875–900.

Elrod, T. and Keane, M. P. (1995) A factor-analyticprobit model for representing the market struc-ture in panel data, Journal of Marketing Research,32(1), 1–16.

Frank, R., Massy, W. and Wind, J. (1972) Market Seg-mentation, Prentice Hall, Englewood Cliffs, NJ.

Garreau, J. (1982) The Nine Nations of North Amer-ica, Houghton Mifflin, Boston.

Gensch, D., Aversa, N. and Moore, S. (1990) Achoice modeling market information systemthat enabled ABB Electric to expand its marketshare, Interfaces, 20, January–February, 6–25.

Gilbride, T. J. and Allenby, G. M. (2004) A choicemodel with conjunctive, disjunctive, and com-pensatory decision rules, Marketing Science,23(3), 391–406.

Gilbride, T. J. and Allenby, G. M. (2006) Estimatingheterogeneous EBA and economic screeningrule choice models, Marketing Science, 25(5),494–511.

Green, P. E. and DeSarbo, W. S. (1979) Componen-tial marketing in the analysis of consumertrade-offs, Journal of Marketing, 43, 83–91.

Green, P. E. and Krieger, A. M. (1985) Models andheuristics for product line selection, MarketingScience, 4, Winter, 1–19.

Green, P. E. and Krieger, A. M. (1991) Segmentingmarketing with conjoint analysis, Journal ofMarketing, 55, October, 20–31.

Green, P. E. and Krieger, A. M. (1994) An Evaluationof Alternative Approaches to Cluster-Based MarketSegmentation, Working Paper, Wharton School,University of Pennsylvania, Pennsylvania.

Ch11-H8566.qxd 8/8/07 2:04 PM Page 242

Green, P. E., Krieger, A. M. and Wind, Y. (2003)Buyer Choice Simulators, Optimizers andDynamic Models, in Wind, J. and Green, P. E.(eds), Market Research and Modeling: A Tribute toPaul E. Green, Kluwer Academic Publishers,Norwell, MA.

Grover, R. and Srinivasan, V. (1987) A simultan-eous approach to market segmentation and mar-ket structuring, Journal of Marketing Research, 24,May, 139–153.

Grover, R. and Srinivasan, V. (1992) Evaluatingthe multiple effects of retail promotions onbrand loyal and brand switching segments,Journal of Marketing Research, 29(1), 76–89.

Gupta, S. and Lehmann, D. R. (2003) Customers asassets, Journal of Interactive Marketing, 17(1), 9–24.

Harris, R., Sleight, P. and Webber, R. (2005) Geode-mographics, GIS and Neighborhood Targeting,Wiley, Hoboken, NJ.

Helsen, K., Jedidi, K. and DeSarbo, W. S. (1993) Anew approach to country segmentation utiliz-ing multinational diffusion patterns, Journal ofMarketing, 57, October, 60–71.

Hill, S., Provost, F. and Volinksy, C. (2006) Network-based marketing: identifying likelyadopters via consumer networks, Statistical Sci-ence, 22(2), 256–276.

Hoch, S. J. and Schkade, D. A. (1996) A psycho-logical approach to decision support systems,Management Science, 42(1), 51–64.

Iyengar, R., Jedidi, K. and Kohli, R. (2007) A con-joint approach to multi-part pricing, Journal ofMarketing Research, forthcoming.

Johnson, B. and Lilien, G. L. (1994) A Frameworkand Procedure for Assessing Market SegmentChange, Penn State ISBM Working Paper, PennState University, University Park, PA.

Kamakura, W. and Russell, G. J. (1989) A prob-abilistic choice model for market segmentationand elasticity structure, Journal of MarketingResearch, 26(4), 379–390.

Kamakura, W. and Wedel, M. (2004) An empiricalBayes procedure for improving individual-levelestimates and predictions from finite mixturesof multinomial logit models, Journal of Businessand Economic Statistics, 22(1), 121–125.

Kamakura, W., Kim, B.-D. and Lee, J. (1996) Mod-eling preference and structural heterogeneityin consumer choice, Marketing Science, 15(2),152–172.

Kamakura, W., Kossar, B. S. and Wedel, M. (2004)Identifying innovators for the cross-selling of new products, Management Science, 50(8),1120–1133.

Kotler, P. and Keller, K. L. (2005) MarketingManagement, Prentice Hall, Englewood Cliffs,NJ.

Levin, N. and Zahavi, J. (2001) Predictive model-ing using segmentation, Journal of InteractiveMarketing, 15(2), Spring, 2–22.

Lewinson, L. (1994) Data mining: tapping into themother code, Database Programming and Design,February, 50–56.

Lilien, G. and Rangaswamy, A. (2002) MarketingEngineering: Computer-Assisted Marketing Analy-sis and Planning, Trafford, Victoria, BC.

Little, J. D. C. (1979) Decision support systems formarketing managers, Journal of Marketing,43(3), 9–26.

Lodish, L. (2001) Building marketing models thatmake money, Interfaces, 31(3), 45–55.

Moe, W. W. and Fader, P. S. (2001) Modeling hedonic portfolio products: a joint segmentationanalysis, Journal of Marketing Research, 38(3),376–385.

Moriarty, R. T. and Reibstein, D. J. (1986) Benefitsegmentation in industrial markets, Journal ofBusiness Research, 14(6), 463–486.

Peppers, D. and Rogers, M. (1993) The One to OneFuture, Doubleday, New York.

Peterson, R. A. (1974) Market structuring bysequential cluster analysis, Journal of BusinessResearch, 2, July, 249–264.

Politis, D., Romano, J. P. and Wolf, M. (1999) Sub-sampling, Springer Series in Statistics, Springer-Verlag, New York, NY.

Rangaswamy, A. and Wind, Y. (1994) Don’t WalkIn, Just Log In! Electronic Commerce on the Infor-mation Highway, Working Paper, WhartonSchool, February.

Rayward-Smith, V. (2002) Fuzzy cluster analysis:methods for classification, data analysis andimage recognition, The Journal of the OperationalResearch Society, 51(6), 769–780.

Reutterer, T., Mild, A., Natter, M. and Taudes, A.(2006) A dynamic segmentation approach fortargeting and customizing direct marketingcampaigns, Journal of Interactive Marketing,20(3/4), 43–52.

Rossi, P. E. and Allenby, G. M. (2003) Bayesian stat-istics and marketing, Marketing Science, 22(3),304–328.

Rossi, P. E., McCulloch, R. E. and Allenby, G. M.(1996) The value of purchase history data intarget marketing, Marketing Science, 15(3),321–340.

Saaty, T. L. (1992) Decision Marketing for Leaders,RWS Pittsburgh Publications, Pittsburgh.


Ch11-H8566.qxd 8/8/07 2:04 PM Page 243

Sandor, Z. and Wedel, M. (2005) Heterogenousconjoint designs, Journal of Marketing Research,42(2), 210–220.

Schweidel, D. A., Bradlow, E. T. and Williams, P.(2006) A feature-based approach to assessingadvertising similarity, Journal of MarketingResearch, 43(2), 237–243.

Sleight, P. (2004) Targeting Customers, 3rd edn,World Advertising Research Council (WARC),London.

Srinivasan, V., Lovejoy, W. S. and Beach, D. (1997)Integrated product design for marketability andmanufacturing, Journal of Marketing Research,34(1), 154–163.

Steenkamp, J.-B. E. M. and Ter Hofstede, F. (2002)International market segmentation: issues andperspectives, International Journal of MarketingResearch, 19, 185–213.

Ter Hofstede, F., and Park, K. (2007) On the SpatialStructure of Consumer Needs: Cross-National Evi-dence from the European Union, Working Paper,Coombs School of Business, UT Austin.

Ter Hofstede, F., Wedel, M. and Steenkamp, J.-B.E. M. (2002) Identifying spatial segments ininternational markets, Marketing Science, 21(2),160–177.

Toubia, O., Hauser, J. R. and Garcia, R. (2007)Probabilistic polyhedral methods for adaptivechoice-based conjoint analysis: theory and appli-cation, Marketing Science, forthcoming.

Wedel, M. and Steenkamp, J.-B. E. M. (1991) Aclusterwise regression method for simultane-ous fuzzy market structuring and benefit

segmentation, Journal of Marketing Research,XXVIII, November, 385–396.

Weiss, M. (1994) Latitudes of Altitudes: An Atlas of American Tastes, Trends, Politics and Passions, Little Brown and Company, Boston.

Wilson, E. J., Lilien, G. L. and Wilson, D. T. (1991)Developing and testing a contingency para-digm of group choice in organizational buying,Journal of Marketing Research, 28, November,452–466.

Wind, Y. (1978) Issue and advances in segmenta-tion research, Journal of Marketing Research, XV,August, 317–337.

Wind, Y. (2007, forthcoming) The power of adap-tive experimentation, Marketing Research.

Wind, Y. and Cardozo, R. N. (1974) Industrialmarketing segmentation, Industrial MarketingManagement, March, 153–165.

Wind, Y. and Robertson, T. (1983) Marketing strat-egy: new directions for theory and research,Journal of Marketing, Spring, 12–25.

Wind, Y. and Saaty, T. L. (1980) Marketing appli-cations of the analytic hierarchy process, Man-agement Science, 26, July, 641–658.

Wind, Y., Crook, C. and Gunther, R. (2004) ThePower of Impossible Thinking: Transform the Busi-ness of Your Life and the Life of Your Business,Wharton School Publishing, Upper SaddleRiver, NJ.

Yankelovich, D. and Meer, D. (2006) Rediscover-ing market segmentation, Harvard BusinessReview, 84(2), 122–132.


Ch11-H8566.qxd 8/8/07 2:04 PM Page 244

Documents

Market segmentation - Wharton Marketing