Journal of Marketing Management, Volume 25, Issue 3 & 4 April 2009, Pages 227-252 Segmentation and Customer Insight

in Contemporary Services Marketing Practice: Why Grouping Customers Is No Longer Enough Christine Bailey Paul R Baines Hugh Wilson Moira Clark* Cranfield School of Management, United Kingdom * Henley Management College, United Kingdom

Address for Correspondence: Dr. Paul Baines Cranfield School of Management Cranfield Bedfordshire MK43 0AL Tel.: 01234 758041 Email: paul.baines@cranfield.ac.uk

Biographies Christine Bailey is Marketing Director, Services, European Markets at Cisco Systems. She holds a bachelors degree in German and Business Studies from Warwick University and a DBA in Customer Relationship Management (CRM) from Cranfield School of Management. Christine has 17 years experience in technology and services marketing and speaks frequently at academic and industry conferences. Previously she spent five years as Marketing Director of Extraprise (now Detica), a specialist CRM consultancy firm. Other employers include Cambridge Technology Partners (now Novell), Insight Marketing and Communications and Hewlett-Packard in Germany. Paul Baines is Senior Lecturer in Marketing and Director, MSc in Strategic Marketing at Cranfield School of Management. He holds bachelors, masters and doctoral degrees from The University of Manchester. He is a Chartered Marketer and Fellow of the Chartered Institute of Marketing, and a full member of the Market Research Society and Institute of Directors. Pauls specialist research interest is market segmentation and positioning,

particularly in political markets. He has co-written and co-edited nine books, the most recently published of which is `Marketing` with Chris Fill and Kelly Page (Oxford University Press, 2008). Paul has conducted strategy and market research consultancy for a variety of not-for-profit and private sector clients over the last 11 years. Hugh Wilson is Professor of Strategic Marketing and Director of the Customer Management Forum at Cranfield School of Management. Hugh has a mathematics degree from Oxford University, a postgraduate computer science degree from Cambridge University and a PhD in marketing planning from Cranfield. He has extensive industrial experience, including thirteen years in the IT industry working for IBM, Logica, Artificial Intelligence Ltd and NCR in marketing, consulting and business development. He has published in Industrial Marketing Management, British Journal of Management, European Journal of Information Systems and Journal of Strategic Information Systems amongst other journals. His books include `The New Marketing` (2002, with Malcolm McDonald) and The Multichannel Challenge (2008, with Rod Street and Lindsay Bruce). Moira Clark is Professor of Strategic Marketing at Henley Management College and Director of The Henley Centre for Customer Management. Her major areas of research are in customer relationship management, customer retention and organizational climate. She is co-author of Relationship Marketing for Competitive Advantage and Relationship Marketing: Strategy and Implementation. Moira is also a judge for the Management Today/Unisys Customer Service Excellence Awards, which has led to her latest book 'Business Success through Service Excellence'. Prior to joining academia, she was international marketing consultant, worked as a marketing director for an international health food manufacturer and as a marketing manager for the toiletries subsidiary of Dunhill International. Keywords: market segmentation, services marketing, customer insight, customer analytics, propensity modelling

Segmentation and Customer Insight in Contemporary Services Marketing Practice: Why Grouping Customers Is No Longer Enough

Abstract The bulk of market segmentation literature has concerned the generation of segments, with far less attention on what segmentation is used for particularly surprising given the common speculations that the role of segmentation is changing due to CRM practices and the wider range of forms of customer insight which they enable. We explore market segmentation in the services and product-service systems context through twenty-five interviews in five UK-based companies, highlighting practical considerations in implementing market segmentation programs (see Young, Ott and Feigin, 1978, for a similar early approach). Within this case set, market segmentation, using a variety of segmentation bases, is still regarded as essential for customer selection, proposition development and mass communication. Addressable and interactive communications with individual customers, though, are increasingly based on individualised customer analytics and propensity modeling, which aid the determination of the likelihood of uptake of specific propositions. Events and triggers informing companies of how to deal with customers individually are also considered to be particularly effective rather than simple allocation of the customer to a particular characteristic segment. Implications for theory and practice in market segmentation are outlined and further research is called for to explore this important area further. Introduction Although the process of generating market segments has been much studied, what segmentation is actually used for has received relatively limited attention until recently (Yankelovich and Meer, 2006). This seems surprising given the common speculations that the role of segmentation is changing due to CRM practices and the wider range of analyses which they enable (Dibb, 2001). The question arises as to what extent one-toone techniques such as propensity modeling (which predicts the likelihood of an individual customer acting in a certain way, such as responding positively to an offer) might cast doubt on the ethos and need for segmentation, as it could be argued that they perform the same purpose of treating different customers differently but with more granularity (Dibb, 2001; Kumar et al., 2006). However, others argue that the reason for CRMs failure is that it should be based on good old-fashioned segmentation analysis (Rigby et al. 2002). While there are differences in perceptions of the role of segmentation in a marketing environment which includes CRM, there is at least some consensus that customer insight is a key resource required to achieve effective CRM, although segmentations contribution to this insight is still unclear. Customer insight as a term has come into vogue in both academic and practitioner discourse (Hirschowitz, 2001; Wills and Williams, 2004; Smith, Wilson, and Clark, 2006b; Wills and Webb, 2007) to reflect the multiple informational sources in existence about customers, ranging from market research to transactional data-mining, as

well as data from marketplace experimentation and third-party sources (Slater and Narver 2000). The few empirical studies reviewing customer insight in practice (Wills and Williams, 2004; Wills and Webb, 2007; Smith, Wilson, and Clark, 2006a) concur that customer insight arises from multiple data sources, yet there is a paucity of evidence on the forms of insight that companies are generating and from where (Forsyth, Galante, and Guild, 2006; Langford and Schulz, 2006). Hirschowitz (2001) regards segmentation as a subset of customer insight, citing such segmentation bases as socio-demographic, geo-demographic and value-based as customer insight examples. But is customer insight just a new name for market segmentation or does it encompass separate sets of activities and processes? In this study, we take the view that the debate on such questions is under-informed by an adequate understanding of current segmentation practice, and in particular of the role of segmentation within the wider process of actioning customer insight (Sinkula et al., 1997). We consider the question of whether or not market segmentation has been superseded by other forms of customer insight, and the problem of how companies use or action market segmentation programmes. In particular, we seek to explore the following research questions: RQ1: Have market segmentation processes been superseded by distinct customer insight processes? RQ2: How do contemporary companies define their segments? RQ3: How is segmentation being implemented or (as we prefer to refer to it) actioned? We focus specifically on services companies, and on companies offering integrated combinations of products and services, as this context provides direct, rich relationships between the firm and its customers in which the potential, at least, exists for both individualised customer insight and individualised treatment of the customer based on that insight; thereby considering the challenge to segmentation thinking based on the notion of groups of customers posed by CRM scholars to the test. Literature Review Wendell Smith (1956) first proposed market segmentation as an alternative market development technique to product differentiation in imperfectly competitive markets. Since few markets correspond with an idealized perfect market, and as market-oriented companies tend to be more profitable because they define products from the perspective of the customer rather than their own needs (Wong and Saunders, 1993; Day, 1994), the rationale for market segmentation seems self-evident. However, to date the literature on market segmentation has focused quite narrowly around what segmentation bases to use, particularly advocating customer characteristics (Foote, 1969), product attributes (Botschen, Thelen, and Pieters, 1997), benefits sought (Haley, 1968), service qualities (Gronroos, 1998), values (Claeys, Swinnen, and Van den Abeele, 1995), and buying behaviour (McDonald and Dunbar, 2005). Such bases are particularly skewed towards the consumer marketing field, with a more limited treatment in the business to business literature (Verhallen, Frambach and Prabhu, 1998; Smith, 2002), where implementation problems are equally paramount and less wellconsidered (Dibb and Simkin, 1994; Palmer and Millier, 2004; Laiderman, 2005). From the services marketing field, segmentation has been suggested on the basis of customers

propensity to switch suppliers: Payne and Frow (1999), for example, argue for four segments of customers who definitely will not switch, probably will not switch, might switch, and definitely will switch. The main reasons identified when switching behaviour does actually occur include pricing, inconvenience, core service and service encounter failures (so-called critical incidents), competition, ethical problems, and involuntary switching (Keaveney, 1995). Nevertheless, service-based customers can and do frequently stay with a service provider despite their dissatisfaction (Kelley, Hoffman and Davis, 1993). Recently, the ethos and need for segmentation has been cast into doubt, as it juxtaposes with relationship marketing, where individualization rather than grouping customers is key (Coviello et al., 2002). In theory, it has been argued, firms with a direct relationship with their customers as is typically the case in service industries in particular - should be able to use CRM technology to identify the right future customers, understand their needs, predict their behaviour, develop tailored propositions and have more relevant conversations with them, all at finer levels of granularity than has previously been possible (Payne and Frow, 2005). However, numerous research studies (Rigby, Reichheld, and Schefter, 2002; Wilson, Daniel, and McDonald, 2002; Kale and Sudhir, 2004; Boulding et al, 2005) have found that CRM projects have often failed to deliver the expected benefits. Counter to the previous argument that CRM replaces segmentation with the segment-of-one, Rigby et al (2002) liken implementing CRM without segmentation to trying to build a house without engineering measures or an architectural plan. Common to these different views, though, is the assumption that in services contexts, CRM technology can enable more appropriate customer conversations in which different customers are treated differently. When ten leading services scholars were questioned on the direction that services marketing research should take in the future (Grove, Fisk, & John, 2003, p. 116), nearly every panel member challenged scholars to chart new territory and examine the impact of technology on services. This addressed Bitner et als (2000) concern that the growing role of technology in service encounters has been largely ignored and that virtually all of the service research has instead explored the interpersonal dynamics of the encounter. A number of authors (Chase & Hayes, 1991; Kelley, 1993; Evans, Arnold, & Grant, 1999; Spencer-Matthews & Lawley, 2006) have suggested that technology-enabled service encounters have the potential to provide a direct financial benefit (Evans et al., 1999; Spencer-Matthews & Lawley, 2006). They propose that if agents manage to initiate conversations that uncover customer needs, this could lead to cross-selling (selling new products), up-selling (selling upgrades of existing products), and specific offers that enhance customer retention. In support of this contention, Beatty et al. (1996) report that investing the time to investigate customer needs may indeed improve both service and sales performance. Cross-selling is attractive to firms because it usually costs less than acquiring new customers (Reichheld & Sasser Jr, 1990). In addition, the more products and services a customer holds, the more likely they are to develop a more durable relationship with the firm, the less likely they are to consider switching to another provider, and the stronger their profitable lifetime duration (Beatty et al., 1996; Ansell, Harrison, & Archibald, 2007). This link between sales and service was proposed nearly two decades ago, with Zeithaml et al. (1988) demonstrating that offers made during service encounters - if underpinned by the delivery of good customer service - can help companies to attract new customers and

develop existing ones. Combining sales and service roles extends services operations into the marketing domain (Evans et al., 1999), and indeed practitioners are now frequently referring to this concept of sales through service as inbound marketing. The popular press suggests that the growing trend towards such inbound marketing is caused by increased restrictions imposed by data privacy and communications legislation, combined with claims of higher conversion rates on offers made on inbound calls as compared with outbound contact by telephone, mail or email. Gartner Group (2006) suggests that companies can expect 1020 times the response rate on analytical inbound marketing compared to traditional marketing and Doyle (2005) hypothesised that response rates are commonly in the 20-30 percent range. However, according to Eichfeld et al (2006, p.1), companies have failed to tap the full revenue potential of their call centres because they just dont understand the extent of the opportunity. One issue holding back the adoption of sales through service, according to Maister (1997), is that the accurate picture of customer needs that it requires can only be achieved through intensive customer-employee interaction. With advances in the generation of customer insight, though, technology is now enabling contact employees to handle service situations with a complexity that could never be managed manually (Bitner et al., 2000). Advances in technology have also fuelled a growth in the popularity of customisation strategies aimed at providing customers with individually tailored products and services (Gwinner, Bitner, Brown, & Kumar, 2005), and real-time marketing strategies (Oliver, Rust, & Varki, 1998) are now being deployed in customer service centres. What place this real-time, individualised world of service leaves for segmentation, though, is left unexplored. In summary, segmentation research has focused around choosing segmentation bases, as opposed to how a segmentation programme is used once generated - with some notable exceptions (Dibb and Simkin, 1997; Dibb and Wensley, 2002; Dibb, 2005; Laiderman, 2005) which include how segments may be used in the marketing planning process to (in)form propositions (McDonald and Dunbar, 2005). There are few prescriptions on how segments can be used in individual customer interactions, and evidence on what actually works in practice is very limited indeed (Wind, 1978; Wedel and Kamakura, 2002; Palmer and Millier, 2003; Laiderman, 2005). The limited understanding of the role of market segmentation in practice is keenly felt by practitioners, who bewail the lack of guidance on actionable segmentation models (Marketing Leadership Council, 2007). This therefore constitutes the raison detre of this paper: to inform the development of more actionable segmentation models and to inform the reader on the strategic and operational role of segmentation as opposed to how to segment. Research Methodology We used a case-based qualitative research methodology, collecting data through 25 in-depth interviews in five UK-based large companies from multiple industries. This approach was deemed the most appropriate because we are seeking to uncover leading edge practice in an area where the theory base is comparatively weak and the environment under study is complex (McCutcheon and Meredith, 1993; Harrison, 2002). Purposive sampling (Gill and Johnson, 1991) was used to select companies using customer insight at all three of Smith, Wilson, and Clarks (2006) levels: 1) generating customer insight, 2) actioning insight in

service and packaging design, and 3) using insight to tailor the core product/service itself. The companies selected were predominantly service providers; Cisco and BT are partial exceptions, being suppliers of integrated product-service systems (Davies et al. 2006) for whom some customers are intermediated by third party organisations. To identify possible sample cases, we considered: companies speaking on customer insight or market segmentation at major industry conferences; member organisations of CRM bestpractice research centers run by two UK business schools; and clients of a major CRM consultancy. This process resulted in a shortlist of 15 companies. Negotiations with these companies led to final selection of five organizations (see Table 1). This final selection took account of where access could be agreed, as well as seeking a non-competing set of companies to ease concerns about confidentiality. The final case study selection comprised Barclays, BT Global Services (Major Customers Division), Cisco Systems, O2 and the Post Office, of which the BT and Cisco cases were B2B and the remainder spanned B2B and B2C customers. In-depth interviews with key personnel were adopted as the best format to use when physical and psychological proximity to the detail of the subject matter is required (Carson and Coviello, 1996). A balanced group of four to six interviewees per company was obtained, involving senior personnel from the disciplines of marketing, customer insight, CRM, call centre and sales management functions. The data were collected between March and August 2006. The interviews were recorded and transcribed and the information was collated into cases, along with any supporting documentation. Each case was then studied in detail, to ensure a rich familiarity. A number of logical sections emerged, each with sub-themes. Sections of text were highlighted and a note of the section/sub-theme was made in the margin. As every new sub-theme emerged, it was listed in an Excel spreadsheet. This spreadsheet then formed the basis of a coding framework, which was entered into the software programme NVivo. Having imported the documents into NVivo, the highlighted passages were coded in batches, per company. This process led to some nodes being created, re-organised or deleted. In order to write up the findings, a coding report was run for each node, displaying comments across all cases on each section/sub-theme. This facilitated a cross-case search for patterns. The NVivo coding framework resulted in the structure of Tables 2 and 3.

Table 1: Case Details

CATEGORY Industry Sector Market Sector Company size (revenue) Interviewee Job Roles BARCLAYS Financial services Business to Consumer and Business to Business $44.7bn (year to Dec 2007) 1.Commercial Manager, North East 2. Director of Marketing Services 3. Head of CRM 4. Senior Manager, CRM delivery 5. Commercial Director, Direct Channels BT Telecommunications Business to Business $39.3bn (year to March 2007) 6. Programmes & Campaign Analysis (Reporting) 7. Head of Marketing Information Management & Systems 8. Senior Market Sizing & Segmentation Manager 9. Insight Manager, UK Major Customers CISCO Technology Business to Business $34.9bn (year to July 2007) 10. Market & Customer Understanding Manager 11. Head of CRM 12. Call Centre Manager 13. Head of UK Marketing O2 Mobile telecommunications Mostly Business to Consumer $5.5bn (Telefonica O2 UK, year to Dec 2007) 14. Head of Insight 15. Head of UK Channel Marketing 16. Head of Real-Time Marketing 17. Head of Business Customer Service 18. Head of CRM, Strategy & Architecture PO Retail & financial services Business to Consumer and Business to Business $1.7bn (year to Dec 2007) 19. Head of Insight Products, Consumer 20. Director of Customer Insight 21. Head of Insight Products 22. Business Segmentation Manager 23. Senior Insight Manager 24. Network Specification Manager 25. Senior Insight Manager UK only

Firms Geographic Reach

Multi-national

Multi-national

Multi-national

Pan-European

Findings We begin by summarising four broad categories of insight generated across the case study organizations: 1) market predictions, 2) customer segments, 3) need and opportunity-focused analytics and 4) customer value analytics. Table 2 provides an overview of how each of our case study companies uses different data types in each of these four broad insight categories. All five of our case study companies were using data to make market predictions and were using events and triggers to identify revenue opportunities. Each of these four broad areas is explained in further detail below. Table 2: Customer Insight Generated
CATEGORY Market Predictions Customer Segments DATA TYPE Various By Business Type By Attitudes By Value By Vertical Market By Needs By Customer Lifecycle By Buyer Behaviour By Geography By Demographics Propensity Models Events & Triggers Customer Profitability Product Profitability Customer Lifetime Value Share Of Customer Wallet Customer Lifecycle Analysis BARCLAYS BT CISCO O2 PO

Opportunity Analytics Customer Value Analytics

Note: One tick indicates that this type of insight is being generated. Two ticks comparison to the other cases, this type of insight is being extensively generated.

indicates that in

Market Predictions: Traditionally the domain of market research, this area has been expanded to include transactional customer data to help make predictions not just about the total market size but also about the companys sales potential and the market trends and issues likely to affect a companys ability to achieve this potential. Respondents considered market predictions to be of greater importance as competition intensifies, as customer demand shifts, as cost pressures to invest wisely increase, and as a result of the greater need to diversify into new markets. Customer Segments: All companies were dividing up their existing and potential customers into groups for customer selection purposes, although they took a variety of approaches. Although there were several instances of traditional a priori segmentation by size of company, geography and demographics, all of the companies were also attempting to segment by needs and behaviour, although they all acknowledged the struggle necessary to implement such approaches.

Opportunity Analytics: Two of the companies, Barclays and O2, were able to segment right down to the individual level using propensity modelling techniques in their consumer markets, while O2 were also beginning to apply similar techniques to business customers. These techniques were being used to determine the likelihood of individual customers buying a cross-sold product or defecting, and were being actioned at the front line through proactive sales and retention offers. All the companies claimed that identifying customers needs based on events and triggers - unusual occurrences in customers lives or their transactions which impact upon their purchase behaviour - were the most actionable forms of insight, providing actionable opportunities for cross-sell, upsell and retention offers which resulted in high reported conversion rates. Customer Value Analytics: The fourth category of customer insight was customer value analytics. Whereas the previous category focused on the most attractive proposition to an individual customer, this category focuses on the most attractive customers to the firm. Companies undertook analyses to build up profiles of customer value more generally not only for the purpose of segmentation but also for other purposes including marketing communications, product development and financial planning. Customer profitability and customer lifetime value are notoriously difficult to calculate in practice, due in large part to the difficulty in appropriately allocating costs (van Raaij, 2005; Wang and Hong, 2006). Only Barclays and Cisco claimed to calculate customer profitability and three companies calculated product profitability (Barclays, O2 and Post Office). Only Cisco mentioned customer lifetime value, and only BT and Cisco are trying to measure share of customer wallet, a measure of customers relative spend amongst competing offers. We next consider the role of segmentation in each of our company cases in greater depth from the perspective of the process of generating segments and then applying how those segments are used in marketing practice. Barclays Generating Segmentation In the past, Barclays, a large UK-based bank, generated attitudinal profiles, but struggled to attribute those profiles back to its customer base to allow customers to be served differently. Barclays now adopts a hybrid segmentation model, incorporating attitudinal profiles and four types of segmentation (see Figure 1): 1. Business type (personal, premier, private, small business); 2. Operational segmentation by age and wealth; 3. Attitudinal segmentation ; 4. Executional segmentation, operationalised as triggers, events and propensity models.

Figure 1: Hybrid Segmentation Model at Barclays

Personal Customers
Business Operational Attitudinal Executional

Premier

Private Clients

Small Business
Medium Business & Agriculture

Customer Categories (Age by Income) 5 Groups 19 Types

On demand Managed

Larger Business Value Segments e.g. deciles

Group Attitudinal 8 Types Triggers, events & propensities

Direct Nature of Relationship - Managed

Respondents regard the executional segmentation category as increasingly important because its focuses on how individual customers are to be treated. The category covers a number of customer-level analyses which have predictive power as to how a customer might respond in the future to a specific offer. One such analysis is triggers, occurrences in dealings with the customer which are believed to be commercially significant. These include out-of-the-ordinary events that occur on the customers account and about which Barclays would want to talk to customers. For example, if a late payment fee has been charged or if a significant credit has come into the customers account, this might indicate that the customers needs have changed in some way. Similarly, a customer who has just taken out cash on their credit card presumably has an urgent need for credit and can be targeted for a loan. The trigger information usually needs to be combined with additional data such as how credit-worthy the customer is. This helps to determine whether a loan would be an appropriate offer for a customer who has just been charged a late payment fee or an overdraft extension. Respondents claimed that such an analysis should enhance the customer experience as the bank can actually help customers to avoid paying fees in the future. Another trigger is evidence of customers holdings of competitor products/services, obtained through direct debit stream analysis. For example, a customer with a direct debit for Halifax home insurance might then become a target for Barclays home insurance products. Barclays considers the importance of events on customer accounts as well: for example, when a customers insurance is coming up for renewal, or a mortgage or loan is coming to the end of its term. Examples of non-product-oriented events are moving house, getting married and having children. This approach might be loosely correlated with the more familiar lifecycle (now known as lifestage) approach to market segmentation (Wells and Gubar, 1966) but with one key difference: what is important is the timing of the transition and its implications for

time-specific communications, not just in which lifestage category a particular customer resides. Barclays main focus in executional segmentation, though, is on propensity modelling to generate Customer Service Opportunities or CSOs, that indicate the the next best product or offer for services representative to talk to customers about, taking into account their contact and transactional history. Barclays first combines its own transactional data from current accounts and credit cards with external data sources, to give it a rich picture of customers lifestyles, finances, careers, spending habits and travel. It then uses technology to combine twenty propensity models, which predict the customers likelihood of responding to particular offers, using an optimisation engine to determine what offer to present to the customer. The engine takes into account multiple variables including propensity or likelihood to respond; value of the offer; cost of contact by channel; operational capacity; and target volumes by product. It can calculate the most profitable product to offer to a customer (if there is more than one choice), taking into account a variety of contact and business rules, including quarantine rules, campaign selection rules, total budgets, campaign volumes, campaign timings, crosscampaign rules and channel priority rules. The engine also takes into account a customers values and behaviour so that each offer can be personalized. The optimization is refreshed daily. Applying Segmentation Generally speaking, operational segmentation by age and wealth is used for strategy, marketing planning and evaluation. The attitudinal segmentation is used for proposition development, media planning and communication treatments, while the executional segmentation is used to optimize the management of the existing customer base. The Customer Service Opportunities are used for direct mail and bank telephony, both outbound (i.e. telemarketing) and inbound (i.e. service call handling). In the case of direct mail, up-sell and cross-sell offers are made alongside bank statements. An up-sell offer might prompt a current account holder to upgrade their basic account with additional services, such as the inclusion of a dedicated relationship manager or a guaranteed overdraft. A cross-sell offer might prompt a mortgage account holder to take out home insurance. In outbound telephony, the CSOs drive thousands of calls each day, for example encouraging certain customers to increase their loan the next month, or inviting customers who had been regularly overdrawn on their current account to come into a branch for a financial review. More recently, the banks focus has shifted from outbound to inbound communications - targeting customers with a tailored message when they contact the bank. Barclays handles around 35 40 million inbound calls each year, typically with service requests such as to check the account balance or to pay bills. Interviewees reported that since CSOs had been made available to inbound call centre agents, the sales ratio had increased from one sale per 14 inbound calls to one sale per 11 inbound calls. Like many organisations, Barclays segmentation efforts are hampered by its organisational structure, which is aligned to business unit segments rather than customer-centric segments, and they face significant operational challenges in responding quickly enough to the opportunities provided (although this is a common problem reported recently elsewhere - see Gulati, 2007).

10

British Telecommunications plc Generating Segmentation BTs Major Customers Division, part of the largest telecommunications provider in the UK by market share, uses three means of defining customer segments as follows: i) by value using tiers labelled A, B and C which take into account current and potential value; ii) by industry sector; and iii) by business needs for telecommunications, with segments such as expanding, serving consumers/citizens, running the business and low needs/DIY. The business needs segmentation is calculated using a mixture of customer interviews and survey questions from BTs ongoing client landscape survey. The CRM system contains smart scripts to prompt account managers to ask customers certain questions, which inform the survey procedures and results. The segmentation distinguishes four types of buyer: i) low involvement; ii) demanding; iii) relationship seekers, who want a dedicated account manager; and iv) price-sensitive, relationship-shy, transactional buyers. Account managers run their accounts through the segmentation model on an annual basis to determine whether the initial allocations still hold true. Interviews with BT managers indicated that although BT did not use the terminology of events and triggers in connection with their segmentation strategy, they do utilize such an approach implicitly, by prompting account managers to act when there is an opportunity for relevant dialogue with the customer whenever such information about a customers circumstances arises. Applying Segmentation BT has a segmentation governance board which consists of heads of marketing from different business units, the segmentation manager, a strategy representative and a sales representative. This board is responsible for agreeing the overall segmentation model, communicating it within the company more generally and gaining buy-in from the sales community. A customer base team runs new accounts through the segmentation model before allocating them to a particular sales group which makes recommendations about how the account should be managed. This is then communicated to BTs service, billing and customer satisfaction teams. Although the new segmentation model is still in the early stages of development, it demonstrates how BT is beginning to use segmentation not only to determine what to sell, but also how to sell. At the moment, the model is primarily used to develop new propositions, and to help in the identification of potential early innovators. The implementation challenge focuses around how the sales team learns to identify customer needs and adapt their sales behaviours accordingly, a shift from selling based only on product features and benefits. The desk-based account teams currently use the Siebel CRM system to view a matrix of what they have sold, what is outstanding, and what campaigns customers have been included in, as well as what opportunities are open to them at the moment. BTs CRM system indicates events and triggers, such as prompts to indicate a contract is up for renewal or has lain dormant.

11

Cisco Generating Segmentation Cisco, a US-based IT networking company, has traditionally segmented by geography and size of company, but more recently has started segmenting by technology lifecycle and by purchasing method. With this latest approach, Cisco believes that it is at the beginning of understanding what managers call the customers next best move. Technology lifecycle equates to a customer life-cycle characteristic of the companys status in adopting new technologies (Rogers, 1962). Respondents claim that it is now possible to predict whether a particular product is needed for certain types of businesses, depending on where they are in the technology lifecycle. For example, using data from the UKs Companies House (or other statutory company accounts-filing sources in other countries), Cisco can pin-point companies with 150% increase in earnings year on year. These companies may need some of the technologies that commonly help deal with growth and expansion, such as call centre technology with call-waiting and call-queueing. They may want to introduce workflow technology so they can start diverting some of the queries onto their website, or to homeworkers in other parts of the company who have been hired on a pay-per-call basis. More mature companies are more likely to need technologies that help diversification, or entry into new markets. Within the customer lifecycle segmentation approach, Cisco also identifies events and triggers - compelling events that are driving technology purchases. These are categorised as inherent and known, inherent but unknown and created and unknown. Inherent and known events are within the business itself, such as an office move. Many of Ciscos solutions apply to companies that are starting up or moving premises, so it purchases lists of officemovers and targets them very successfully with outbound calls. Inherent but unknown events are classified as such because they may be known to Cisco, but are not necessarily known to the customer. For example, if the service contract on a product has expired, Cisco could call the customer with a view to renewing the contract. Created and unknown events are where Cisco envisages and communicates a commercial benefit from technology of which the customer was not previously aware. The purchasing method segmentation acknowledges that customers can buy products in a number of ways. They can purchase products and install them on their own premises, or they can have the service provided through a third-party distributor such as BT. Either of these models can be applied with a lease plan as opposed to paying for the product outright. Applying Segmentation Traditionally a sales-driven company, worsening market conditions prompted sales teams in Cisco to ask their marketing colleagues for help in defining their markets and segments. At this stage, the European HQ was beginning to deliver customer packs to different countries, designed to help sales teams plan their operations, workforce and activities around market objectives. This segmentation model has to date been launched in six European countries, where small and medium-sized businesses (SMB) were divided into three tiers using a segmentation model based on technology type. Tier 1 are experimenters; Tier 2 companies

12

would be considered the early majority in a typical lifecycle (Rogers 1962); and Tier 3 typically purchase from a service provider. The tiers direct Ciscos approach. For example, Cisco reports that Tier 2 customers have tended to be the most profitable, so if a country has a limited marketing budget, they would be encouraged to focus on Tier 2 customers. Experimenters are service-heavy, so Cisco concentrates on selling a service contract to this group. To maximize the margin in Tier 3, Cisco has concentrated on working effectively with the service provider to ensure they carry enough stock to service demand. Ciscos challenge is to support the data collection which drives segmentation programs and the actioning of them, without over-administering the whole process. A secondary problem is that budgets do not stretch to targeting smaller and smaller segments, and the whole ethos of market segmentation, is defeated, because resources are not allocated strategically in such an instance. O2 Generating Segmentation O2 is a UK-based mobile telecommunications provider (originally demerged from BT and now part of Telefonica O2 Europe plc). This company has traditionally segmented by value, industry sector, needs and behaviour. Value measures are used to determine the highest value customers, particularly in the consumer business (e.g., young socials). O2 respondents believed that targeting industry sectors was more appropriate in the corporate business, because bespoke solutions are built for them due to the size and scale of their businesses. O2 considers the needs and attitudes of customers, particularly in relation to how they use their mobile phone, in order to prioritise them based on what managers refer to as their mindstates. Segments are created (particularly in the SME business) by analysing the behaviour of existing customers, combined with data from external sources such as CACI, TGI and others. O2 also uses events and triggers in the existing customer base - for example, selecting customers whose contract is up for renewal, or corporate accounts which have not upgraded their handsets or reviewed their pricing structure for some time. Until recently, the segmentation strategy at O2 has been used predominantly to guide the customer acquisition process and to assist in outbound activity, rather than to define segments amongst existing customers. More recently, segmentation of the customer base has moved towards a one-to-one marketing model. Three years prior to the interviews, O2 had a large department of analysts carrying out project-based analysis of existing customers in order to create customer selection models for outbound promotional purposes. However, this approach was felt to be too slow and did not deliver actionable results immediately. Consequently, O2 installed predictive data-mining software, which enabled it to produce large volumes of customer response predictions very quickly. The resulting system - christened VISION - incorporates 45 propensity models predicting across the product range each customers probability of purchasing a particular product if it is offered to them, as well as their probability of ceasing to use a product or defecting entirely. The analysis is based on a combination of transactional and externally obtained data. Compared to O2s previous technology, the models can be calculated in one-seventh of the time and

13

furthermore, operate in real-time during a customer interaction, so the estimated propensities will be updated during a telephone call depending on what has happened earlier in the call. The software is customized into six versions, aligned to different product lines, sales teams and service call centres. Its goal is to enable front-line staff to make the right offer to customers, in the right place, at the right time to improve the customer experience, as well as their engagement and loyalty. The decision logic of the system allows customer service agents to assess the callers concerns, interests and risk. Propensity models are then constructed using a two-step analytical approach. Firstly, univariate analysis is used to establish the top ten independent variables for the dependent variable such as propensity to buy a given product: for example, a handset upgrade propensity model might include the time to the end of the existing contract as an independent variable. Secondly, multivariate analysis is used to predict purchase or likelihood of defection for a given customer at a given moment. Applying Segmentation The needs-based segmentation approach is used at the strategic level, to understand which markets O2 should operate in. The behavioural segmentation is still used to target some outbound marketing communications activity. However, using segmentation insights in this way was perceived to be excessively coarse when dealing with individual customers, hence the perceived need to develop the VISION system. In the inbound service call centres (handling around 50 million calls per year), the VISION system offers advice to agents on how to handle service enquiries and problems. For example, it alerts an agent if a customer has a high propensity to churn or signifies a payment risk. Once the enquiry or problem has been resolved, the system suggests the top three most appropriate products or services to discuss with individual customers, and provides a script to help agents to discuss the top one. The VISION initiative is perceived as a considerable success within the company, with impressive conversion rates being cited. Use of the system is voluntary and on average, at the time of the interviews, 73% of all customer service calls (from consumer and small-to-medium sized businesses) were entered onto VISION. Of those 73% of calls, about 38% of customers were being offered an additional product or service. In 41% of these cases, this resulted in an agent processing an order (this figure increasing to 47% if cases are included where a customer was recorded to have taken an action as a result of the offer). Approximately 50% of the offers were non-revenue generating for O2, such as a new handset or extra minutes, but the overall effect, respondents claimed, was that bill value increased by an average of 15% in the month after the offer was accepted. Retention costs were also reduced by 150-200k per month and customer churn reduced by 3%. All of these results were achieved without an increase in average call-handling time. The outbound sales teams use the system to significantly reduce the preparation time for sales calls. Often the sales department would place a call and not get through, meaning that preparation time had been wasted. As the offers were relevant and timely, it was easier to build rapport and the conversion rate went up consequently.

14

Post Office Counters Limited Generating Segmentation In 2003, the UK Post Office, responsible for post and parcel delivery, marketing and logistics services to public and business users, decided that it wanted to develop a detailed strategic segmentation strategy to be used predominantly by marketing teams, but understood and used throughout the business to help design products, size markets, differentiate marketing messages and understand their customers. When the segmentation strategy was launched in 2004, managers had no products for which they owned the customer data, only anonymous transactions, relying on external market research data to understand product usage, attitudes and behaviour. Based on a customer survey, the Post Office identified 19 segments, organized into 8 groups, which then clustered and mapped onto the general population using data from information provider Experian. By 2006, according to Experian, major changes were taking place in the UK population and, consequently, new products and new marketing initiatives were launched (e.g. Ants advertising campaigns). Additional research was also commissioned. The Post Office had decided it was time to refresh its segmentation strategy by developing its own data sources, together with those from its market research partners including Hall and Partners (Brand and Communications), Forrester (technology and internet), Millward-Brown International (Post Office usage), Experian Canvasse Lifestyle (UK level database of lifestyle characteristics), Experian UKCSD (UK level database of demographics and classifications) and MFS (financial services). The refreshed segmentation strategy contained over 1,000 main variables (compared to just eight previously), based around three key customer components: i) demographics, ii) attitudes and iii) behaviours. For each segment, the Post Office now understands: what percentage of the UK population they represent; their dominant age, income and lifestage; their preferred channel, what Post Office products they have purchased and when; how often they visit a branch; and the main reason for their last visit. It is able to list detailed key features of each segment and understands how the different segments relate to each other over time. All adults over 18 years old in the UK are now classified into one of the 19 segments or 8 groups, plotted against two key demographic variables, affluence and lifestage. The eight main groups are outlined below as follows: A: Maturing Affluence: Technology Embracing Careerists; Conservative Values B: Starting Out: Prospering Graduates; Young Active Fun Males, Females Finding Their Feet C: Optimistic Families: Aspiring Midmarket; Nestmaking Nuptials D: Autumnal Comfort: Contented Retirement; Community Mainstays; Traditional Resistors E: Blue Collar Survivors: Burdened Blokes; Hardship Balancing Females; Transient Dependants F: Cash Strapped Youth: Welfare Young Mums; Struggling Singles G: Traditional Elders: Industrial Backbone; Dignified Elders; Dependent Elders H: Welfare Reliants

15

The eight groups address the consumer side of the business. The Post Office is also starting to identify events and triggers in certain markets: for example, school and public holidays trigger demand for travel products. On the corporate side, The Royal Mail segments by value, by size of company, by industry sector and by type of relationship. Applying Segmentation The Post Office uses segmentation to drive strategy, marketing and new product development. Branch managers use the segmentation tool to determine what products are most likely to appeal to the segments that are most represented in their postal area. The segmentation tool is used to retrain the sales department to focus on understanding and identifying customer needs, rather than selling on features and benefits. The segmentation model also helps to determine which markets should be tested for which new products. By targeting segments that are likely to be receptive to particular new products, Post Office representatives report that the effectiveness of pilot launches has much improved. However, due to an historical operating model, a lack of investment in technology and a culture of serving not selling, borne of its public sector history, the Post Office is not making use of the segmentation model to create value through inbound contacts. Its inbound call centres are outsourced and managed by different suppliers on long-term contracts, aligned to particular products, making it impossible (from both a technology and a process point of view) for call centre agents to offer products that are managed by a different call centre. Discussion We discuss our findings firstly with respect to the generation of segmentation, and secondly with respect to its actioning. The actioning of customer insight, including segmentation, across the case study companies is summarised in Table 3. For simplicity, only the dominant applications of insight, in the perception of the interviewees, are shown in the table.

16

Table 3: Applying Customer Insight: Cross-Case Analysis

PERIODIC TARGET SETTING AND EVALUATION All cases: targets and progress against plan by market Barclays (lifecycle) BT (sector; value tier) Cisco (geography; size) O2 (B2B: industry sector; B2C: sociodemographic value clusters) PROPOSITION DEVELOPMENT EXECUTION

Market Predictions Customer Segments

Barclays (attitudinal) BT (needs-based; sector; value tier) Cisco (technology lifecycle) O2 (B2B: industry sector, behaviour segmentation; B2C: needs-based) Post Office (integrated segmentation)

Opportunity Analytics

Customer Value Analytics

BT (value tier) O2 (share of wallet and customer value targets)

Barclays (attitudinal: mass marcoms) BT (needs-based, sector and value tier: campaign lists, advertising) Cisco (campaign lists) O2 (customer acquisition) Post Office (test cell definition, geographic targeting, advertising) Barclays (propensity, events & triggers to prompt outbound & inbound sales offers) BT (events & triggers to prompt outbound sales offers) Cisco (events & triggers to prompt outbound offers) O2 (propensities, events & triggers to prompt outbound & inbound sales offers) Post Office (limited use of events to prompt sales offers) Barclays (customer profitability) BT (customer value, sales opportunity value) Cisco (company growth; opportunity value) O2 (product profitability, not customer profitability, influences sales offers)

Generating Segmentation The cases suggest that segmentation has not been superseded as an actionable form of customer insight, but indicate instead that segmentation forms a key component of customer insight programmes. We define customer insight as a broader term encompassing the domains of market research, segmentation and customer analytics based on a mix of transactional and external customer data. All the cases in this study were found to be aggregating and synthesizing multiple data sources to generate customer insight. However, there is no empirically-derived list of the different types of customer insight in the marketing literature, although Wills & Williams (2004) propose the following as data sources: customer database analysis, market intelligence, competitor intelligence, feedback from sales and customer service staff, including customer complaints, and financial and planning data, for examples. We have confirmed these and proposed several new sub-categories. Hirschowitz (2001) cites strategic segmentation, loyalty indicators, channel propensity, campaign propensity scores and response value scores as examples of customer

17

insight. We confirm and expand these areas. Only two of the five companies (Barclays and O2) made extensive use of propensity models, confirming previous suggestions that most companies do a poor job predicting the behaviour of their customers (Reinartz et al., 2005). Smith et al (2006a; 2006b) found practice to be hindered by a lack of appropriate data and understanding of needs-based segmentation: we support this finding, given the patchy application of needs-based segmentation in our sample. We synthesise this discussion into the following proposition: P1. Segmentation needs to be complemented by other forms of customer insight from multiple data sources in order to support both strategic and operational marketing objectives. In support of this proposition, we now turn to how companies use or apply segmentation and other complementary forms of customer insight. Applying Segmentation Customer segmentation, the action of dividing customers into like-minded groups, remains in use in all the case studies for marketing planning purposes: identifying potential target groups, prioritising these, and developing propositions for them. How this process works, including the segmentation bases used, differs substantially across the cases, but respondents did not question the need for such a process to be engaged. It is in the communication of value propositions that more significant differences began to arise across the set of cases. While mass marketing communications were reported as benefiting from attitudinal segmentation in the cases of BT and Post Office, communications through interactive channels showed an interesting trend across all the cases towards the usage of customer analytics at individual, rather than group, level. Respondents in the cases where this trend was most mature, O2 and Barclays, cited impressive improvements in communications effectiveness to support their assertion that this trend was a beneficial one. In both cases, the combination of propensity models, which allow the determination of likelihood of acceptance of a particular product/service, and rules which take into account incremental cost/benefit were used to fine-tune one-to-one communications through both outbound and inbound channels. All cases, however, claimed substantial benefits from the use of another under-studied insight category: the use of events and triggers. Gartner Group (2006) has argued that event-triggered marketing (practiced by all the companies in this study to a greater or lesser extent), yields a level of response rate typically five times greater than that obtained using traditional marketing approaches. The above discussion gives rise to some further speculative propositions, which while by no means proven, are consistent with our data and which deserve further investigation: P2. Market segmentation is most appropriate for decisions on customer selection and proposition development which need to be taken at the level of a group of

18

customers for example, broad product design, branding, pricing strategy and mass communications. P3. Where a company is interacting with an individual customer, aspects of the proposition that can be tailored to the individual level (such as cross-sale offers in inbound channels and outgoing targeted direct mail) will be most effectively informed by one-to-one customer analytics rather than just by using segment membership. With regard to P3, it is worth noting that it is not just cross-sale and retention offers which are being made on the basis of individualised insight. The cases illustrate tailoring of the customer value proposition for other aspects of the marketing mix as well: product specification (all cases), pricing (Cisco) and the choice of communications and delivery channels (all cases). These findings support various authors who have asserted that technology can play a critical role in the ability of firms to customise their service offerings (Peppers & Rogers, 1993; Pine II, 2004). Our findings support Gwinner et als (2005) proposal that inbound customers can be provided with individually tailored products and services, and Bitner, Brown, and Meuters (2000) suggestion that contact employees can handle service situations with a growing complexity that are incapable of being handled manually, improving performance through the tailoring of the customer-employee interaction to a specific customer (Sujan, Weitz and Kumar, 1994). Maister (1997) has claimed that intensive customer-employee interaction was necessary to build an accurate picture of customer needs required for successful cross-selling. Contrary to this claim, our research indicates from two case studies at least that customer insight delivered to call centre agents can lead to successful cross-selling without intensive customer-employee interaction. The fine-tuning of this individualised insight was a moving target in several of the case study organisations, with continuing refinements and consequent improvements in metrics such as conversion rates reported by Barclays and O2, in particular. The need for recalculation of propensity models even during an inbound telephone call at O2 is a logical conclusion for the insight-into-action loop once it has been technology-enabled. While the purposes for which the use of market segmentation is proposed, such as product design and the design of mass communications, have no particular requirement for fast application, this is not the case when individual customers present themselves in an outlet, on the website or in an inbound call centre. We therefore conjecture: P4. While segmentation may need to be periodically refreshed, individualised customer analytics and propensity models are most effective when generated and actioned in real-time customer interactions. We note a further implication of P3 regarding customer acquisition versus customer development. Sufficient data are more likely to exist to calculate customer analytics and propensity models in the case of existing customers than in the case of new ones, therefore the opportunity for individualised interactions is frequently greater in the case of existing customers. Consequently, we propose that:

19

P5. Segmentation is most likely to be applicable to communications with new customers, while customer analytics and propensity models are most effectively used with existing customers where transactional history data are available. This generalisation may prove, however, to be decreasingly the case as interactive IT-enabled channels continue to grow in importance and customer data are increasingly collected at market level rather than at the customer level. Next, we note the implications for measurement of effectiveness. A characteristic of propensity models is that they embed explicitly an estimate of the effectiveness of the communication based on them. As models are refined, so effectiveness improves. P3 can be tested through field experimentation. The same is not the case, however, for proposition development at the level of marketing planning, the effectiveness of which is famously difficult to assess (Harris and Ogbonna 2006). We therefore propose that: P6: Companies do not directly measure the success of their market segmentation programmes per se but are increasingly likely to measure the success of CRM programmes based on individualised customer insight provided by customer analytic data and propensity models. It is difficult to envisage methods by which companies could assess quantitatively the effectiveness of their segmentation programmes, and indeed in none of the companies was this assessed explicitly. Qualitative evaluation might, however, be feasible, through the use of multiple segmentation approaches in parallel in different business units in order to evaluate their relative usefulness, but most assessments of effectiveness are based on subjective managerial perceptions, which may or may not be accurate. Finally, we conjecture as to how a possible feedback loop can be incorporated from one-to-one customer analyses to re(de)fine segmentation approaches. We propose that companies that decide to use propensity modeling to identify customer transaction likelihood indicators can then work backwards, not only estimating customer lifetime values but also re-constructing different segment arrangements (thereby incorporating propensities) until they achieve an optimal segment configuration. Segmentation in such a situation is not only a process of deriving desirable (micro) customer segments from a mass (heterogeneous, macro) market, but could also be the opposite, a process of building-up micro-segments based on transaction likelihoods to form maps of customer groups at the macro-level. We propose that this area of market segmentation, the link between propensity modeling, customer analytics and the determination of optimal customer segments over time, is particularly worthy of further research, given the lack of understanding of dynamism in market segmentation programmes (Brangule-Vlagsma, Pieters and Wedel, 2002). We therefore propose that: P7: Highly sophisticated companies will not only manage their interactions with and tailor offerings to selected customer groups on an individual basis, but will also measure the likelihood of customers taking up new offerings using customer analytic data and propensity modelling, which when combined with segmentation

20

programmes will allow the measurement of customer lifetime value and the dynamic optimization of market segments. Nevertheless, we recognize that not all companies will be able to undertake the above approach since only those companies with a direct relationship with the end-user have the possibility of providing one-to-one service interaction, which are not present for FMCG companies, for example, LOreal who deal with trade intermediaries more often than endusers. The act of service provision per se provides both a data collection opportunity at the individual end-user level, and a context in which individualised offers can be made. Conversely, segmentation as practiced in the FMCG context - based on market research studies may be the best available method if a survey is the only way of getting customer data and the means of communication used (e.g. broadcasting) are aimed at groups rather than individual customers. Managerial Implications Our study demonstrates that in todays environment of fierce competition and intense cost pressure, the impact of customer insight extends well beyond the one outcome studied by Sinkula et al. (1997), that of marketing programme dynamism. Organisations need to develop a common understanding of what customer insight is and how it is used throughout their organization, including the link with market segmentation and CRM. Segmentation continues to be vital for marketing planning, but organizations with rich data on individual customers should consider developing and implementing propensity modelling systems (where customer data are available) and the use of triggers and events in individualised customer interactions. This will then provide the possibility of optimizing customer segments using propensity modelling data derived from CRM systems which when clustered can determine the most likely customers for specific products; an invaluable piece of information to determine the effectiveness of segmentation approaches when the two are combined. Consequently, marketers future skills requirements will change dramatically. Analytical ability will become increasingly imperative as firms shift from defining and actioning marketing programmes in identified segments to generating individualised customer insight among customer groups to craft profitable conversations with individual customers. Developing such high-powered insight is likely to require the development of new relationships with sophisticated market research agencies/IT consultancies to leverage the potential that such customer knowledge can bestow. The sales role will consequently change as their focus shifts from selling on product features and benefits to understanding and responding to customers individualised needs for a long time an aspiration, but increasingly becoming a reality. Further Research and Conclusions Further research is needed in several areas. Firstly, the claims from some respondents about the promising efficacy of events and triggers require further in-depth qualitative exploration. This could usefully uncover, via in-depth case study work among early adopters, the nature of this emerging customer insight technique both analytically and in the processes which apply it to form individualised offers. Secondly, similar in-depth exploration with early adopters is needed into the use of propensity models particularly to formulate offers in inbound channels such as call centres and the web contexts in which individualisation has in general been less advanced

21

than in outbound direct communications. In both cases, the boundary between group-level segmentation and individual insight needs further elucidation. Thirdly, these more specific studies would be well complemented by descriptive survey work exploring current practice in the generation and actioning of all types of customer insight, hence adding granularity on the issue of customer insight to the more holistic study of contemporary marketing practice by Coviello et al (2002). Finally, the efficacy of different approaches to the actioning of customer insight might be best explored through field experiments (Ryals and Wilson, 2005) using randomised assignment of customers or transactions to two or more treatments, such as one based on a conventional segmentation approach and one based on individualised propensity models. Given the clear benefits to a collaborating company of such action research, this might prove possible in terms of research access. In summary, there is precious little written in the marketing literature on customer insight, or the link between customer insight and market segmentation, or indeed the link between CRM and market segmentation. Our study is the first we are aware of to empirically identify different types of customer insight, and indicate their link to market segment definition and actioning, especially in organizations which have a relationship with their end-users. As this study was based on five largely multinational, mainly service-based companies, it represents only one particular set of practices and concerns regarding the actioning of market segmentation, indicating the presence of techniques used to action market segmentation rather than confirming their ubiquity in commerce more generally. We should note that propensity modelling, in particular, might be of much less importance to manufacturers at least in FMCG markets (although not retailers if product sales and customer data are both collected through loyalty cards) because these companies do not tend, at present, to collect data directly from their end-users. Pharmaceutical manufacturers are another case in point, because they are legally barred from marketing to end-users, although they are starting to collect data on patients for product usage/informing purposes (e.g. related to the correct use of medications). So, even in this area, we may well see change in the future. Acknowledgements The authors would like to thank representatives of Barclays, BT Global Services, Cisco Systems, O2 and the Post Office for agreeing to take part in the study. References Ansell, J., Harrison, T., & Archibald, T. (2007), Identifying cross-selling opportunities, using lifestyle segmentation and survival analysis, Marketing Intelligence & Planning, Vol.25, No.4, pp394-410. Beatty, S. E., Mayer, M., Coleman, J. E., Reynolds, K. E., & Lee, J. (1996), Customersales associate retail relationships, Journal of Retailing, Vol.72, No.3, pp.223-247. Bitner, M. J., Brown, S. W., & Meuter, M. L. (2000), Technology infusion in service encounters, Journal of the Academy of Marketing Science, Vol.28, No.1, pp.138-149. Botschen, G.; Thelen, E.M. and Pieters, R. (1997), Using means-end chain structures for benefit segmentation: an application to services, European Journal of Marketing, Vol.33, No.1/2, pp.38-58.

22

Boulding, W., Staelin, R., Ehret, M., and Johnston, W. J. (2005), A customer relationship management roadmap: what is known, potential pitfalls, and where to go, Journal of Marketing, Vol.69, No.4, pp.155-166. Brangule-Vlagsma, K.; Pieters, R.G.M. and Wedel, M. (2002), The dynamics of value segments: modeling framework and empirical illustration, International Journal of Research in Marketing, Vol.19, No.3, pp.267-285. Carson, D., and Coviello, N. (1996), Qualitative research issues at the marketing/ entrepreneurship interface, Marketing Intelligence and Planning, Vol.14, No.6, pp.51-58. Chase, R. B. , & Hayes, R. H. (1991), Beefing up operations in service firms, Sloan Management Review, Vol.33, No.1, p.15. Claeys, C.; Swinnen, A. and Van den Abeele, P. (1995), Consumers means-end chain for think and feel products, International Journal for Research in Marketing, Special Issue: Means-End Chains, Gerry Olson (Ed.) pp.193-208. Coviello, N.E., Brodie, R.J., Danaher, P.J., and W. Johnston (2002), How Firms Relate to Their Markets: An Empirical Examination of Contemporary Marketing Practices, Journal of Marketing, Vol. 66 No.3, pp. 33-46. Davies, A., Brady, T. and Hobday, M. (2006), Charting a path toward integrated solutions, MIT Sloan Management Review, 47, 39-48. Day, G.S. (1994), The capabilities of market-driven organizations, Journal of Marketing, Vol.58, No.4, pp.37-52. Dibb, S. (2001), New millennium, new segments: moving towards the segment of one?, Journal of Strategic Marketing, Vol.9, No.3, pp.193-213. Dibb, S. (2005), Market segmentation: implementation barriers and how to overcome them, Marketing Review, Vol.5, No.1, pp.13-30. Dibb, S. and Simkin, L. (1994), Implementation problems in industrial market segmentation, Industrial Marketing Management, Vol.23, No.1, pp.55-63. Dibb, S. and Simkin, L. (1997), A program for implementing market segmentation, Journal of Business and Industrial Marketing, Vol.12, No.1, pp.51-66. Dibb, S. and Wensley, R. (2002), Segmentation analysis for industrial markets: problems of integrating customer requirements into operations strategy, European Journal of Marketing, Vol.36, No.1/2, pp.231-251. Doyle, S. (2005), Real-time technologies in marketing - interaction management, Journal of Database Marketing & Customer Strategy Management, Vol.12, No.3, pp.272-278. Eichfeld, A. , Morse, T. D., & Scott, K. W. (2006), Using call centers to boost revenue, available at: http://www.mckinseyquarterly.com/article_abstract.aspx?ar=1781&L2=1, accessed 2007. Evans, K. R. , Arnold, T. J., & Grant, J. A. (1999), Combining service and sales at the point of customer contact: a retail banking example, Journal of Service Research, Vol.2, No.1, pp.34-49. Foote, N.N (1969), Market Segmentation as Competitive Strategy. In: Current Controversies in Marketing Research (Ed.) Bogart, L., Chicago, Markham Publishing Company, pp.129-139. Forsyth, J. E., Galante, N., and Guild, T. (2006), Capitalizing on customer insights, The McKinsey Quarterly, Vol.3, pp.42-53.

23

Gartner Group (2006), Improving Offer Response with Intelligent Inbound Micro-Direct Marketing, available at: http://www.marketingpower.com/webcast261.php, accessed xx month 2006. Gill, J., and Johnson, P. (1991), Research Methods for Managers, London: Chapman. Gronroos, C. (1998), Marketing services: a case of a missing product, Journal of Business and Industrial Marketing, Vol.13, No.4/5, pp.322-338. Grove, S. J. , Fisk, R. P., & John, J. (2003), The future of services marketing: forecasts from ten services experts, Journal of Services Marketing, Vol.17, No.2, p.107. Gulati, R. (2007), Silo busting: how to execute on the promise of customer focus, Harvard Business Review, Vol.85, No.5, pp.98-108. Gwinner, K. P., Bitner, M. J., Brown, S. W., & Kumar, A. (2005), Service customization through employee adaptiveness, Journal of Service Research, Vol.8, No.2, pp. 131-148. Haley, R.I. (1968), Benefit segmentation: a decision-oriented research tool, Journal of Marketing, Vol.32, pp.30-35. Harris, L. and Ogbonna, E. (2006), Initiating strategic planning, Journal of Business Research, 59, 100-111. Harrison, A. (2002), Case Study Research. In: David Partington (Ed.), Essential Skills for Management Research: London: Sage Publications, pp.158-180. Hirschowitz, A. (2001), Closing the CRM loop: The 21st century marketer's challenge: Transforming customer insight into customer value, Journal of Targeting, Measurement and Analysis for Marketing, Vol.10, No.2, pp.168-179. Kale, and Sudhir H. (2004), CRM Failure and the Seven Deadly Sins, Marketing Management, Vol.13, No.5, pp.42-47. Keaveney, S.M. (1995), Customer switching behavior in service industries: an exploratory study, Journal of Marketing, Vol.59, April, pp.71-82. Kelley, S. W. (1993), Discretion and the service employee, Journal of Retailing, Vol.69, No.1, p.104. Kelley, S.W.; Hoffman, D. and Davis, M.A. (1993), A typology of retail failures and recoveries, Journal of Retailing, Vol.69, Winter, pp.429-452. Kumar, V., Venkatesan, R. and Reinartz, W. (2006), Knowing what to sell, when, and to whom, Harvard Business Review, Vol.84, No.3, pp.131-137. Laiderman, J. (2005), A structured approach to B2B segmentation, Database Marketing and Customer Strategy Management, Vol.13, No.1, pp.64.75. Langford, R., and Schulz, K. (2006), Gaining 3-D customer insight to drive profitable growth, Strategy and Leadership, Vol.34, No.2, pp.21-27. Maister, D.H. (1997), True Professionalism: The Courage To Care About Your People, Your Clients And Your Career, New York: Free Press Marketing Leadership Council (2007), Mastering Marketing Fundamentals Series: Marketing Segmentation, Teleconference on 27 March, available at: https://www.mlc.executiveboard.com/Public/Default.aspx McCutcheon, D. M., and Meredith, J. R. (1993), Conducting case study research in operations management, Journal of Operations Management, Vol.11, No.3, pp.239-256. McDonald, M. and Dunbar, I. (2005) Market segmentation. Butterworth Heinemann, Oxford. Oliver, R. W., Rust, R. T., & Varki, S. (1998), Real-time marketing, Marketing Management, Vol.7, No.4, pp.28-37.

24

Palmer, R. and Millier, P. (2003), Segmentation: identification, intuition, and implementation, Industrial Marketing Management, Vol.33, pp.779-785. Payne, A. and Frow, P. (1999), Developing a segmented service strategy: improving measurement in relationship marketing, Journal of Marketing Management, Vol.15, pp.797-818. Payne, A., and Frow, P. (2005), A strategic framework for customer relationship management, Journal of Marketing, Vol.69, No.4, pp.167-176. Peppers, D., and Rogers, M. (1993), The One to One Future, London: Piatkus. Pine II, B. J., Peppers, D., and Rogers, M. (1995), Do You Want to Keep Your Customers Forever?, Harvard Business Review, Vol.73, No.2, p.103. Reichheld, F., & Sasser Jr, W. E. (1990), Zero defections: quality comes to services, Harvard Business Review, Vol.68, pp.105-11. Rigby, D. K., Reichheld, F. F., and Schefter, P. (2002), Avoid the four perils of CRM, Harvard Business Review, Vol.80, No.2, pp.101-9. Rogers, E.M. (1962), Diffusion of Innovations, New York: Free Press. Ryals, L. and Wilson, H. (2005), Experimental methods in market research: from information to insight, International Journal of Market Research, Vol.47, No.4, pp.347366. Sinkula, J. M., Baker, W. E., and Noordewier, T. (1997), A framework for market-based organizational learning: Linking values, knowledge, and behavior, Journal of the Academy of Marketing Science, Vol.25, No.4, pp.305-319. Slater, S. F., & Narver, J. C. (2000), Intelligence generation and superior customer value, Journal of the Academy of Marketing Science, Vol.28, No.1, pp.120-127. Smith, B., Wilson, H., and Clark, M. (2006a), Creating and using customer insight: 12 rules of best practice, Journal of Medical Marketing, Vol.6, No.2, pp.135-139. Smith, B., Wilson, H., and Clark, M. (2006b), From data to dividends: what makes some firms better than others at turning information into value?, Working Paper, Cranfield School of Management, Cranfield, Bedford, UK. Smith, G.E. (2002), Segmenting B2B markets with economic value analysis, Marketing Management, March/April, pp.35-39. Smith, W.R. (1956), Product differentiation and market segmentation as alternative marketing strategies, Journal of Marketing, July, pp.3-8. Spencer-Matthews, S., & Lawley, M. (2006), Improving customer service: issues in customer contact management, European Journal of Marketing, Vol.40, No.s1/2, pp.218232. Sujan, H.; Weitz, B.A. and Kumar, N. (1994), Learning orientation, working smart and effective selling, Journal of Marketing, Vol.58 (July), pp.39-52. Van Raaij, E. M. (2005), The strategic value of customer profitability analysis, Marketing Intelligence and Planning, Vol.23, No.4, pp.372-381. Vargo, S. and Lusch, R. (2004) Evolving to a new dominant logic for marketing. Journal of Marketing, 68(Jan), 1-17. Verhallen, T.M.; Frambach, R.T. and Prabhu, J. (1998), Strategy-based segmentation of industrial markets, Industrial Marketing Management, Vol.27, No.4, pp.305-313. Wang, H.-F., and Hong, W.-K. (2006), Managing customer profitability in a competitive market by continuous data mining, Industrial Marketing Management, Vol.35, No.6, pp.715-723.

25

Wedel, M. and Kamakura, W. (2002), Introduction to the special issue on market segmentation, International Journal of Research in Marketing, Vol. 19, No.3, pp.181-183. Wells, W.D. and Gubar, G. (1966), Life cycle concept in marketing research, Journal of Marketing Research, Vol.III, November, pp.355-363. Wills, S., and Webb, S. (2007), Measuring the value of insight - it can and must be done, International Journal of Market Research, Vol.49, No.2, p.155. Wills, S., and Williams, P. (2004), Insight as a strategic asset - the opportunity and the stark reality, International Journal of Market Research, Vol.46, No.4, pp.393-410. Wilson, H., Daniel, E., and McDonald, M. (2002), Factors for success in Customer Relationship Management (CRM), Journal of Marketing Management, Vol.18, No.1/2, pp.193-219. Wind, Y. (1978), Issues and advances in segmentation research, Journal of Marketing Research, XV, August, pp.317-337. Wong, V. and Saunders, J. (1993), Business orientations and corporate success, Journal of Strategic Marketing, Vol.1, No.1, pp.20-40. Yankelovich, D. and Meer, D. (2006), Rediscovering market segmentation, Harvard Business Review, Vol.84, No.6, pp.141-145. Yin, R. K. (2003), Case Study Research: Design and Methods, 3rd Edition, Thousand Oaks, CA.: Sage. Young, S.; Ott, L. and Feigin, B. (1978), Some practical considerations in market segmentation, Journal of Marketing Research, Vol.XV, August, pp.405-412. Zeithaml, V. A., Berry, L. L., & Parasuraman, A. (1988), Communication and control processes in the delivery of service quality, Journal of Marketing, Vol.52, No.2, p.35.

26