Industrial Marketing Management 39 (2010) 784–792

Contents lists available at ScienceDirect

Industrial Marketing Management

Segmenting a market in the making: Industrial market segmentation as construction

Debbie Harrison a,⁎, Hans Kjellberg b,1

a Department of Strategy and Logistics, Norwegian School of Management BI, Nydalsveien 37, 0442 Oslo, Norwayb Stockholm School of Economics, Department of Marketing and Strategy, PO Box 6501, S-11383 Stockholm, Sweden

⁎ Corresponding author. Tel.: +47 46 41 04 71.E-mail addresses: Debbie.Harrison@bi.no (D. Harriso

(H. Kjellberg).1 Tel.: +46 87 36 95 23.

0019-8501/$ – see front matter © 2009 Elsevier Inc. Aldoi:10.1016/j.indmarman.2009.05.016

a b s t r a c t

a r t i c l e i n f o

Article history:Received 16 November 2007Received in revised form 30 March 2009Accepted 16 May 2009Available online 8 September 2009

Keywords:Market segmentationMarket shapingInteractionMarketsNetworks

This paper takes an empirical starting point in a claim that Biacore, a pioneering Swedish producer of affinitybiosensors, was “in the enviable position of creating its own market” (Abelin, 1997). An in-depth case studytraces how Biacore undertook segmentation activities while shaping the market for its new producttechnology, affinity biosensors. This involved stabilising the modes of exchange with customers, the productand the identity of the company. The efforts of Biacore highlight a constructive dimension of market seg-mentation that hitherto has received little attention. Rather than a process of describing, deciding and takingaction, Biacore engaged in the gradual construction of market segments through an interactive and iterativeprocess involving close collaboration with early users. Simultaneously, the market for the new technologygained shaped. Thus, the paper reports a ‘markets from networks’ story illustrating how the practice ofsegmenting a market may have consequences for that market.

© 2009 Elsevier Inc. All rights reserved.

1. Introduction

In connection to the initial stock market listing of Biacore, a pio-neering Swedish producer of affinity biosensors, stock market analystsEnskilda Securities observed that Biacorewas “in the enviable position ofcreating its own market” (Abelin, 1997:3). For Biacore, the work ofcreating thismarket had startedmore than a decade earlier as part of thedevelopment project that led to the launch of the first biosensor in late1990. During product development, the company had been forced toabandon their envisioned major market segment, medical clinics, andinstead focus on research labs, a much smaller potential segment.Following the launch, Biacore worked closely with a gradually growingnumber of such customers to develop applications for the product. Later,thefirmmadeuseof these experienceswhendevisingmore standardisedmarketing programs for different customer groups. This work couldbroadly be characterised as a process of industrial market segmentation.

As part of marketing theory, the principal view of what marketsegmentation is has remained fairly stable. The basic assumption isthat there are groups of customers ‘out there’ in an existing market,with specific characteristics and responses to a marketing mix. As aresult, considerable research efforts have been put into refining thesegmentation process and the diagnostic techniques both forconsumer and industrial markets in order to obtain the best possible‘market map’ or description. As such, market segmentation is con-sidered as the process of identifying relatively homogeneous customer

n), hans.kjellberg@hhs.se

l rights reserved.

groupswithin a definedmarket (Smith, 1956;Wind & Cardozo, 1974).More recentwork argues that the homogeneity of customer groupingsis a managerial assessment, rather than a naturally occurring marketphenomenon (Quinn, Hines & Bennison, 2007).

In the case discussed in this paper, there was nomarket for affinitybiosensors at the outset. Indeed, therewere no biosensors. Under thesecircumstances, we argue that the production of an actionable map ofthe market— segmentation — requires more than the employment ofdescriptive techniques. Market segmentation becomes an emergentand interactive process of shaping that which is to be described. Thus,Biacore undertook segmentation activities alongside efforts to stabilisethe modes of exchange with customers, the product and their ownidentity as a firm.

The overall purpose of the paper is to increase our understanding ofindustrial segmentation processes. Two specific research questions areaddressed. First, what is the character of the market segmentationprocess when performed on a market for a new technology that is ‘in themaking’? Furthermore, recent literature in industrial market segmen-tation suggests that the practice of segmenting a market potentiallyhas consequences for that market. However, there is a paucity ofempirical research on how ex ante efforts to segment markets con-tribute to shape those markets. Thus, we also address a secondresearch question: how do a firm's segmentation activities contribute to(re)shape the market being segmented?

Four sections follow this Introduction. In the Section 2, we reviewrelevant segmentation literature. In Section 3, we present the casestudy. In Section 4, we analyse the case in order to make comparisonswith the literature. Lastly, in Section 5 we conclude by discussing ourfindings and their implications. The case shows how Biacore engagedin an interactive process of segment construction when their initial

785D. Harrison, H. Kjellberg / Industrial Marketing Management 39 (2010) 784–792

attempt to address an envisioned target market segment failed. Theco-development of use applications with users/customers generatedseeds for futuremarket segments iteratively over time. Based on theseobservations, we argue that different market segmentation activitiesmay be called for when segmenting markets in the making ascompared to segmenting ready-made markets. Our major conclusionis that current conceptualizations of industrial market segmentationfail to acknowledge this by exclusively focusing on segmentation as adescriptive practice.

2. Industrial market segmentation: identification, diagnosisand shaping

Research on industrial market segmentation conducted in the 1970sand 1980s regarded segmentation as a means for making informedchoices concerning which customers the company should serve. Effortswere consequently focused on refining the bases and techniques foridentifying segments (e.g. see Wind & Cardozo, 1974; Bonoma &Shapiro, 1983; Webster, 1984). Broadly, existing research defines amarket segment as “a group of present or potential customers withsome common characteristic which is relevant in explaining (andpredicting) their response to a supplier's marketing strategy” (Wind &Cardozo, 1974:155). The managerial objective is one of segmentidentification or “identifying relatively homogenous customer groupswithin a defined market” (Ibid.). It is vital that groupings of customershave homogeneous needs and respond to marketing programs, such aspersonal selling and advertising, in similar ways (Choffray & Lilien,1980; Dibb, 1998) and that segments are of sufficient size and areaccessible (Wind & Cardozo, 1974; Choffray & Lilien, 1980; Quinn et al.,2007).

Wind and Cardozo (1974: 155) express the tension betweenheterogeneity of customer needs and economic necessity thus: “some-times identification may be very difficult or economically unfeasible, inwhich case the industrial marketer faces essentially one undifferentiatedset of buyers. At the other extreme, each individual customer mightconceivably constitute a segment. Unless the customers were few innumber and each economically significant, the marketer would face anarray of virtually unmanageable variety”. Section 2.1 below outlines thegeneral processes for identifying segments within a market.

2.1. Processes for identifying and diagnosing industrial markets

There are two classic approaches tomarket segmentation (Shapiro&Bonoma, 1984). Both are primarily concerned with identifyinghomogenous groups of customers in terms of descriptive characteristicsrather than response patterns (Wilson, 1986). The ‘customer needs’approach groups customers with shared needs into segments on theassumption that needs are both known and relatively stable. However,‘benefits sought’ is not a categoryoftenused in industrialmarketing (e.g.Moriarty & Reibstein, 1986), perhaps because it is difficult to obtaininformation about micro-level variables (Wilson, 1986).

The second approach is to start with the firms in the market, andwork through a disaggregation process to locate the most appropriatesegmentation bases (e.g. Wind & Cardozo, 1974). This reflects a viewof segmentation as a technique for diagnosing markets (cf. Plank,1985). The basis of segmentation can be: industry, organisational,DMU/buyer and product characteristics, e.g. application, organisa-tional structure, new versus repeat purchase (Wind & Cardozo, 1974),structure of the purchasing decision process (Choffray & Lilien, 1980)and organisational strategy (Verhallen, Frambach, & Prabhu, 1998).

The macro–micro conceptual model of Wind and Cardozo (1974)is the classic diagnostic segmentation process. The first stage is toidentify macro segments based on organisational characteristics. Thisis followed by the selection of acceptable segments, and anassessment of whether these segments will respond to a marketingmix. If this is the case, the segmentation process can be halted. If not,

the second stage is to identify micro segments within the “acceptable”macro segments, based on DMU characteristics. Afterwards, microsegments are selected based on some assessment of costs andbenefits.

Shapiro and Bonoma's (1984) multi-level segmentation processmodel of ‘nests’ is also well known. The authors argue the disaggrega-tion approach is cheaper and easier for the firm to attempt, but thatthere can be problems in implementing such approaches. These arerelated to the complexity of segmentation techniques, barriers blockingthe process and the ‘academic-practitioner gulf’ (Plank, 1985; Dibb,Stern, & Wensley, 2002; Palmer & Millier, 2004). As Dibb and Simkin(1994: 57) suggest; “although academics stress the need to identify themost suitable and statistically valid segmentation schemes, the priorityof the practitioner is to identify segments for which an effectivemarketing program can be developed”.

One explanation for problems in implementing market segmen-tation is that “segmentation appears to be largely an after-the-factexplanation of why a marketing program did or did not work” (Wind& Cardozo, 1974:155). Recent literature has suggested directly work-ing with practitioners in order to improve the accuracy of segmen-tation processes, such as through the use of workshops. For example,Palmer and Millier's (2004) three-part process is that of; ‘definecriteria’, ‘develop scales’, ‘test and iterate’ against the existing cus-tomer base. As the authors state, “by working within an organisation,generic issues of segmentation aremademuchmore specific when setwithin context” (p. 783).

2.2. Segmenting industrial markets in the making

The complexity of diagnostic segmentation techniques, barriersblocking the process and the ‘academic-practitioner’ gulf are well-recognised concerns in industrial market segmentation. For markets ‘inthemaking’, i.e.marketswhose characteristics are still in importantwaysundetermined, additional concerns can be raised. In particular, the ideaof a clear separation between thinking and action becomes problematic,as do certain basic assumptions about the markets being segmented.

Both the main approaches for segmenting industrial marketsproceed by way of identifying segments on existing markets usingvarious descriptive techniques (Plank, 1985). The managerial objec-tive is one of identifying segments using the best possible descriptivetechnique. In other words, while the specific approaches differ, thebasic rationale is the same— that of obtaining the best possible map ofthe market terms of segments. In turn, this should allow the firm todevise the most appropriate marketing program based on e.g. price,product, place, and promotion modules. Thus, there is an assumedone-way causal relationship between the description of the marketand the marketing program.

The availability of information for suchanundertakinghas longbeenquestioned (e.g. Balakrishna, 1978), along with the lack of explanationof buyerpreferences for particular suppliers (e.g.Wind&Cardozo, 1974;Cunningham & White, 1973; Webster & Wind, 1972). In particular,segmentation practice needs to be re-assessed if customers are activeand there is interaction between multiple customers and the supplier(Freytag & Clarke, 2001), rather than a firm-centred process. Ratherthan a process clearly separating thinking from action, interactivemarket segmentation would involve incremental sequences of actionsand decisions forming a pattern retrospectively based on activenetworking (cf. Mintzberg & Waters, 1985; Håkansson & Ford, 2002):the market map is shaped in a ‘live’ emergent process.

Moreover, arguing that the managerial objective is to obtain thebest possible description of the market, suggests that this marketexists and is known or knowable. That is, it is assumed that themarketboundaries, suppliers, users, products and uses are recognised, wellknown and set. The market is already made up of potential and actualcustomers that have varying characteristics and behaviours. It is theirspecific characteristics that can be uncovered through segmenting “…

2 BIA is short for Biomolecular Interaction Analysis.3 All of the information provided in describing the features of the BIAcore tool is

taken from http://www.Biacore.com/technology.

786 D. Harrison, H. Kjellberg / Industrial Marketing Management 39 (2010) 784–792

after-the-fact”. But if markets are becoming more diverse orfragmented (Firat & Shultz, 1997; Sheth, Sisodia, & Sharma, 2000;Quinn et al., 2007) marketers need to re-assess their segmentationpractice in various ways.

Sheth et al. (2000) suggests that marketers need to move fromtrying to identify segments to focus on the needs, wants and resourcesof individual customers. Quinn et al. (2007) advise marketers to relyless on the traditional, planned and static segmentation approach andallow for more intuitive understandings to handle complexity andambiguity. In their reconsideration of segmentation practice, Palmerand Millier (2004) discuss the role of intuition, knowledge gained byexperience, in market segmentation practice in established markets,while Millier (2000) focuses on intuition in emergingmarkets for newtechnologies. As he neatly puts it, “…all these mechanistic methodsare limited when the clients are few, when the markets are concen-trated, or when they do not yet exist...” (p. 149).

Although these ideas differ from the classic approaches in usingworkshop methods and taking into account a variety of additionalconcepts for re-assessing segmentation practice, the assumptionsunderpinning the processes of mapping markets are essentially thesame. There is a clear focus on improving the accuracy of practice, ratherthan concentrating on acting and the consequences from this. In otherwords, if fewauthors have questioned the standard assumptionof a pre-existing market, even fewer have addressed that segmentation effortshave consequences.

If segmentation is a descriptive practice only, the most importantissue is to obtain an actionable map of the market. The actionsundertaken by a firm over time, and the actions feeding back fromcustomers, are largely left unexplored. Managerial action andsupplier–customer interaction may result in considerable changes inthe way in which a market map appears and what activities can beundertaken. Freytag and Clarke (2001: 474) argue that segmentation“shares in shaping the environment” and can affect the internalworkings of the company. Furthermore, Griffith and Pol (1994: 39)argue: “… some of the benefits of segmentation have not beenrealized due to […] the tendency to segment markets after they havebeen established (at which point management becomes descriptiveand reactive rather than prescriptive and proactive)”.

These ideas connect well with the recent attention given to theshaping of markets. Jaworski, Kohli and Sahay (2000) and Kjellberg andHelgesson(2007)have suggested thatfirmshaveamuchwider repertoireof action alternatives than what is commonly recognised in marketing.More specifically, they argue that firms may seek to drive or managemarkets, in addition to reacting to a given market structure. As Araujo(2007: 223) concludes: “Marketing contributes to the reproduction andtransformation ofmarket structures.” In otherwords,firms' segmentationefforts may influence the workings of the markets they partake in.

Below, we present a case study of an emergent segmentationprocess conducted by a firm within a market that is very much in themaking. The case illustrates how the firm's segmentation efforts con-tribute in important ways to shape that market beyond the broadcategories of internal firm structure and environment suggested bythe existing industrial market segmentation literature.

3. The case of Biacore and the affinity biosensor market

3.1. Notes on method

The single case study is the research methodology adopted for thepaper. Various authors provide both practical suggestions for carryingout case research and the philosophical bases of case studies (e.g.Abbott, 1992; Easton, 1995; Dubois & Araujo, 2007). Case studiesembed an object in context, allow depth, detail, and richness of data,are longitudinal by default (Easton, 1998) and are process-oriented. Asingle case was appropriate to obtain an in-depth account of seg-mentation in a market for a new technology that was in the making.

The selection of the case was based on a mixture of ongoing researchlinks and theoretical interests (Easton, 1995; Halinen & Törnroos,2005; Dubois & Gadde, 2002). The Biacore project reported herewas part of a larger study regarding the use of science in business inUppsala, Sweden (Harrison & Waluszewski, 2008).

In terms of data collection, a mixture of primary and secondarydata sources was used to construct the case study. Thirty six face-to-face and telephone interviews were conducted over an 18-monthperiod between 2002 and 2004, supplemented with four additionaltelephone interviews carried out in summer 2006. At Biacore inter-views were conducted with individuals in management, sales andmarketing, technical specialist and senior applications developerroles. Both former and current employees were interviewed, with theduration of each interview being one to 6 h. Twenty users were alsointerviewed. These individual scientists were located in bothuniversity (9 users) and commercial (11 users) settings in the UKand Scandinavia. The university users held either post-doctoral orprofessorial roles in areas such as molecular or microbiology, and inimmunology. The commercial users included senior research scien-tists in pharmaceutical companies, technical developers and R&Dtechnicians. The interviews with users were between 1 to 2 h induration. It was necessary to maintain detailed records of the dataanalysis process in order to limit the possibility for bias in the inter-pretation of data.

In addition, descriptive material relating to a publications databaseand financial data has been collected in order to complement theinterview material. This includes the Biacore website (www.biacore.com), along with press releases and clippings accessed through twodatabases for business information, the Swedish Affärsdata (http://www.ad.se/) and the international Factiva (http://global.factiva.com).

The multiple data sources were combined to generate a chronol-ogy of the development of the firm, product and customers, using thelogic of longitudinal process research (Huber & Ven de Ven, 1995;Langley, 2007). The data collected from the various interviews andsecondary sources were then divided into three time periods: Biacoreas a product development project (1984–1990); Biacore as a businessunit (1990–1995); and Biacore becoming a formally independentcompany (1995–2000). This organisation of the descriptive accountwas based on two important events in the development of Biacore astold by the interviewees; the planned product launch in 1990 and theplanned stock market launch in 1996.

The data analysis involved a comparison of the Biacore case andcurrent industrial market segmentation literature in a search forsimilarities and differences (Eisenhardt, 1989). There are clear limit-ations in terms of statistical generalisability from exploratory casestudy. Instead, the basis of generalisation beyond the boundaries of acase study is analytical (Mitchell, 1983). Hence, our aim was to makeanalytical or theoretical generalisations to the existing body of re-search concerned with market segmentation in industrial settings.That is, the findings can be used to further develop current theoreticalideas regarding industrial market segmentation, but not to makeclaims about populations of firms or markets.

3.2. Background: biosensors and their use

In brief, the BIAcore2 biosensor instrument combines three mainfeatures: sensor chip technology, microfluidics and surface plasmonresonance (SPR) detection.3 A target molecule is immobilised on tothe sensor chip in order to measure its binding interaction betweenproteins, nucleic acids, carbohydrates, lipids, etc. SPR detects changesinmass on the surface of the chip as the target molecule interacts withmolecules in a test solution. A scientist is able to follow the molecular

Table 1Biacore's combinations of user type, application areas and instrument versions in the Protein Interactions Analysis market.

User types Applications Instrument types

Life Science (in particular academic customers) University laboratoriesResearch centresCommercial organisations

Molecular causes of disease• cancer• neuroscience• immunologyFunctional proteomics• ligand fishing• protein–protein interactions

BIAcore XBIAcore 3000

Pharmaceutical/biotechnology Commercial organisationsResearch centres

Antibody characterisation for drugdiscovery and development• identify and screen target compounds• antibody developmentLead optimisation

BIAcore 2000 and BIAcore 1000(depending upon sensitivity requirements)BIAcore S51

787D. Harrison, H. Kjellberg / Industrial Marketing Management 39 (2010) 784–792

interaction in real-time via a software that produces a sensorgramsummarising the quantitative information generated by the SPRdetector.

The contemporary use of the product is in protein interaction appli-cations such as antibody characterisation and proteomics.4 Scientistscan use methods such as chromatography and electrophoresis to sep-arate and purify mixtures of proteins. A BIAcore can then be used tostudy the interactions between these proteins. University laboratories,research centres and commercial organisations in life science are usersfor applications such as the identification of DNA damage. ‘Hit to leadcharacterisation’ (rapid affinity ranking and detailed kinetics of inter-action for small molecules binding to target proteins) is an example of adrug discovery and development application for commercial users inpharmaceutical and biotech firms.

The first product was launched in 1990 and a decade later approx-imately 1000 Biacore systems had been installed in universities andcommercial research labs across the world. During this period, theproduct range had also grown to four instruments: BIAcore 3000,BIAcore 2000, BIAcore 1000 (which had replaced the originalBIAcore), and BIAcore X. (see Table 1). The company was growingrapidly with annual sales reaching MSEK 439 during the year 2000(+29%). The following three sections will provide an account of howBiacore developed from a product idea to a rapidly growing biotechequipment supplier.

3.3. 1984–1989: Biacore as a development project

Biacore started as a development project in the early 1980s withinPharmacia Biotech, at the time one of the world's largest suppliers ofbiotechnology analytical tools. The basis for the project was to inves-tigate whether it was possible to utilise surface plasmon resonance(SPR) technology (a light phenomenon that enables tracing interac-tions between molecules) to develop a biosensor instrument. At thetime, SPR was a well-recognised phenomenon within applied physicsand the project group interacted with one of the leading researchunits in the area, Applied Physics at Linköping Institute of Technology,Sweden. The project was formalised as a separate subsidiary in 1984.By 1987, the total cost for the biosensor project was estimated toabout 400–500 MSEK with the product launch planned for mid-1990.

The main supporter of Biacore was Pharmacia Diagnostics, a well-established and rapidly growing producer of allergy diagnosticdevices and a sister company to Pharmacia Biotech.5 Diagnosticsprovided the project with a first potential application area and a set ofpotential customers; medical personnel would use the new instru-

4 Biacore press releases 2004, “Biacore announces update on ‘Strategic BusinessReview’”.

5 Dagens Industri “Fortias försäljning” 1982-08-12; Veckans Affärer “Starkt Pharmaciaändrar ej prognos” 1984-05-10; TT Nyhetsbanken “Pharmacia” 1985-03-05; DagensIndustri “Pharmacia: +17%” 1985-11-15.

ment to analyse samples on-site, rather than sending away blood teststo an external laboratory. The choice of diagnostics as the majorapplication area for the new instrument promoted it as a comple-mentary technology to Pharmacia Diagnostics' existing product range,in particular the fast protein liquid chromatography system (FPLC).Biacore would piggyback on Pharmacia Diagnostics' established salesand marketing organisation. In addition, Biacore could also draw onthe experience of Pharmacia Biotech concerning the development ofsystem solutions for separation and characterisation of bio-molecules.It was considered that the price of the product needed to be relativelylow if it was to be located in many medical clinics. The productionvolume for the new instrument was estimated at 10,000 units peryear.

A second potential market was also identified within the project:selling the new instrument as a research tool for characterising bio-molecules. This implied a different and much smaller set of potentialcustomers — research labs located at universities and in bigpharmaceutical companies. Clearly, the diagnostics application wasthe more commercially interesting. In the words of Erik Danielsson,the then-general manager of Pharmacia: “Biosensors are what we gohardest for today. If we succeed biosensors have a good prerequisite tobe our largest product ever.”6

In January 1989, with the launch of the instrument estimated totake place within a year and a half, the planned diagnosticsapplication was still put forward as an important future market. Thehead of research and development for the Pharmacia group estimatedthe potential for biosensors in the medical diagnostics market at€400–1000 million, whereas the market for biosensors as researchinstruments was estimated at €100 million.7 However, PharmaciaDiagnostics' main customers in diagnostics laboratories had started toraise concerns. There were potential issues of cannibalisation, as themedical clinics were important users of the diagnostics laboratories.In addition, the geographical market where biosensors for diagnosticuse were expected to take off first andmost rapidly was the US, wherePharmacia Diagnostics' position was weak.8 The outcome was thatPharmacia Diagnostics left the Biacore project. In so doing, theyessentially withdrew the diagnostic application and access to thepotential customer base.

As a result, only the smaller potential market for biosensors asresearch instruments remained. Officially, the head of research anddevelopment for the Pharmacia group emphasized that the targetgroup in the research market, besides being “well definable andhomogenous”, had been anticipating the new technology for years.9

Still, the estimated production volumes had to be reduced to a few

6 Interview in the Swedish business magazine Veckans Affärer, 1987, No 3, p. 4.7 Veckans Affärer, “Huvudmarknaden värd 4–10 miljarder”, 1989-01-19.8 Veckans Affärer, “Pharmacias biosensor når marknaden 1990”, 1989-01-19.9 Ibid.

788 D. Harrison, H. Kjellberg / Industrial Marketing Management 39 (2010) 784–792

hundred instruments per year. At the same time, investments in anew building and in recruiting 90 employees had already been made.

3.4. 1990–1995: Co-development in interaction with users

The disappointing developments with respect to the diagnosticsapplication resulted in the biosensor project becoming a problem forthe project group, Pharmacia Biotech and for the external financersthat had invested in the project. Activities now centred on launchingthe instrument for the research market. Biacore staff worked on whatthey considered to be two key application areas internally. Thesewerecharacterisation methods such as epitope mapping and concentrationmeasurement. “At this time it was hard to provide convincingapplications. We needed convincing examples in order to be able torun demonstrations or workshops at customer sites”.10 Employeesrecruited from Pharmacia Biotech were able to identify currentPharmacia customers that were interested in new technologies. Asmall number of researchers from institutions such as Institute deBiologie Moleculaire et Celulaire in Strasbourg, France were invited toexperiment with a biosensor instrument. Together with these andother early potential opinion leaders some characterisation experi-ments were developed.

In September 1990 the first workshop presentation of thebiosensor occurred at Pharmacia Biotech's European sales office inFreiburg, Germany. The selected Pharmacia customers were used aspresenters of some first results. Eight in-house technical staff, ratherthan sales people, presented the tool to potential customers in aworkshop format over 2 days. The potential customers were scientistsknown to the Pharmacia sales representatives in Germany. It was “realexperiments with all the risks…full exposure of the product.Biosensors were perceived as promising, but questions remainedregarding the value of the instrument vis à vis what other techniquescould do”.11

Similar presentations took place throughout Europe in 1990 and1991. The potential customers — scientists from the pharmaceuticalindustry and universities—were responding to advertising flyers thatdescribed the basic features of the product. They perceived potentialbenefits from using the new biosensor ‘ahead of the general market’.During the workshops there was ‘lots of talking, sharing anddiscussing the possibilities. It was not very driven in a sales processsort of way…’.12 For example, researchers from the Medical SciencesDivision at Oxford University attended a UK workshop at PharmaciaBiotech's offices in Milton Keynes. They were initially sceptical as towhether a biosensor was suitable for their research in the area ofstudying weak interactions in antibody binding. However, a BIAcorewas purchased after some initial results were obtained in interactionwith Biacore technicians. As one of the ongoing users commented,“What was interesting was that Biacore did not sell the machine forweak interactions”.13

The multiple launch workshops started the process of pre-salesinteraction between Biacore technicians and potential customers. Thiswas not a process of providing a potential customer with existingknowledge. Biacore staff also helped the potential users to learn howto use the biosensor. The interaction was an intensive problem-solving process in order to connect a scientist's research ideas to thebiosensor instrument. The interaction occurred both at Biacore's labsand at the user's site, often for many weeks at a time. The results ofexperiments conducted at the workshops, the customer's site or atBiacore's labs were recorded. The work was frequently performedwith a machine on loan from the company. “It was not important to

10 Interview with a Biacore Applications Developer, Product Development Depart-ment.11 Ibid.12 Interview with a Biacore Product Development Technician.13 Interview with users at Oxford University, UK.

Biacore where the customers were. We were just going to talk andhope that the customer bought…”.14

The after-sales process was similar. It included discussions ofresults, on-site visits and development projects. This time-consuminginteraction was free of charge and Biacore effectively developed closerelationships with all of the early users. In the words of a productdevelopment technician, “in 1990 one question could get you lots ofattention. We spent lots of time early on with a small number ofcustomers. It was important for positioning and credibility”. For ex-ample, for one scientist at Astra Zeneca R&D, the after-sales processdeveloped into an ongoing discussion about how to use the instrumentin studies of small molecule interactions.

Early in 1992, while these activities were taking place, Biacore wasmoved out of Pharmacia Biotech and placed as a separate businessproject reporting directly to group management as part of an internalreorganisation of the group that owned Pharmacia (Andersson,1996:85). Furthermore, the relations between Biacore and themarketing organisation of Pharmacia Biotech also changed as Biacoresuccessively established its own sales forces in the UK, France andGermany, while signing distribution contracts with Pharmacia Biotechfor sales in other national markets. Overall, these developmentscontributed to make Biacore more independent.

After the launch workshops, multiple customers/users began toapply “all kinds of different applications…there was an explosion inthe literature of articles assessing the different ways of using themachine”,15 aided and encouraged by Biacore. Many of the uses werenot foreseen or intended by Biacore. The early users offered feedbackto the continued development of the instrument in a variety of ways,both concerning its use and its features per se, e.g. software upgrades.Further, publications began to appear, in particular those written byusers in university environments.

Biacore marketing staff recorded and amassed these publicationsand formed a database that could be accessed by users of the ins-trument. As could be expected, the early articles often includedBiacorestaff. However, the number of publications referencing BIAcore grewdramatically over the following years, and the prevalence of in-houseauthors decreased. By the end of 1995 approximately 400 articles hadbeen published. Five years later the figure was 2000. Further adding tothe scientific touch of their marketing approach, Biacore establishedthe ‘BIA Journal’ in 1994. Biacore employees selected published papersfrom the ongoing reference database, or requested papers fromcustomers “who were doing interesting work”.16 In all, there were22 issues of the BIA journal between 1994 and 2005.

A second important marketing channel was established when theinitial launch workshops were replaced with seminars where userscould present their results. The first of these ‘BIAsymposia’ took placein 1992. For Biacore, these meetings were a place for intensivediscussions about how customers used their biosensor. For the users,they were “a place to talk ‘real science’ with others who were usingthe same technology in a range of different applications…a fantasticand real discussion forum”.17 Reports from the BIAsymposia were alsopublished in the BIA journal to try to increase impact.

Biacore also took measures of its own to further develop theinstrument. A lower priced version of BIAcore called BIAlite, to someextent born out of the original plan to concentrate on the diagnos-tics application, was introduced in September 1993.18 However, thenew instrument was far from the initially envisioned diagnosticstool; essentially a no-frills version of BIAcore requiring more manual

14 Interview with a Biacore Product Development Manager.15 Interview with senior scientist at Astra Zeneca Gothenburg.16 Ibid.17 Interview with senior scientist at Leeds University.18 Biosensors — A Rapidly Expanding Sector, Kemisk Tidskrift, Vol. 106, No. 7, 55–56,16th June 1994.

789D. Harrison, H. Kjellberg / Industrial Marketing Management 39 (2010) 784–792

operations.19 Although considerably cheaper than BIAcore, the pricetag for BIAlite was still €10,000. Late in 1994, a third instrument wasintroduced, the BIAcore 2000, which was presented as the newflagship instrument, offering significant improvements in accuracy,sensitivity and stability over the existing two versions.20

During these years, instrument sales did grow steadily as did thevolume of users. Still, in June 1994 as outside observers noted, “majorcommercial success thus far has eluded biosensors.”21 Despite this,the growing number of users meant that the close interaction withcustomers was becoming increasingly time-consuming and costly forBiacore. From spring 1994 to early 1995, an internal re-organisation ofthe technical support and sales functions took place. This wasconsidered to be a very natural step to take: “the way we had beenworking had not changed very much”.22

3.5. 1995–1999: Becoming a formally independent company

In 1995 a new general manager was appointed and a rationalisa-tion project was initiated. There was a perceived need to improve the‘market orientation’ of the organisation, not least because Biacore wasto be launched on the stock market in 1996. In particular, as sales ofthe instrument increased — in September 1995, more than 400 werein use around the world 23 — it was considered necessary to make theinteraction between the company and new customers more cost-efficient. A trade off was recognised between the level of interactionwith individual customers and the cost of generating an experiencebase via interaction in multiple dyads. “Biacore was not new, volumeswere increasing, and non-users knew how they worked in principle…It is an established technique and it is positioned in a research contextin terms of what it can and cannot do”.24

The potential customers were considered to be both many enoughand informed enough to change the marketing approach. 400instruments were in place with users that had been identified throughinteraction. The connections with multiple early users had developeda series of application areas based on actual use. These can be broadlyclassified under two main headings: defining antibody characteristicsand defining the characteristics of protein interactions. The formerincluded monoclonal antibody development and the identification oftherapeutic targets. In terms of the latter, disease mechanisms, drugtarget analysis, thermodynamic measurements and proteomics wereimportant. The database, journal and BIAsymposia assisted thecompany in grouping individual customer experiments over time.

Thus, in late 1996 Biacore was able to give the following char-acterisation of the customers of its various product versions: BIAcore2000 customers were “large research labs and pharmaceutical com-panies with very high requirements concerning sensitivity, quality ofdata and productivity”; BIAcore 1000 customers were those researchlabs and pharmaceutical companies that did not have the same highrequirements; finally, BIAcore X customers were “shared laboratoryenvironments where the technology is used together with other es-tablished laboratory tools such as spectroscope, fluorescent and sep-aration methods”.25

Overall, it was the users from pre-1995 that engaged in developingnew applications, and in developing the limits for how to use theinstrument. Many of these users perceive they have a long-termrelationship with Biacore and describe themselves as belonging to aBiacore user community. They continued to develop new applications,

19 Making the most of the opportunities, BIA Journal, vol.1:1, p.10–11.20 BIA gets set for the next century, BIA Journal, vol. 2:1, p.3–4.21 Biosensors Seek Future Markets, Industries in Transition, Business Comm. Co., 22:2,Jun 1994.22 Interview with Product Manager for Consumables, Biacore.23 Moving into new spheres, BIA Journal, vol. 2:2, p.10–13.24 Interview with Product Development Manager, Biacore.25 Inbjudan till förvärv av aktier i Biacore International AB (Prospect), Pharmacia &Upjohn, 8 November 1996, p.28.

the results of which were reported to Biacore and distributed viaapplication newsletters.

“The new market organisation was based on the assumption thatother users would use a BIAcore in a similar way…rather than anendless undercover sales function through extensive sampling…weshifted the new users to look at publications…” (earlier work by theauthors). In other words, potential customers would follow theapplications paths developed by the earlier users. The way of relatingto the post-1995 user in general was to provide support in terms ofhow the instrument worked, not how it connected with a specificresearch experiment. The pre-sales support process now involved anApplication Specialist demonstrating a biosensor either at Biacore orat the customer's site. There were a number of self-training products,such as CDs and books, alongside a series of Biacore-run trainingcourses such as ‘kinetics one’. On-line support for users included web-based tutorials, interactive CDs, the reference database and theBIAjournal. In addition, new users were able to get help from built-insupport or modifications to the product as product line extensions(see Section 3.2) were brought out.

4. Case analysis: segmenting a market in the making

Our literature review showed that the established approach toindustrial market segmentation is based on the assumption that themarket, customers and product uses are known. Under such circum-stances, segmentation proceeds through descriptive techniques thatidentify relatively homogenous groups that can be approached throughstandardised marketing programs. The Biacore case presented aboveclearly highlights a different logic. Overall, it describes a long-term,iterative process during which product characteristics and uses,customer identities and interaction patterns gradually evolve. Thepurpose of this analysis section is to systematically explore how the casecan add to our understanding of segmentation. In so doing we willaddress both the character ofmarket segmentationwhen performed ona market in the making and how a firm's segmentation activities mayaffect the shape that the market assumes.

The Biacore project essentially started with a clear idea about twopotential customer segments, each accompanied by an ideal customerpersona: the medical clinic/practicing doctor and the research lab/practicing researcher. One of these ideal customers—medical clinics—was selected as the most relevant to target, due both to the con-siderable size of the segment and Pharmacia Diagnostics' establishedmarketing channels. With clinics as the chosen targets, the newinstrument was envisioned as a diagnostic tool that would comple-ment existing equipment produced for the diagnostics market.Although we learn relatively little about the process of identifyingthese market segments, the fact that these medical clinics alreadyconstituted a major market for Pharmacia Diagnostics suggest thattheywere not something the project group simply dreamt up. Initially,then, the project could be said to follow a more traditional productdevelopment approach, aiming to provide a complementary piece ofkit to an existing product range targeting a specific market segment.

But the initially envisioned target segment became increasinglydifficult to retain as the project went on. For several reasons, the initialpromoter of the practicing-doctor segment, Pharmacia Diagnostics,withdrew their support of the project. This failure to retain the initialtargetmarket seems to have beendecisive in throwing the project off itsinitial segmenting-targeting-positioning like course. The subsequentreassessment of the potential market for the instrument marks thetransition to very different type of segmentation logic.While the secondof the identified segments— practicing researchers—was considerablysmaller, it was still held to be “well definable and homogenous”.However, as this segment came into focus in the project, it soon becameclear that it was neither of these things.

The relation between the researchers and the new instrument hadnot been a central issue during the development of the diagnostics

790 D. Harrison, H. Kjellberg / Industrial Marketing Management 39 (2010) 784–792

application. Hence, two fundamental questions now had to berevisited: what the instrument was — or would become — and whothe users were. It soon became clear that these key characteristicsdepended on how the instrument could be linked both to othertechnologies in lab settings and to specific research questions thatinterested potential users within the new primary target segment.With no clear relation in place between the potential users and thenew tool, the new target segment was falling apart. The success of theproject now depended on establishing working relationships withpracticing researchers in order to establish what the instrument coulddo for them, or how they could use a BIAcore instrument.

Since Biacore was leading the development of biosensors for thelab/practicing researcher segment, they could not rely on assistancefrom competitors or market co-creators (cf. Johanson & Mattsson1985). In other words, there were no rival suppliers with whom toshare the market shaping/making investment costs and efforts withthe transition to another type of segmentation logic.

The result was an expensive, yet necessary, interaction and cus-tomisation of activities in collaborationswith individual users. Althoughthese interactions did generate knowledge about users and uses, theywere primarily constructive rather than descriptive in character. Theycontributed to put things in place rather than depict. The interactionstransformed the instrument from being a product whose uses andusefulness were largely unknown, to gradually acquiring an identity as‘an instrument for’ certain applications. The features of the instrumentwere also changed in response to both ongoing product developmentin-house and early user feedback, e.g. newauxiliary software for specificapplications. The interactions also configured the users. From the testsand experiments on the instrument conducted by the researcherstogether with Biacore's technical staff, emerged a series of individualand unique users, or ‘segments of one’ (e.g. Shapiro & Bonoma 1984;Turnbull & Valla, 1986; Freytag & Clarke 2001). Finally, Biacore itselfgained identity from these interactions. Until the research collabora-tions started, Biacore had been defined largely through its internalrelations within Pharmacia, remaining a technology-focused project.Now, it was transforming into an actor involved in solving challengingcustomer problems.

As noted above, the interactions also generated knowledge forBiacore. First of all, the ‘segments of one’ that emerged replaced theprevious idea of a homogeneous segment of customers in a generalresearch application. Second, the intensive interactions between theinstrument, Biacore staff and prospective customers enabled theelaboration of a number of dimensions for characterising the ins-trument, including specific applications. Biacore also systematicallycollected and disseminated these experiences for reputation and trustbuilding, e.g. through the BIA symposia, BIA journal, and BIA databaseetc. This knowledge proved important when Biacore once again at-tempted to take a more aggregated view of the market.

Can these efforts, which followed the initial segment failure, be seenas part ofmarket segmentation? Ifwe limit segmentation to the processof identifying relatively homogenous customer groups within a definedmarket, then the efforts clearly fall outside of segmentation proper.Indeed, a major reason for undertaking them seems to have been thelack of one of the basic conditions for market segmentation, i.e. theexistence of a defined or at least definable market. So far, then, ouranalysis suggests that for markets in themaking there may be a need toengage in certain preliminary activities before market segmentationproper can be meaningfully pursued. Even if we employ the concept ofintuition (Palmer & Millier, 2004), which fits well with how Biacoregradually gained knowledge from experience, we are left with theobservation that their activities also did something to the biosensormarket in the making.

As a result of Biacore's interactive approach during the post-launchyears, sales grew along with the number of BIAcore users. But as thevolume of users increased, it became evident that it was expensive tomaintain the contact patterns involved in this customized approach

(the central problem of “segments of one” strategies). These problemswere accentuated by growing financial pressures placed on Biacore inconnection to the planned stock market launch. Once again, an effortwas made to group customers and users into categories.

This time, the situation was different. First of all, bio-molecularinteraction analysis using a BIAcore instrument had become an estab-lished technique for a number of research applications, e.g. antibodycharacterisation and protein interactions, as evidenced by the growingnumber of scientific publications referencing the instrument. Second,and related, there were a growing number of actual BIAcore custom-ers that had bought one or more instruments and were using them intheir labs. Third, a range of BIAcore instrumentswas now available andthere was a pattern emerging as to the character of the customersbuying each of them. Fourth, Biacore was generating ideas about newapplication areas from user-experiences, indicating that completelynew market segments might be accessible in the future. Fifth andfinally, Biacore had first-hand experience of all these circumstancesthat it could utilise in an attempt to standardise its interaction withnew customers.

The intensive interactions between Biacore employees and earlyusers appear to have been critical to this attempt. Based on particularapplications that had been developed, Biacore identified a number of‘ideal customers’ for whom their offers could be somewhat standar-dised. Clearly, this was an attempt to move from “segments of one” to“segments of many” by bundling together some of the numerousapplications that had been collected in the database. This consciouseffort to aggregate was largely triggered by the new commercialperspective that was being espoused by the management of Biacore.

To address these potential “segments ofmany” a substantial part ofthe previous interactive contact pattern was packaged into saleableproducts or auxiliary objects, e.g. CDs and training courses, for a moreeconomic sales support process. Here, Biacore's early efforts to collectand disseminate experiences among users were important since theyallowed Biacore to rely on mediated interactions with customers viaresearch notes and other documents. This growing body of publica-tions testifying to the functionality of the instrument also allowed a‘shift of blame’. In the early post-launch years, Biacore had clearlyassumed accountability for non-results, i.e. for failures to successfullyuse the instrument for particular analyses. Such failures were linkedto the characteristics of the instrument, illustrated for instance by thedevelopment of new software. By amassing publications, producingmanuals, analysis software and instruction CD's, offering educationetc., Biacore increased the ability of new, first-time users to assumeaccountability for non-results. A post-1995 new user of a BIAcoreneeded to adapt to aggregations based on the results of Biacore'sinteractions with early users and was also assisted in doing so. Newusers were fitted into “a few segments of many”.

Although Biacore clearly sought to aggregate customers into groups,the specific approach employed did not follow the standard firm-centred ‘description-to-marketing-program’-logic. Nor did it exactlymatch the more iterative, interactive approach suggested by industrialmarketing researchers (e.g. Freytag & Clarke 2001, Håkansson & Ford2002). First, the attempted aggregation of customers was not based onan unobtrusive and distal description of the market, but on the actualresults of the collaborative efforts of Biacore and others to create thatmarket. Second, while the aggregation thus was based on previousinteractions with customers, it constituted a clear disruption of thislogic by being directed at other customers with whom Biacore neitherhad nor sought to develop an interactive history. Indeed, one importanteffect of the new standardised market approach was that it providednew users with a history that they did not actually have.

The more aggregated, segment-like treatment of customers at-tempted by Biacore did not replace all the previous interactive activ-ities. The series of application segments decided on for post-1995 userswere placed alongside ongoing relationships with the early users. Inother words, Biacore continued to relate to the early users in the same

791D. Harrison, H. Kjellberg / Industrial Marketing Management 39 (2010) 784–792

intensive, interactive way, while standardising how new users werehandled. Through this parallel, relational approach Biacorewas able tocontinue to learn and form visions for new application areas.

From a segmentation perspective, Biacore's maintenance ofintensive relationships with early users appears to have worked as aresource for ongoing re-segmentation. That is, the relationshipsprovided continuous updates on customer needs that could beemployed in adjusting the BIAcore offers. There is little evidence ofany static period following the efforts to standardise interactions,even if the firm did make choices and sought to stabilise a limitednumber of segments. In contrast to the literature, then, the casesuggests that re-segmentation may proceed as a mixture of activitiesaiming to describe the market and ongoing interactive, constructiveactivities with customers. Once again, our analysis suggests thatindustrial market segmentation is neither one-way, nor exclusivelydescriptive. In addition to refining descriptive techniques detachedfrom on-going marketing activities, the firm can also perform re-segmentation activities in collaboration with customers. The outcomeis that the market is always in the making as a result of being shapedby segmentation and re-segmentation activities.

5. Discussion and conclusions

Our analysis suggests that segmenting a market in themakingmaycall for other activities than the ones proposed in the segmentationliterature. Ourmajor conclusion is that the conceptualization of indus-trial market segmentation needs to be expanded to account for cons-tructive aspects in parallel to descriptive ones. Such an expansion willalso improve the applicability of segmentation in situations wheremarkets are undergoing significant changes.

The analysis showed that the initial activities in the Biacore projectfit relatively well with a firm-centred, segmentation-like approach tothe working-up of a market, including a relatively clear image of thecustomers and their needs. When this approach failed, the conditionsfor the project changed dramatically as did the activities undertakenby Biacore. Although the reasons for this kind of failure constitute aninteresting research topic by themselves, they are not our primaryconcern. Generally speaking, both proactive firm-initiated strategicmeasures to alter markets and more reactive responses to extra-organisational developments seem plausible. The subsequent activi-ties followed a pattern very different from classic market segmenta-tion. Still, therewas an ostensive similarity to segmentation in that theactivities indeed produced aggregated groups of customers sharingcertain important characteristics. Rather than dismissing theseactivities as not being about segmentation, we want to discuss howthey indeed could be viewed as such.

5.1. Identifying relatively homogenous groups of customers

In the segmentation literature, the managerial objective is toidentify sets of relatively homogenous customer groups within adefined market. Based on the assumption that there are groups of(actual and potential) customers of a product or service out there withspecific characteristics, this part of the segmentation process canproceed via descriptive techniques. Our analysis of a case where thisassumption did not hold highlights an alternative way of producingsuch groups. Rather than unobtrusively probing the practicingresearchers to identify their character traits and what they thoughtabout affinity biosensors, Biacore engaged in a process aiming to puttogether viable users and uses of their offer. In this process, the offeritself also evolved as a result of the experiences gained throughinteractions between Biacore staff, BIAcore machines, practicing re-searchers and specific research questions. In our analysis, then, cus-tomer identification wasmore about constructing identity, than aboutdescribing it.

5.2. Stabilising the relation between an envisioned market map and theenacted market

Successful segmentation attempts require firms to stabilise therelation between the image they have generated of themarket and theactual market. Working with a descriptive approach to segmentationthe question of how well the generated segments fit the market atlarge is one of generalisability. It is first and foremost an epistemo-logical issue: how can we say something that is valid for the entiremarket based on information drawn from only part of it?

Once again, our analysis highlights a different pattern in Biacore'sattempt to handle new customers in amore transaction-orientedway.Drawing on the experiences gained from their interactions with earlyusers, Biacore constructed their own user segments (and therebymapped the market) by aggregating various applications, researchissues, etc. Rather than being concerned about the descriptive validityof these aggregations, Biacore engaged in a process aiming to producestandardised users that fit them by combining documentation, edu-cation, systemdemonstrations, etc. In our analysis, then, stabilising therelation between the envisioned segments and the market was moreabout adjusting theworld to themap than about improving the qualityof the map.

As a result, our major conclusion is that the established concep-tualization of industrialmarket segmentation needs to be revised to beable to handle markets in the making. Extant work on segmentationhas attended primarily to its descriptive dimension, although someauthors have pointed out that segmentation may have consequencesfor the market being segmented (cf. Freytag & Clarke, 2001). Based onour case analysis we suggest that segmentation is a process involvingboth description and construction. Depending on the market, therelative importance of these two dimensions will vary. Thus, onestablishedmarkets, where offers, uses, modes of exchange, and buyerand seller identities are well developed, descriptive practices maysuffice to identify and make generalisations about segments. To theextent that one ormore of these facets of themarket are still unsettled,or once again are being called into question, the constructive dimen-sion of segmentation is likely to become increasingly important. Byprimarily studying established markets, the segmentation literaturehas systematically downplayed the constructive dimension of marketsegmentation. This, we believe, has helped to reinforce the strongassumptions typically associated with segmentation, i.e. that marketboundaries, applications, suppliers, and users are both established andfixed.

A key facet of the constructive dimension of market segmentationthat emerges fromour analysis is interaction. The identity constructionprocess triggered by Biacore's failed initial segmentation revolvedaround an interactive development of users' needs and productcharacteristics. To us, this is an alternative way of generating marketsegmentation criteria or dimensions. The interactions between thefirmand the early users both shaped themarket andprovided the basisfor the subsequent attempt to aggregate new users into segments.Similarly, this attempt itself relied on interaction to shape the newusers to fit the mould being provided for them. Moreover, the newtransaction-like market approach continued to evolve based on theexperiences gained from still on-going relations with early users.Overall, our analysis supports the idea that transaction-based marketsmay rely on both an interaction-based history and scaffolding allowingthem to develop further (cf. Johanson&Mattsson, 1985; Araujo, 2007).

The proposed re-conceptualization of industrial market segmen-tation has important implications for practicingmanagers in the sensethat it may improve the normative quality of the model (cf. Plank,1985). Situations where development projects imply a significant re-organisation of some specific market are likely to be relatively com-mon. At the same time, normative recommendations based on a viewof segmentation as an exclusively descriptive practice are likely toleave out many activities related to the shaping of that market. Under

792 D. Harrison, H. Kjellberg / Industrial Marketing Management 39 (2010) 784–792

fortunate circumstances the firmwill experience the familiar problemof generating useful market segments documented in the industrialmarket segmentation literature. Under less fortunate circumstances,potentially viable projects may either be terminated prematurely dueto ‘absence of proof’, or severely hampered by a circumscribed set ofaction alternatives.

5.3. Suggestions for further research

In closing this articlewewould like to indicate four important areasfor further research triggered by this study. First, given the limitedpossibility of making claims about populations of firms or marketsbased on the current study, there is a need to further investigate theprevalence and import of the constructive element in industrialmarket segmentation in other product and market contexts. Second,given that the study only has provided a few examples of what theconstructive dimension ofmarket segmentationmight entail, there is aneed for a more systematic exploration of the major classes of actionsinvolved in constructive market segmentation. Third, given that wehave studied a market that might be considered idiosyncratic; there isa need to further explore under what market conditions the cons-tructive dimension of market segmentation becomes pronounced.Finally, since there is no principle bar against a constructive dimensionbeing important also for segmenting consumer markets, it would bevery interesting to explore towhat extent and how itmanifests itself ina segmentation process related to a consumermarket “in themaking”.

Acknowledgements

We would like to thank three anonymous reviewers for their manyinsightful and encouraging comments on previous versions of thisarticle. We also acknowledge the input of Prof AlexandraWaluszewski,Uppsala University and Prof Thomas Ritter, Copenhagen BusinessSchool. Partial financial support was provided by the Bank of SwedenTercentenary Foundation, and the research project P2005-0517:1“Market-making: shaping markets and market actors in practice”.

References

Abelin, P. (1997-02-03). Biacore, EnskildaResearch, Equities, EnskildaSecurities, Stockholm,12.

Abbott, A. (1992). What do cases do? Some notes on activity in sociological analysis. InCharles C. Ragin & H. S. Becker (Eds.), What is a case?: Exploring the foundations ofsocial inquiry (pp. 53−82). UK: University of Cambridge.

Andersson, P. (1996). The emergence and change of Pharmacia Biotech 1959–1995: thepower of the slow flow and the drama of great events, PhD diss., Stockholm,Stockholm School of Economics.

Araujo, L. (2007). Markets, market-making andmarketing.Marketing Theory, 7, 211−226.Balakrishna, G. (1978). Better use of the industrialmarketing concept. IndustrialMarketing

Management, 7, 71−76.Bonoma, T. V., & Shapiro, B. P. (1983). Segmenting the industrial market. Lexington,

Mass.: Lexington Books.Choffray, J. -M., & Lilien, G. L. (1980). Industrial market segmentation by the structure of

the purchasing process. Industrial Marketing Management, 9, 331−342.Cunningham,M. T., &White, J. G. (1973). Thedeterminants of choice of supplier:A studyof

purchase behaviour for capital goods. European Journal of Marketing, 7, 189−202.Dibb, S. (1998). Market segmentation: Strategies for success.Marketing Intelligence and

Planning, 16(7), 394−406.Dibb, S., & Simkin, L. (1994). Implementation problems in industrialmarket segmentation.

Industrial Marketing Management, 23, 55−63.Dibb, S., Stern, P., & Wensley, R. (2002). Marketing knowledge and the value of segmen-

tation.Marketing Intelligence and Planning, 20(2), 113−119.Dubois, A., & Gadde, L. -E. (2002). Systematic combining: An abductive approach to case

research. Journal of Business Research, 55, 553−560.Dubois, A., & Araujo, L. (2007). Case research in purchasing and supply management:

Opportunities and challenges. Journal of Purchasing and Supply Management, 13(3),170−181.

Easton, G. (1995). Methodology in industrial networks. In Kris Moller & David T. Wilson(Eds.), Business Marketing: an Interaction and Network Perspective (pp. 411−492).Norwell, Mass: Kluwer.

Easton, G. (1998). Case research as a methodology for industrial networks: A realistapologia. In P. Naude & P. Turnbull (Eds.), Network dynamics in internationalmarketing (pp. 73−87). Oxford: Pergamon.

Eisenhardt, K. M. (1989). Building theories from case study research. Academy ofManagement Review, 14(4), 532−550.

Firat, A. F., & Shultz, C. J., II (1997). From segmentation to fragmentation: Markets andmarketing strategy in the postmodern era. European Journal ofMarketing, 31, 183−207.

Freytag, P., & Clarke, A. (2001). Business to business market segmentation. IndustrialMarketing Management, 30(3), 473−486.

Griffith, R., & Pol, L. (1994). Segmenting industrial markets. Industrial MarketingManagement, 23(1), 39−46.

Håkansson, H., & Ford, D. (2002). How should companies interact in businessnetworks? Journal of Business Research, 55, 133−139.

Halinen, A., & Törnroos, J. -A. (2005). Using case methods in the study of contemporarybusiness networks. Journal of Business Research, 58, 1285−1297.

Harrison, D., & Waluszewski, A. (2008). The development of a user network as a way tore-launch an unwanted product. Research Policy, 37(1), 115−130.

Huber, G. P., & Ven de Ven, A. H. (Eds.). (1995). Longitudinal field research methods —Studying processes of organisational change Thousand Oaks, CA: SAGE.

Jaworski, B., Kohli, A. K., & Sahay, A. (2000). Market-driven versus driving markets.Journal of the Academy of Marketing Science, 28, 45−54.

Johanson, J., & Mattsson, L. -G. (1985). Marketing investments andmarket investments inindustrial networks. International Journal of Research in Marketing, 3(2), 185−195.

Kjellberg, H., & Helgesson, C. -F. (2007). The mode of exchange and shaping of markets:Distributor influence on the Swedish post-war food industry. Industrial MarketingManagement, 36, 861−878.

Langley, A. (2007). Process thinking in strategic organisation. Strategic Organisation, 5(3),271−282.

Millier, P. (2000). Intuition can help in segmenting industrialmarkets. IndustrialMarketingManagement, 29, 147−155.

Mintzberg, H., & Waters, J. A. (1985). Of strategies, deliberate and emergent. StrategicManagement Journal, 6, 257−272.

Mitchell, J. C. (1983). Case and situation analysis. Sociological Review, 31, 187−211.Moriarty, R. T., & Reibstein, D. J. (1986). Benefit segmentation in industrial markets.

Journal of Business Research, 14, 463−486.Palmer, R. A., & Millier, P. (2004). Segmentation: Identification, intuition, and imple-

mentation. Industrial Marketing Management, 33, 779−785.Plank, R. E. (1985). A critical review of industrial market segmentation. Industrial

Marketing Management, 14, 79−91.Quinn, L., Hines, T., & Bennison, D. (2007). Making sense of market segmentation: A

fashion retailing case. European Journal of Marketing, 41, 439−465.Shapiro, B., & Bonoma, T. (1984). How to segment industrial markets. Harvard Business

Review, 62(3), 104−110.Sheth, J. N., Sisodia, R. S., & Sharma, A. (2000). The antecedents and consequences of

customer-centric marketing. Journal of the Academy of Marketing Science, 28, 55−66.Smith, W. R. (1956). Product differentiation and market segmentation as alternative

marketing strategies. Journal of Marketing, 21, 3−8.Turnbull, P. W., & Valla, J-P. (Eds.). (1986). Strategies for International Industrial

Marketing: The Management of Customer Relationships in European IndustrialMarkets. Croom Helm, London.

Verhallen, T. M. M., Frambach, R. T., & Prabhu, J. (1998). Strategy-based segmentation ofindustrial markets. Industrial Marketing Management, 27, 305−313.

Webster, F. E. J., & Wind, Y. (1972). A general model for understanding organizationalbuying behavior. Journal of Marketing, 36, 12−19.

Webster, F. E. J. (1984). Industrial marketing strategy, 2nd edNewYork: JohnWiley & Sons.Wilson, R. D. (1986). Segmentation and communication in the industrial market place.

Journal of Business Research, 14, 487−500.Wind, Y., & Cardozo, R. N. (1974). Industrial market segmentation. Industrial Marketing

Management, 3, 153−165.

Debbie Harrison is an Associate Professor in the department of Strategy and Logisticsat the Norwegian School of Management BI, Oslo, Norway. Her research interests are inthe area of inter-organisational relationships. She has published papers in the areas ofnetwork strategizing, the role of contracts in business relationships, user networks andresource development in journals such as Journal of Management Studies, IndustrialMarketing Management and Research Policy.

Hans Kjellberg is an Associate Professor at the Department of Marketing and Strategyat the Stockholm School of Economics. His research interests concern economicorganising in general and the shaping of markets and economic exchanges inparticular. He is currently heading a project studying mundane market practices andhow marketing theories and tools participate in shaping markets. Previous publica-tions include articles in Industrial Marketing Management, Marketing Theory and Con-sumption, Markets and Culture.