“Segmenting Industrial Buyers by Loyalty and Value”

The process of “market segmentation” is a cornerstone of modern marketing and an essential task for relatively smaller seller organisations as they address themselves to markets much larger and more diverse than their limited output capacity. In mass consumer markets the process is, arguably, well developed and sophisticated. In industrial markets, it is generally acknowledged that the process is less well understood and even less well practised. The obvious diversity of industrial markets and frequently idiosyncratic buyer-seller relationships explain, in part, this lack of generalisability. This paper proposes a methodology for industrial market segmentation which is based on the joint consideration, and explicit calculation, of both the value of the customer to the seller, and the value of the seller to the customer. Implicit in the process of measuring value are the central concepts of relationships, and customer loyalty. Introduction The fundamentally important process of market segmentation is well understood and widely documented in relation to mass consumer markets. However, in industrial or business-tobusiness markets, the subject has received less discussion and / or critical evaluation. The aim of this paper is to develop a methodology for segmenting industrial markets in a way which reflects: (1) the value of the customer to the seller, and (2) the value of the seller to the customer. Implicit in the process of measuring value are the central concepts of relationships and customer loyalty. Market Segmentation The concept of market segmentation is today a central tenet of modern marketing so that its origins and principles are seldom critically examined. The origins of the term can be traced back to Wendel Smith’s trailblazing article of the 1950’s which introduced the concept as being ‘based upon developments on the demand side of the market and represents a more precise adjustment of product and marketing effort to consumer and user requirements’ (Smith, 1956). In support of his position, Smith pointed to the variety inherent in contemporary markets among users and suppliers of products. Market segmentation was identified as a prerequisite for any organisation endeavouring to create products or services fitting customers’ needs. Smith’s market segmentation concept was rapidly adopted in theory and practice and was further developed by a number of researchers in the sixties and seventies (Yankelovich, 1964; Barnett, 1969; Myers and Tauber, 1977; Wilkie and Cohen, 1977). By the end of the seventies, a number of authors were able to confidently assert generalised market segmentation principles. In a key summary article, Wind declared market segmentation to be "...a dominant concept in marketing literature and practice" (Wind, 1978). The concept of market segmentation remains a central tenet of modern marketing and, arguably, is linked inextricably to the marketing concept itself (Neidell, 1983). As organisations move from serving individual customers to addressing larger markets, the focus shifts from identifying and satisfying the needs of individual customers to the task of satisfying the needs of target market segments. This is inevitable as the firm typically only has a limited ability to match the variety of demand with a corresponding variety of supply (Ashby, 1960). Put simply, most organisations serving mass markets have only a finite ability to vary their offer before production or cost constraints prevail. Several authors have complained that there is no generally accepted and validated way to segment markets (Beane and Ennis, 1987; Schauerman, 1990). However, four forms of market segmentation have emerged as the most popular: • Geographic Market Segmentation: segmenting markets by geographic region, population density or climate. • Demographic Market Segmentation: variables including age, sex, size and family type, income, educational level, race and nationality or combinations thereof. The validity of using demographic variables is well supported in the literature (Frank, Massey and Wind, 1972). Clearly defined segments can be identified using demographic variables, but entire markets cannot usually be segmented by this method alone (Beane and Ennis, 1987). • Psychographic Market Segmentation: involves the less easily measured social class and way of living or life-style variables. This form of segmentation attempts to incorporate part of the inner person or their underlying motivations into the understanding of the market. Psychographic segmentation has proved superior to demographic segmentation alone (Plummer, 1974; Wells, 1975). • Behaviouristic (sic.) Market Segmentation: This form of segmentation includes variables such as purchase occasion, benefits sought, user status, degree of usage and loyalty, buyer readiness stage, and marketing focus sensitivity. Knowledge of product, attitude and response to the product are used to segment consumers (McDonald and Goldman, 1979). A combination of psychographic and behavioural segmentation has also been proposed as a means of segmenting markets on the basis of the consumer's self-image or self-concept and its relationship to the image of the product (Sirgy, 1982). An important, but seldom discussed, issue is that behavioural segmentation variables differ from geographic, demographic or psychographic approaches in being based on consumer responses (as distinct from consumer characteristics); ie based on what they do, rather than 1 This situation may not apply in all cases, however. The characteristics of professional services, and, in particular, their almost infinite heterogeneity, provide an obvious example. Furthermore, as industry commentators, often with vested interests, are fond of proclaiming, the concept of the mass market may be a thing of the past with the development of such technology as the Internet, the process of mass customization and the growth of direct marketing. Whilst their arguments have some face validity, the proposition that all firms will be able to customise their offer for each individual customer is patently unrealistic, as, for example, the prospect of negotiating your individual train journey with the engine driver ! who they are. This explains the enduring popularity of Haley’s (1968) “benefit” segmentation which is explicitly grounded in consumers’ behaviour and their fundamental purchasing and usage motivations. A number of useful criteria to help guide researchers in selecting appropriate market segments have been identified (Plummer, 1974; Kotler, 1980; Martin, 1986; Zeithaml and Bitner, 1996): • measurability: the size, location and content of a segment can be easily measured; • accessibility: the segment can be reached and effectively served; • substantiality: the segment is sufficiently large and profitable to merit investment; • actionability: effective marketing strategies can be implemented to attract and serve the segment(s); • determinant: buyers' decision factors can be clearly identified; • appropriate: the basis of segmentation is viewed as being rational by management; • predictive: the segmentation basis links market behaviour to segment membership. Segmentation in Service Industries In the services literature, the treatment of the central concept of market segmentation is noticeably lightweight. To date, researchers have tended to concentrate on the segmentation of external customer markets in the targeting and positioning of services (Gilmour, 1977; Easingwood and Mahajan, 1989; Payne, 1993; Gavin et al., 1996; Zeithaml and Bitner, 1996). Customer service cost reduction, customer retention and service level improvements have been cited as reasons to segment external service markets (Payne and Clarke, 1995; Zeithaml and Bitner, 1996). Service organisations have also tended to adopt an unsophisticated approach to segmentation. For the most part, service organisations have concentrated on demographic and geographic data to segment markets (Palmer 1994). One author commented that many organisations and researchers are currently paying only lipservice to the concept of market segmentation (Payne, 1993). In an earlier paper Elliott and Glynn (1998) have linked the process of market segmentation to the valuation of customer relationships over the expected duration of the relationships with particular reference to financial services markets. The proposed methodology produced a typology of customer segmentation based around “current profitability” (transaction-based) and “long-term attractiveness” (relationship value) of the customer connection. The underlying principles inherent in the process of valuing customer relationships were also discussed. However, as far as this methodology advanced the principles of segmenting services markets, it displayed a weakness common in most discussion of the principles and processes of market segmentationnamely that such segmentation is almost invariably based on the attractiveness of the customer to the seller. As such, it ignores, or assumes away, the interests and perceptions of the customer towards the seller. Of course, there is no guarantee that such relationships will be reciprocal and symmetrical and that the value placed on the customer by the seller will be mirrored in the customer’s perception of the seller. This then leads to the liklihood of a mis-match, especially as attractive customers are frequently valued more by the seller than vice versa. It is thus tempting to conclude that, as a cornerstone of the modern marketing concept, the practice of market segmentation, when based around the attractiveness of the customer to the seller, breaches the fundamental tenets of the marketing concept (ie the primacy of the customer’s interests). Haley’s (1968) “benefit segmentation” is a notably exception to this generalisation. Segmenting Industrial Markets In the case of industrial markets, there are particular issues which suggest that the practice and process of market segmentation will be less well grounded conceptually and less sophisticated in practice. Haas (1992, p.297), for example, concludes “although it is generally conceded that segmentation is as appropriate for business marketers as it is for consumer marketers, there is considerable evidence to conclude that the former have been slow to appreciate the concept.” One reason for this comparative lack of attention to segmenting industrial markets is undoubtedly, and rightly, the widespread presumption that industrial markets are not mass markets and that, consequently, there is little to be served in decomposing what are, already, inherently small markets. Clearly this will be the case in many highly specialised industrial markets in which close customer-seller relationships are the norm. Examples might include the relationship between IBM and Daimler-Chrysler or between Motorola and Apple Computer or Boeing and General Electric. Nevertheless, there are also clearly many industrial markets in which there are potentially a large number of industrial or business customers, where the product or service serves to satisfy very general needs or requirements, and where there is also a consequent need to identify meaningful and valuable segments of buyers. Examples would be the market for information technology services, industrial chemicals, office supplies, energy requirements, telecommunications services and so on. Recognising the need to segment large industrial or business-to-business markets, much of the writing has focussed on segmentation variables and how these should reflect the range of diverse industrial settings and purchasing practices. In this light, the Bonoma and Shapiro (1984) framework is widely cited. Their framework involves the following segmentation variables: § Demographic (industry, company size, location) § Operating (technology, user / non-user status, customer capabilities) § Purchasing approaches (purchasing-function organisation, power structure, nature of existing relationships, general purchase policies, purchasing criteria) § Situational factors (urgency, specific application, size of order) § Personal characteristics (buyer-seller similarity, attitudes towards risk, loyalty). Wind and Cardozo (1974), in a seminal paper, argue for a two-stage approach to the process, involving (1) identifying meaningful “macrosegments” and (2) sub-dividing those “macrosegments” into meaningful “microsegments”. Variables forming the macrosegments include (Weiner, 1983): § Industry § Organisational characteristics (size, plant characteristics, location, economic factors, customers’ industry, competitive forces, purchasing factors) § End-use markets § Product application. § Microsegmentation variables include: § Organisational variables (purchasing stage, customer experience stage, customer interaction needs, product innovativeness, organisational capabilities) § Purchase situation variables (inventory requirements, purchase importance, purchasing policies, purchasing criteria, structure of the buying centre) § Individual variables (personal characteristics, power structure). Whilst, it is not intended to be critical of this framework, a common problem of this and other segmentation frameworks is that they provide little insight into which of these variables may be most useful and in what combination or sequence. When a sequence is suggested (McKoll-Kennedy and Kiel, 2000) it is likely to be based on “common sense” with little theoretical or empirical justification. A result of this lack of a commonly accepted template is that most industrial market segmentation processes therefore suffer from being piecemeal and idiosyncratic. This situation also largely applies in the segmentation of consumer markets, although Haley’s (1968) benefit segmentation has received widespread acceptance. The concept of benefit segmentation has been extended to industrial markets by Doyle and Saunders (1985), although their emphasis on product benefits, rather than user benefits is questionable. Customer Segment Portfolios Central to the segmentation methodology proposed in this paper is the valuation of customers and their relationships with the seller organisation. Moreover, it is also proposed that customers can, and should, be regarded as assets and, like a firm’s financial assets and investments, managed as a “portfolio”. In evaluating businesses (and customer segments) in terms of their value both in the present and over time, an obvious advantage lies in the capability to make comparative judgments of the relative attractiveness of customer segments. Integral to this process of comparative analysis is the concept of portfolio analysis. Following this argument, a central proposition of this paper is that it is valid to analyse customer segments in terms of their value, to the seller organisation, in the present and over time Further, it is argued that organisations may analyse, and manage, their customer base from the perspective of a balanced portfolio, in the same fashion as is widely accepted that organisations should maintain balanced product or business portfolios. The concept of the desirability of looking at the customer base as a portfolio is not new. For example, Shapiro and Bonoma (1984) classify customers in two dimensions based on their sensitivity to price and the cost to the organisation to service the customer. Customers fall into four resultant cells of their matrix, labelled as “passive”, “carriage trade”, “bargain basement” and “aggressive” and there are clear parallels with the segments proposed in the current paper. More recently, other authors (Krapfel et als, 1991; Yorke and Droussiotis, 1994) have further illustrated the importance of viewing the customer base from the perspective of a portfolio. With these considerations in mind, the author proposes a portfolio-based matrix approach to the evaluation and selection of target market segments. The portfolio will, ideally, comprise a mix of customer segments which yield to the seller an optimal blend of current profit and future growth and economic value. To the buyer, the approach focuses on appropriate relationships with the seller which meet the buyer’s shortand longer-term expectations. The key to the model lies in valuing relationships from both the seller’s and the buyer’s point of view over time. The incorporation of these important considerations, to the authors’ knowledge, presents a new approach. Segmentation and Relationships This paper thus presents an alternative framework for the segmentation of industrial markets one based on the nature of the buyer-seller relationships and which seeks to tap into the interests of both parties. In common with the widely discussed portfolio models (BCG, Directional Policy Matrix, GE Business Screen) the proposed model is two dimensional. These dimensions are proposed as: • Buyer loyalty; and • Value to the seller. These dimensions are closely indicative of the "relationship" value to both the buyer and the customer.. The simplified two-dimensional framework is shown below: Figure 1: Proposed Segmentation Framework Buyer Loyalty Low High Low (1) Simple (2) Buyer Value to Exchange Exploitation Seller High (3) Seller’s (4) Partnership Over-investment The interpretation of the matrix is simple and largely self-explanatory, although, the underlying rationale, the authors argue, is both sound and capable of implementation. The fundamental premise of the methodology is that market segmentation is most effective when it reflects genuinely the interests and practices of both buyer and seller. The same cannot be claimed for conventional industrial segmentation practices which commonly revolve around the use of available secondary data (eg demographic data) and/or the existing production or sales organisation (eg industrial application). Thus, the logic for the proposed approach is that, while it is acknowledged that value to the buyer is arguably complex and multi-dimensional (Barnes and Cumby, 1999) the buyers’ interests will be fundamentally reflected in their loyalty to the seller (either historically or in the future). This reflects the fact that while, undoubtedly, many industrial buyer-seller relationships are mutual and long-lasting, in a significant proportion of cases, particularly in larger mass industrial markets, some buyers do not seek a deep and reciprocal relationship. Such buyers will be commonly seeking to maximise value around a single transaction or short-term relationship. They have no loyalty; nor do they expect it. (A familiar example is the commercial/business market for electricity. Partly as a result of history, in which the government was a monopoly supplier, most buyers neither have experience with, nor seek long-term relationships with their supplier, and many will buy simply on the basis of a fractionally lower price, whilst others are dominated by “inertia”, (as distinct from “loyalty”). Much of government purchasing which requires competitive tendering will reinforce this lack of loyalty. Conversely, the seller’s interests will be reflected in the “lifetime financial value” of the customer. This approach addresses concern expressed by Reeder et al (1987, p.224) that “before target markets are actually chosen by the firm, identified segments must be evaluated with respect to their profitability – the cost of serving and maintaining them as it relates to the potential sales revenue.” This involves an empirical investigation of such factors as average purchase size, purchase frequency, liklihood of switching, costs of servicing, purchase of consumables and after-sale service, recommendations etc. Clearly, in the ideal situation there will be a symmetry in the interests of both parties as shown in cells (1) and (4); with sellers’ preferring the “partnership” relationship shown in cell (4). Where the buyer clearly is not looking for a close or long-term relationship, as shown in cell (1), the seller’s best interests are served by focussing on the current transaction, lowering its cost and maximising the value of the transaction without risking losing the deal. When there is a mismatch in both parties’ expectations, then the likely consequence is that, while the deal will proceed, it will result in adverse consequences for both parties. In the case of cell (2), the seller will receive a windfall gain on the transaction, but will create illwill as the buyer will feel exploited and will not re-purchase, even though that would be his preference. In the case of cell (3), the seller will over-invest in what he perceives as a valuable customer, only to see the customer desert the seller after the transaction, notwithstanding the investment of the seller in the relationship. Thus, the logic of the proposed typology, appears sound and overcomes the shortcoming which the author perceives in most conventional approaches to industrial market segmentation (ie that they are based on what is easy and suits the seller’s modus operandi). Moreover, the proposed typology embraces the central concepts of customer relationships, customer value and customer loyalty and incorporates them into the pivotally important process of market segmentation in industrial markets. Estimating Customer Value The principal empirical challenge in operationalising the models is that of establishing the value to the seller of each customer and customer segment. This is a formidable operational challenge, although the principles guiding such a task can be discussed broadly. In Barnes and Cumby’s view, the estimation of true customer value requires five stages of analysis in order fully to account for all costs and benefits relevant to individual customers and, by extension, customer segments. These comprise allocatable and non-allocatable; monetary and non-monetary costs and benefits (Barnes and Cumby, 1995). Benefits would include conventional income through sales, fees and charges, interest etc., but would also include measures such as customer satisfaction, referral sales, repeat business, loyalty and goodwill. Similarly, costs include the usual direct labour charges but further include training, compensation, motivation, service quality expenses etc. “The objective is to link revenues with the appropriate customer or market segment, and to match costs with their related revenues, in some systematic and rational manner” (Ibid). In common with other portfolio models, the constituent "independent" (predictor) variables and their relative weightings could be expected to vary between sellers due in part to their varying histories, markets, structures and financial objectives. Alternative approaches to valuing buyer-seller relationships have been provided by Rexeisen and Sauter’s “Customer Profitability Analysis” (1997) and Gattorna and Walters’ “Customer Account Profitability” (1996), although it is argued that these focus primarily on the more tangible aspects of relationships as measured in the net financial flows. An alternative view of the task of valuing relationships is provided by Baxter and Matear (1999) who advocate treating buyer-seller relationships as “Intellectual Capital”, although they concede that there is no widespread agreement on how such intellectual capital as relationships should be accurately estimated. In the task of calculating the value of chosen customer segments, it is beyond the scope of this paper to itemise the components and their relative importance in such a system. However, in general, the following factors would commonly be relevant (for each customer segment): § Annual sale revenue § Cost of sales § Marketing and selling cost § Distribution costs § Inventory § Discounts and other terms of sales § Possible referrals and recommendations § Expected “life span” of relationship. Calculation of this value to the seller will rely, typically, on the analysis of a representative sample of longitudinal customer histories extending perhaps over ten years or more. Moreover, such a value will be probabilistic as there is a finite and calculable probability that the customer will remain a customer over the span of the chosen period. (The identification of this period is also important, especially in light of the widely accepted view that buyerseller relationships with some key customer groups have become appreciably shorter in recent years as these key segments display an increasing willingness to “shop around”.) Measurement of Loyalty Similarly, from the buyer’s perspective, the measurement of loyalty can arguably serve as a proxy measure of the buyer’s valuation of the relationship. The profitability of loyal customers is well recognised and is the key to the concept of “relationship marketing”. Whilst the value of loyal customers is widely appreciated, the prior identification of such customers is more difficult, since loyalty can only be observed once a relationship has had sufficient time to develop. This is, of course, impossible with new or prospective customers. The isolation of demographic, psychographic and behavioural indicators of loyalty remains a major challenge for analysts and researchers. Notwithstanding, an important advantage in using loyalty as a proxy measure of the buyer’s value of the relationship is, of course, that loyalty data are readily available and based on actual historical behaviour, although Barnes and Cumby’s caveat is apposite: “To maximise profitability, relationship building should be forward looking, not entrenched in historic monetary revenue patterns” (Barnes and Cumby, 1995). Nevertheless, the use of such “hard data” does ensure that the model is grounded in measurable and verifiable facts. Implementation Issues Having demonstrated, prima facie, the value of such an approach to segmentation, the focus shifts, quite naturally, to its implementation. The information requirements of such a system are formidable, although not insurmountable. Equally, the payoffs of such a system should reward the seller handsomely. The financial payoff from focussing efforts and resources on the most valuable, and loyal, customers would be substantial. This issue, while clearly significant in its operational demands, should not be beyond industrial marketers. The key concerns in its implementation are the data collection and analysis demands. In particular, the model requires systematic classification of customers in terms of their historical or likely loyalty and historical or likely future value to the seller. Both these issues are currently the subject of interest to the accounting and marketing professions and progress seems assured. The currently popular practice of data mining has, at its heart, the task of uncovering the historical and future valuation of customer relationships. Moreover, since these data are largely historical and internal by nature, the data collection task should not be insurmountable. The more formidable challenge lies in the development of the algorithms which reflect both the marketing and accounting realities. A further issue is, of course, the simplistic nature of the 2 X 2 matrix. This is acknowledged, and while it is useful for illustrative purposes, the practical implementation of this framework ought, at the very least, involve a 3 X 3 matrix, if not the use of more finely delineated axes.