SOA adoption in business networks: do service-oriented...

16
SPECIAL THEME SOA adoption in business networks: do service-oriented architectures really advance inter-organizational integration? Jan Löhe & Christine Legner Received: 30 May 2010 / Accepted: 27 October 2010 / Published online: 23 November 2010 # Institute of Information Management, University of St. Gallen 2010 Abstract The emergence of the Internet as a global communication infrastructure has dramatically reduced interaction costs within and across organizations, with significant impact on inter-organizational relationships, vertical industry structures, and markets. More recently, service-oriented architectures (SOA) and Web services have introduced the next paradigm shift and foster the idea of dynamic business networks with quick connect and disconnect relationships. However, little research has systematically analyzed how companies leverage SOA to improve their inter-organizational relationships and reshape their business networks. In addition, the mature research stream on inter-organizational information systems (IOS) has not yet sufficiently considered SOA. In order to close this gap, our research seeks to improve the fundamental understanding of how SOA is applied in business networks and how it differs from prior forms of IOS. Using an exploratory research approach, we investigate 33 SOA cases to identify focus areas and patterns of SOA adoption in business networks. Our case analysis builds on a multi- dimensional classification scheme which we derived from prior literature. While our empirical findings do not confirm all promising propositions related to SOA, they underline the specific contribution of SOA compared to prior forms of IOS. We conclude by suggesting five clusters of SOA adoption in the inter-organizational domain, each of those introducing new aspects in the coordination of distributed business networks. Keywords Service-oriented architecture (SOA) . Web services . Business networking . Inter-organizational systems (IOS) . Integration . Business-to-Business (B2B) JEL Classification L86: Information and Internet Services; Computer Software Introduction Over the past decades, specialization and globalization have lead organizations to engage in new forms of value creation and to rethink their organizational boundaries (Jacobides and Billinger 2006; Österle et al. 2001). The emergence of the Internet as a widely-accepted, global communication infra- structure has dramatically reduced interaction costs within and across organizations, which is said to foster the division of labor between organizations (Christiaanse et al. 2004; Hagel and Brown 2001). Since the early 2000s, service- oriented architectures (SOA) and Web services have intro- duced the next paradigm shift by establishing service-based communication across heterogeneous platforms and among enterprises (Alonso et al. 2003; Erl 2005; Papazoglou and van den Heuvel 2007). As SOA increases internal and external integration capabilities, it is expected to stimulate inter-organizational process integration (Daniel and White 2005) and foster dynamic business networks with quick connect and disconnect relationships (Merrifield et al. 2008). SOA and Web services, however, are still mostly used within company boundaries. Until now, the success of SOA and Web service implementations applied within an Responsible editor: Rainer Alt J. Löhe : C. Legner (*) Institute of Research on Information Systems (IRIS), EBS Business School, Söhnleinstr. 8D, 65201 Wiesbaden, Germany e-mail: [email protected] Electron Markets (2010) 20:181196 DOI 10.1007/s12525-010-0046-7

Transcript of SOA adoption in business networks: do service-oriented...

Page 1: SOA adoption in business networks: do service-oriented ...aws.iwi.uni-leipzig.de/em/fileadmin/user_upload/doc/Issues/Volume_20/Issue_03-04/V20I3...understanding of how SOA is applied

SPECIAL THEME

SOA adoption in business networks: do service-orientedarchitectures really advance inter-organizationalintegration?

Jan Löhe & Christine Legner

Received: 30 May 2010 /Accepted: 27 October 2010 /Published online: 23 November 2010# Institute of Information Management, University of St. Gallen 2010

Abstract The emergence of the Internet as a globalcommunication infrastructure has dramatically reducedinteraction costs within and across organizations, withsignificant impact on inter-organizational relationships,vertical industry structures, and markets. More recently,service-oriented architectures (SOA) and Web services haveintroduced the next paradigm shift and foster the idea ofdynamic business networks with quick connect anddisconnect relationships. However, little research hassystematically analyzed how companies leverage SOA toimprove their inter-organizational relationships and reshapetheir business networks. In addition, the mature researchstream on inter-organizational information systems (IOS)has not yet sufficiently considered SOA. In order to closethis gap, our research seeks to improve the fundamentalunderstanding of how SOA is applied in business networksand how it differs from prior forms of IOS. Using anexploratory research approach, we investigate 33 SOAcases to identify focus areas and patterns of SOA adoptionin business networks. Our case analysis builds on a multi-dimensional classification scheme which we derived fromprior literature. While our empirical findings do not confirmall promising propositions related to SOA, they underlinethe specific contribution of SOA compared to prior formsof IOS. We conclude by suggesting five clusters of SOA

adoption in the inter-organizational domain, each of thoseintroducing new aspects in the coordination of distributedbusiness networks.

Keywords Service-oriented architecture (SOA) .Webservices . Business networking . Inter-organizationalsystems (IOS) . Integration . Business-to-Business (B2B)

JEL Classification L86: Information and InternetServices; Computer Software

Introduction

Over the past decades, specialization and globalization havelead organizations to engage in new forms of value creationand to rethink their organizational boundaries (Jacobides andBillinger 2006; Österle et al. 2001). The emergence of theInternet as a widely-accepted, global communication infra-structure has dramatically reduced interaction costs withinand across organizations, which is said to foster the divisionof labor between organizations (Christiaanse et al. 2004;Hagel and Brown 2001). Since the early 2000s, service-oriented architectures (SOA) and Web services have intro-duced the next paradigm shift by establishing service-basedcommunication across heterogeneous platforms and amongenterprises (Alonso et al. 2003; Erl 2005; Papazoglou andvan den Heuvel 2007). As SOA increases internal andexternal integration capabilities, it is expected to stimulateinter-organizational process integration (Daniel and White2005) and foster dynamic business networks with quickconnect and disconnect relationships (Merrifield et al. 2008).

SOA and Web services, however, are still mostly usedwithin company boundaries. Until now, the success of SOAand Web service implementations applied within an

Responsible editor: Rainer Alt

J. Löhe : C. Legner (*)Institute of Research on Information Systems (IRIS),EBS Business School,Söhnleinstr. 8D,65201 Wiesbaden, Germanye-mail: [email protected]

Electron Markets (2010) 20:181–196DOI 10.1007/s12525-010-0046-7

Page 2: SOA adoption in business networks: do service-oriented ...aws.iwi.uni-leipzig.de/em/fileadmin/user_upload/doc/Issues/Volume_20/Issue_03-04/V20I3...understanding of how SOA is applied

organization have not been translated to SOAs acrossenterprises or industries (Huhns 2008). Correspondingly,little research has systematically analyzed how companiesapply SOA to improve coordination with their businesspartners. In addition, the mature stream of research on inter-organizational systems (IOS) has not yet sufficientlyaddressed SOA. Moreover, prior research did not developa classification scheme or taxonomy that allows foranalyzing SOA adoption in business networks or forcomparing it to prior forms of IOS. Accordingly, fromboth the academic and practitioner perspectives, it is notwell understood how companies should leverage theservice-oriented paradigm in their inter-organizationalrelationships. Our research seeks to bridge this gap as wellas to contribute to the emerging research discussion on“service science” (Buhl et al. 2008; Demirkan et al. 2008)by exploring the adoption of SOA in business networksbased on two research questions: (RQ1) what are the mainfocus areas of SOA implementations that target inter-organizational integration; (RQ2) which patterns can beidentified for SOA adoption in business networks, and howdo they differ from prior forms of IOS?

In order to improve our understanding of SOA’s role andimplementation in inter-organizational settings, we choose anexploratory research approach. For this purpose, we collectedand reviewed 33 case descriptions of SOA projects andconducted a content analysis. In order to assess SOA adoptionin the cases, we used a classification scheme which wederived from prior literature and analyzed three relevantdimensions: the characteristics and configuration of businessnetworks; the electronic collaboration processes between thedifferent organizations; and the technical architecture thatsupports inter-organizational integration. Our empirical find-ings inform researchers as well as practitioners of the SOAparadigm’s contribution and the implications for its furtherapplication in business networks. By anchoring SOA in theestablished research streams on IOS and business networking,our work seeks to provide a stronger foundation for futureresearch in this field.

The remainder of the paper is structured as follows:First, we analyze the current state of research and motivateour research objectives and process. Next, based onliterature work, we justify the need for a multi-dimensional framework for assessing SOA adoption inbusiness networks and introduce that framework, which wehave derived from prior IOS literature. Building on this, wepresent the empirical analysis of real-world SOA imple-mentations and discuss the results in the context of IOS andSOA propositions. We then summarize the findings of ourcase analysis and derive five clusters of SOA adoption. Thearticle closes by summarizing our contributions to researchand practice, the limitations, and an outlook for futureresearch.

Current state of research

Service orientation has gained much popularity since theearly 2000s, owing to the hype around Web servicetechnologies and software vendors’ subsequent announce-ments that they would redesign their product suites (Danieland White 2005; Hagel and Brown 2001). Recent literaturedescribes SOA as a component-based, distributed ISarchitecture paradigm that encapsulates elements of the ITinfrastructure as identifiable, self-described services basedon open standards (Demirkan et al. 2008; Erl 2005;Heutschi 2007; Papazoglou and van den Heuvel 2007).While SOA was initially mostly researched from theperspective of computer scientists, recent publications seekto address the manifold organizational and strategic aspectsthat are associated with SOA implementations in practice(Demirkan et al. 2008; Luthria and Rabhi 2009; Viering etal. 2009). The growing number of literature reviews (Buhlet al. 2008; Siedersleben 2007; Sinz 2008; Zhao and Cheng2005) and propositions for research agendas (Demirkan etal. 2008; Luthria and Rabhi 2009; Papazoglou and van denHeuvel 2007) aim at defining the service-oriented paradigmand initiating “service science” or “Service Science,Management and Engineering (SSME)” as a new, interdis-ciplinary research area (Buhl et al. 2008; Maglio et al.2006; Spohrer and Maglio 2008).

If applied in the inter-organizational context, SOAsprovide a more flexible and much cheaper platform forexternal integration than previous forms of IOS, notablyelectronic data interchange, which built on proprietarytechnologies (Bussler 2003; Christiaanse et al. 2004;Gosain et al. 2004; Linthicum 2001). Dynamically discov-erable business services are said to increase the flexibilityof inter-enterprise collaboration (Demirkan et al. 2008) andallow for quick connect and disconnect relationships(Merrifield et al. 2008). Consequently, a large number ofconceptual papers elaborate on the dynamic sourcingcapabilities of the emerging Web services framework andposit that they will be crucial to enabling loosely coupledbusiness networks (Iyer et al. 2003; Van Heck and Vervest2007). Another stream of research suggests Web service-based reference models for B2B collaboration (Chen et al.2007; Huang and Chung 2003; Li et al. 2006), whereasother authors apply SOA concepts to particular domains orindustries, resulting in platform designs or reference modelsfor distributed SOAs (Baida et al. 2008; Reitbauer et al.2008). In summary, most of the existing research on SOAin the inter-organizational context is design-oriented orconceptual in nature and a systematic analysis as towhether and how SOA concepts are applied in real-worldimplementations is lacking.

In this context, it is important to note that the maturestream of IOS research has not yet sufficiently addressed

182 J. Löhe, C. Legner

Page 3: SOA adoption in business networks: do service-oriented ...aws.iwi.uni-leipzig.de/em/fileadmin/user_upload/doc/Issues/Volume_20/Issue_03-04/V20I3...understanding of how SOA is applied

SOA and Web services. Until now, IOS research hasextensively explored the strategic and operational impactof external integration and analyzed IOS adoption, asunderpinned by the literature reviews by Robey et al.(2008) and Wareham et al. (2005). However, empiricalstudies focus either on electronic data interchange (EDI) oron e-business and have not yet been extended to includeSOA or Web services. While these empirical studiesexplain how IOS have been adopted in the past, Robey etal. (2008) claim that the research results generated from thefirst wave of IOS research are not particularly insightful forunderstanding the next generations of IOS based on openstandards. The same authors recommend opening the blackbox of IOS by identifying specific design features andsystem capabilities. However, besides the lacking empiricalanalysis of real-world SOA implementations, there is littleresearch synthesizing the commonalities and differencesbetween SOA the existing forms of IOS and their adoption.To our knowledge, no classification scheme or taxonomy isavailable that allows for systematically analyzing IOSadoption in business networks and is suitable for comparingSOA to prior forms of IOS. Since IT-based strategies mayreform the patterns of interdependence that exist in inter-organizational relationships, such a classification schemeneeds to consider not only the electronic integration linkages,but also its interplay with a business network’s strategy andstructure (Van Heck and Vervest 2007; Österle et al. 2001).

Research design

Research objectives and method

Given that the academic discussion about SOA in businessnetworks is at an early stage (Zhao et al. 2007), our primarygoals are to understand how companies adopt SOA as wellas to identify patterns in the way companies (re)shape theirexternal relationships by means of SOA. We are particular-ly interested in analyzing how these patterns differ frompatterns that we have seen with prior forms of IOS. Thisshould allow us to critically assess SOAs’ specificcontributions to business networks. We use an exploratoryresearch approach based on case studies because it is stilldifficult to gather enough high-quality and reliable data onreal-world SOA implementations to conduct quantitativestudies with large firm samples. This is due to two factors:First, SOA adoption follows the typical technology diffu-sion trajectory that starts with a relatively small communityof early adopters before reaching higher levels of diffusion.Second, the design and implementation of SOA implysignificant organizational changes and multi-year IT proj-ects before SOA can be used in a productive environment.Additionally, a case study approach is particularly useful in

cases where a contemporary phenomenon is studied in itsnatural context and where research and theory are still attheir early, formative stages (Benbasat et al. 1987; Gibbertet al. 2008). We chose a multiple-case research design sincethe knowledge gained from replicated case studies isusually considered to increase the external validity of thefindings (Dubé and Paré 2003; Yin 2003). Althoughmultiple case studies do not allow for statistical general-ization, they are valuable for analytical generalization, i.e.supporting the process from empirical observations totheory (Dubé and Paré 2003; Gibbert et al. 2008).

Research process

Given that there are only weak links between the SOA and IOSresearch streams, our four-step research process starts with thedevelopment of a multi-dimensional framework for the classifi-cation of our SOA cases. Such amulti-dimensional classificationscheme is also said to support case analysis by means of patternmatching (Yin 2003). In order to deal with the multiple aspectsin inter-organizational settings, we decided to build on theelectronic business networking methodology and architecture(Fleisch 2001; Österle et al. 2001; Senger 2004) and reviewedSOA and IOS literature to identify attributes that characterizeSOA adoption in business networks according to threedimensions: (D1) the characteristics and configuration ofbusiness networks; (D2) the electronic coordination betweenorganizations defined by the inter-organizational businessprocesses; and (D3) the technical architecture that supportsinter-organizational integration. Prior to coding, we discussedthe applicability of the classification framework to our researchand revised the defined coding scheme after a first round ofcoding by one of the authors (Fettke and Loos 2003).Following Yin (2003), who recommends evaluating casestudies by using matrices of categories and placing the evidencewithin such categories, we implemented this coding scheme bymeans of tables to support content analysis. Finally, we comeup with ten attributes and 32 values at the strategic layer (D1,Table 1), four attributes and 16 values at the process layer (D2,Table 2), and 11 attributes and 44 values at the IS/IT layer (D3,Table 3).

Data collection represents the second step in our researchprocess. Here, the key difficulty is the accessibility of case data,given that our main research objective is to explore real-worldSOA implementations. Like other IS researchers (Frick andSchubert 2009; Tafti et al. 2008), we rely on secondary data toget a comprehensive empirical basis. Our data originate froma case study database with 194 descriptions of SOAimplementations acquired from journal databases, printpublications, case collections, press coverage, and materialissued by companies, vendors, consultants, and analystsdealing with SOA. Cases were selected based on theirsubstantive significance (Dubé and Paré 2003) by applying

SOA adoption in business networks 183

Page 4: SOA adoption in business networks: do service-oriented ...aws.iwi.uni-leipzig.de/em/fileadmin/user_upload/doc/Issues/Volume_20/Issue_03-04/V20I3...understanding of how SOA is applied

the following selection criteria: (1) Description of a real-worldSOA implementation; (2) SOA in an inter-organizationalrelationship involving at least two independent organizations;(3) Availability of sufficient information about the SOAimplementation and context for filling the multi-dimensionalframework. 71 case descriptions satisfy the first two criteria.During our coding, we had to eliminate another 38 cases dueto less information according to the third criterion. The finaldata set covers 33 case descriptions from different industries(see Table 6 in the Appendix).

As a third step, we use qualitative content analysis (Mayring2000) and coded the case descriptions according to our multi-dimensional framework. Each case was first analyzed by oneof the researchers, who coded the cases’ arguments accordingto the attribute definitions. Whenever we found several valid,non-exclusive attribute values in the case material, multiplecounting was used. In order to increase the inter-raterreliability, a second researcher analyzed each case. If therewas agreement on the coding, the classification was accepted.The remaining classifications were subject to further discus-sions by the authors. If agreement could not be reached, caseswere excluded from the data set. Descriptive statisticsmethods supported the identification of patterns in ourempirical data and helped us gain insights related to the mostprominent focus areas of SOA adoption according to the pre-defined dimensions and attributes. Although the number ofcases does not allow for statistical generalization, our data setallowed for analytical generalization, i.e. supporting theprocess from empirical observations to theory (Dubé and Paré2003; Gibbert et al. 2008). In addition, through patternmatching (Dubé and Paré 2003; Gibbert et al. 2008; Yin2003), we were able to compare the empirically observedpatterns with those predicted by SOA literature and thoseidentified by prior IOS literature. This would also allow us toassess whether empirical findings are consistent with propo-sitions on SOA and prior forms of IOS.

Lastly, to identify clusters of SOA adoption, weperformed two rounds of agglomerative, hierarchicalclustering procedures on binary data. In a first step, weidentified outliers that were eliminated from the data set(four cases) with single-linkage method. To identify thenumber of homogenous clusters, we applied the Wardmethod and considered large breaks between the rankedcoefficients and the dendrogram. This resulted in fiveclusters characterizing different forms of SOA adoption inbusiness networks.

The need for a multi-dimensional assessment of SOAadoption in business networks

While SOA enthusiasts argue that SOA will significantlyincrease the dynamism in network formation, IOS literature

generally emphasizes that IT advances are an importantfactor in the formation of networked forms of organizing.Prior experiences with EDI demonstrate that IOS adoptionis inherently complex. On the one hand, patterns ofinterdependence exist between multiple economic actors(Kambil and Short 1994). IT-based strategies may reformthese patterns, resulting in a complete redesign of a givenbusiness network. On the other hand, sharing “more”information electronically with external parties is notnecessarily beneficial (Gosain et al. 2004; Premkumar2000). Consequently, the information processing capabili-ties of an IOS need to fit a number of contextual factors,which they summarize as information processing needs.This calls for a congruence of the IOS design and theinstitutional context. Hence, IOS adoption requires anunderstanding of the relationships between a businessnetwork’s structure and characteristics, the coordinationneeds between the different business partners as well as theunderlying infrastructure and systems (Österle et al. 2001).However, we are unaware of any comprehensive classifi-cation scheme that covers these interdependent aspects andis therefore suitable as a basis for comparing SOAimplementations to other, IOS implementations.

In order to fill this gap, we suggest a multi-dimensionalframework that allows for the description of SOA implemen-tations and their comparison to prior forms of IOS.We derivedthis framework based on the review of the businessnetworking concept (Fleisch 2001; Österle et al. 2001;Senger 2004) and classification schemes developed by priorliterature, notably by Klein (1996), Clarke (2001), Senger(2004), Al-Naeem et al. (2004), Löwer (2005), and Rodon(2006). The basic structure of our framework builds on theelectronic business networking methodology and architecture(Fleisch 2001; Österle et al. 2001; Senger 2004). Thisconceptual approach distinguishes three dimensions fordescribing the alignment of strategic and organizationalcapabilities with the IS/IT architecture, and has been widelyused to analyze the design of IT-supported business relation-ships: The strategic layer covers the characteristics andconfiguration of business networks. The process layer furtherrefines the inter-organizational coordination by means ofcooperation processes. The IS/IT layer depicts the technicalarchitecture that supports inter-organizational integration.The following three sections will present our multi-dimensional framework in detail and describe the attributeswhich we identified from prior literature to analyze IOS andSOA implementations in business networks.

Assessing the strategic layer

The strategic layer describes the business network as value-creating system and comprises the different actors (suppli-ers, business partners, allies, and customers) that collabo-

184 J. Löhe, C. Legner

Page 5: SOA adoption in business networks: do service-oriented ...aws.iwi.uni-leipzig.de/em/fileadmin/user_upload/doc/Issues/Volume_20/Issue_03-04/V20I3...understanding of how SOA is applied

rate to produce value and achieve superior customersatisfaction (Kambil and Short 1994). The main designelements are organizational entities that form the businessnetwork and their interrelationships (Österle et al. 2001).

In order to assess SOA adoption, we first consider attributesthat characterize the network structure and size: The number ofpartners (Kauffman and Weill 1989) distinguishes betweennetworks involving a handful of partners (<10), networksinvolving a larger group of partners (10<n<100) and large-scale networks (n>100). The heterogeneity of partnersdescribes whether they are in the same industry (homoge-neous) or in different ones (heterogeneous), whereas valuechain integration can be vertical (e.g. integration withsuppliers) as well as horizontal (e.g. cooperation withpartners) (Robertson and Langlois 1995). The type ofrelationship considers relationships between businesses, con-sumers, and public administrations. The latter have onlyrecently started to electronically collaborate with businessesand citizens with the objective to improve public services andincrease their efficiency (cf. Baida et al. 2008; Vogel et al.2008). The following combinations can be derived: business-to-business (B2B), business-to-consumer (B2C), government-to-government / government-to-business (G2G/B2G), andgovernment-to-citizen (G2C). Network topology differentiatesbetween 1:1, 1:n, and m:n relationships (Clarke 2001).Furthermore, we characterize the network by its stability(stable versus dynamic) and governance (focal versuspolycentric) (Alt et al. 2005; Fleisch 2001).

Besides network size and structure, the network purposeis a relevant characteristic for analyzing SOA adoption.Following Ansoff’s two-by-two matrix, it can be charac-terized either as efficiency network (existing product /existing market), channel extension network (existingproduct / new market), innovation network (new product /existing market), or venture network (new product / newmarket) (Kambil 2008). Secondly, for assessing SOAadoption, we consider the strategic factors that drivebusiness networking. Since business networks pursuespecific objectives, we analyze networking targets (Kambil2008) such as the sharing of core competencies throughstrategic partnerships, operational effectiveness, and strate-gic positioning through new products, new markets, or newdistribution channels. Furthermore, SOA adoption could bedriven by several strategic pressures, such as access tocustomers, improvement of the product & service offering,mergers & acquisitions as well as the development ofstrategic know-how (Senger 2004).

Assessing the process layer

The process layer concerns organizational integration(Bauer and Stickel 1998) and is essential to connect thestrategic layer with the IS layer. Business processes

comprise sequences of activities that may be distributedacross various organizational units and are considered thecore of IT-based value creation (Melville et al. 2004;Österle et al. 2001). Moreover, SOA literature claims thatSOA supports seamless process integration within andacross company boundaries (Bussler 2002; Dorn et al.2009; Papazoglou and van den Heuvel 2007).

Within our classification framework, the attribute coop-eration process characterizes the business processes thatlink two or more independent entities. Following Alt andÖsterle (2004) we distinguish between content & commu-nity, commerce, product life-cycle, supply chain, finance,and maintenance & repair cooperation processes. Anotherdifferentiator is the output of cooperation processes, whichcan be divided into physical product or virtual product,service & outtasking (Alt et al. 2005). Intimately connectedwith cooperation processes is the span of cooperationwhich either covers the entire business network (end-to-end) or focuses on the customer-facing or supplier-facingside. Finally, we examine a list of process-related pressuresthat drive business networking; for example, capacityutilization, coordination effort & error proneness, redun-dancies, as well as process complexity, process cost &throughput times (Senger 2004).

Assessing the IS layer

The IS layer describes how information is processed andshared electronically within and across organizations fromthe technical integration perspective (Bauer and Stickel1998).

Given our research objectives, we first characterize SOAadoption by means of scope, implementation strategy, andthe affected applications: SOA scope describes the focus ofSOA implementations, which may either span the entirebusiness network, an internal implementation with differentinternal and external partners, or be restricted to anorganization’s internal network. SOA implementation strat-egy distinguishes between heterogeneous, best-of-breed,single-vendor, and in-house development SOA approaches.Another analyzable characteristic is that SOA affectsapplications like ERP & other core, core banking &insurance, e-commerce & customer relationship manage-ment, content & document management, groupware &community management, management information systems& business intelligence, e-procurement, supply chainmanagement & product life-cycle management, humanresource, as well as finance (accounting) applications.

To assess how SOA is implemented, we use thefollowing attributes: The coupling intensity defines whethertechnical integration needs previous agreements (stable) orsupports on-the-fly (dynamic) coupling (Al-Naeem et al.2004). We also differentiate between a centralized (one

SOA adoption in business networks 185

Page 6: SOA adoption in business networks: do service-oriented ...aws.iwi.uni-leipzig.de/em/fileadmin/user_upload/doc/Issues/Volume_20/Issue_03-04/V20I3...understanding of how SOA is applied

instance) and distributed (multiple instances) couplingapproach (Al-Naeem et al. 2004). Furthermore, thecommunication type covers human-to-machine andmachine-to-machine interaction (McAfee 2005). Further-more, with SOA, different external information exchangestyles can be realized, such as a message-oriented (e.g.SOAP and WSDL), a file- & data-oriented, a remotemethod invocation (e.g. DCOM and RPC) and a pipe-and-filter (e.g. using queues) style (Al-Naeem et al. 2004).During our analysis, we also considered adding portaltechnology to the information exchange style attribute. Interms of the integration approach, we further distinguishwhether SOA involves business process layer, presentationlayer, or function & data layer integration (Al-Naeem et al.2004; Heutschi 2007). In this context, service granularity isa key design decision (Erl 2005): business process servicesencapsulate entire workflows, process activity & taskservices support single-process-step activities, and utility& entity services support generic infrastructure functional-ity (e.g. authentication) or operations on data. Finally,standardization is a precondition for any form of electroniccooperation (McAfee 2005): Firstly, standardization oftransport & data formats at the technical level, secondly,the meaning & payload of messages at the syntactic andsemantic level, and, thirdly, the process flows betweenparties at the pragmatic level. As before, certain IS-relatedpressures drive SOA adoption in business networks, suchas missing interoperability & different integration mecha-nisms, heterogeneous technologies & platforms, redundantfunctionality & master data, security issues, legacy &monolithic systems, development & operations & mainte-

nance costs as well as missing functionality & imperfectsoftware (Senger 2004).

How is SOA adopted in business networks? – Empiricalfindings

By analyzing 33 real-world SOA cases based on the multi-dimensional framework, we obtained interesting insightsinto the focus areas of SOA adoption, which we willillustrate in the following sections. We will compare theobserved patterns in our data set with propositions fromSOA literature to analyze the differences and commonali-ties with prior forms of IOS. Lastly, we will present fiveclusters representing specific forms of SOA adoption.

Strategic focus of SOA adoption

Our examination (cf. Table 1) reveals that most of the SOAprojects are vertically integrated business networks, involv-ing partners from the same industry in typical forward- andbackward integration scenarios. Interestingly, two cases useSOA for vertical integration (with their service providers)and horizontal integration (with their value chain partners), atthe same time. An example of such a scenario is EmergencyMedical Services providing time-critical services to allow theemergency management to exchange medical informationwith hospitals (vertical integration) using emergency notifi-cation services of telecom providers (horizontal integration).While SOA is still mainly applied between businesses, SOAprojects often involve several groups of partners simulta-

Table 1 Assessing SOA adoption at the strategy layer

Attributes Attribute Values

Value ChainIntegration*

Vertical Integration [#26, 79%] Horizontal Integration [#9, 27%]

PartnerHeterogeneity

]%33,11#[suoenegoreteH]%76,22#[suoenegomoH

RelationshipType*

B2B [#26, 79%] B2C [#11, 33%] G2G/G2B [#8, 24%] G2C [#6, 18%]

Number of Partners*

small n (< 10) [#4, 12%] n (> 10) [#7, 21%] N (> 100) [#22, 67%] na [#1, 3%]

NetworkTopology

1:1 [#2, 6%] 1:n [#23, 70%] m:n [#8, 24%]

NetworkStability

Stable [#27, 82%] Dynamic [#5, 15%] na [#1, 3%]

NetworkGovernance

]%12,7#[cirtnecyloP]%97,62#[lacoF

NetworkingPurpose

Channel Extension[#10, 30%]

Venture[#7, 21%]

Efficiency[#14, 42%]

Innovation[#2, 6%]

NetworkingTarget*

Strategic Partnerships[#13, 39%]

Op. Effectiveness[#18, 55%]

New Market [#7, 21%]

New Product [#10, 30%]

New Channel [#23, 70%]

StrategicPressure*

Customer Access[#20, 61%]

Improvement[#19, 58%]

M&A[#1, 3%]

Know-how[#3, 9%]

No Pressure[#7, 21%]

Legend: [# of occurrence in cases, percentage of cases]; attribute value “na” denotes case(s) without categorization* non-exclusive attribute values (multiple counting of cases).

186 J. Löhe, C. Legner

Page 7: SOA adoption in business networks: do service-oriented ...aws.iwi.uni-leipzig.de/em/fileadmin/user_upload/doc/Issues/Volume_20/Issue_03-04/V20I3...understanding of how SOA is applied

neously and extend electronic business networking toindividuals (B2C) and public administrations (G2G/G2Band G2C). In our empirical material, not only the dominanceof 1:n network topology suggest a centralist design of thebusiness networks, but also network governance which isdominated by one focal organization. As in the case of priorIOS, polycentric arrangements between equal partnersremain rare. Additionally, networks comprise mostly stablerelationships, whereas only a smaller number of dynamicnetworks leverage SOA for supporting their frequentlychanging business relationships.

Furthermore, we see that the SOA approach is mainlydriven by strategic positioning into new channels as well asoperational effectiveness, often accompanied by strongstrategic pressures to gain customer access and improvethe product & service offering. This market-driven focus isalso backed by about a majority of the cases that applySOA to connect with more than 100 partners, whereas onlyfew cases use SOA to electronically interact with a handfulof partners. One case combines the two characteristics byconnecting with a handful of suppliers and using their SOAsimultaneously to serve plenty of customers. Consideringnetwork purposes, our data set consists mostly of efficiencyand channel extension networks. Additionally, when look-ing at pair-wise relationships, we found that almost 50% ofour cases are vertical, homogenous, B2B networks. Theyrealize mostly focal, stable, 1:n relationships with morethan 100 partners, and are mainly driven by product &service improvement as well as customer access pressure.Those SOA implementations which address smaller settings(<10 and 10–100 partners) expose different characteristics:They are vertical B2B networks with homogeneouspartners focussing on operational effectiveness.

Process focus of SOA adoption

To further determine the characteristics and drivers of SOAadoption in business networks, we look to the process layer(cf. Table 2). Concerning cooperation processes, our data

set reveals that commerce processes are most often coveredby SOA projects, followed by finance, content & commu-nity, and supply chain processes. However, as an underly-ing technical platform, SOA often serves differentcooperation processes simultaneously. In line with theoutcomes at the strategic layer, our data set includes eitherend-to-end or customer-facing cooperation processes, andsurprisingly no supplier-facing cooperation processes.

Additionally, our case analysis exposes that SOA ismostly adopted in cooperation processes that producevirtual products, services, or outtasking as outputs. Fortwo order-to-delivery cases SOA adoption facilitates virtualproducts, services, or outtasking and physical productoutput in combination. Furthermore error-prone processes& high-coordination efforts represent the main challengesaddressed by SOA implementation, while capacity utiliza-tion, and process complexity & high process costs are otherimportant drivers for SOA.

IS focus of SOA adoption

Our case analysis confirms that SOA is still mostly appliedto build information systems within an organization’sboundaries (cf. Table 3). Although all SOA implementa-tions seek to integrate different internal and externalentities, in more than half of our cases, firms purely use aSOA-based infrastructure offered by an external partner butdo not necessarily have to implement a SOA themselves.Only some of our cases implement SOA across the entirebusiness network or use group-wide SOAs (e.g., tointegrate their subsidiaries). The organizations frequentlyuse single-vendor strategies. According to the process layerfindings, we identified that mainly e-commerce & CRMapplications as well as ERP & other core systems, or anycombinations of those, are affected by SOA.

Although dynamically discoverable business services areconsidered a key element of SOA, most of the analyzedcouplings are stable, i.e. they require extensive previousagreements, whereas on-the-fly coupling has to date played

Table 2 Assessing SOA adoption at the process layer

Attributes Attribute Values

CooperationProcess*

Content &

[#9, 27%]

Product Life-cycle[#2, 6%]

Commerce[#15, 45%]

Supply Chain [#8, 24%]

Maintenance &Repair

[#3, 9%]

Finance[#11, 33%]

ProcessOutput*

Physical Product [#10, 30%] Virtual Product, Service & Outtasking [#25, 76%]

CooperationSpan End-to-end [#17, 52%] Customer-facing [#16, 48%] Supplier-facing [#0, 0%]

Process-rel.Pressure*

Cap. Utilization[#12, 36%]

Coor.Eff.&Errors[#24, 73%]

Redundancies[#1, 3%]

Complexity Cost &Time [#10, 30%]

No Pressure[#5, 15%]

Legend: [# of occurrence in cases, percentage of cases] ; attribute value “na” denotes case(s) without categorization* non-exclusive attribute values (multiple counting of cases).

Community

SOA adoption in business networks 187

Page 8: SOA adoption in business networks: do service-oriented ...aws.iwi.uni-leipzig.de/em/fileadmin/user_upload/doc/Issues/Volume_20/Issue_03-04/V20I3...understanding of how SOA is applied

a subordinate role. In the same way, SOA is mostlyimplemented within one centralized instance and not as acompletely distributed architecture. In our data set, SOAmainly supports both automated machine-to-machine com-munication as well as human-to-machine communication,thereby offering multiple electronic communication optionswithin the business network. In line with the communica-tion type, the majority of our cases apply message-orientedand about two-thirds portal information exchange as well asany combination of different information exchange styles.A strong focus on function & data integration furtherconfirms that SOA adoption has to date been driven bytechnology.

With regard to standardization, we identified the major-ity of cases relying on technical standards as well as syntax& semantic standards. While Web service standards andnotably XML are increasingly used to share informationwith external parties in a standardized format, only ninecases benefit from standardized processes (pragmatics).This supports that SOA adoption is still in an early stagewith little standardization of coarse-grained service func-tionality. Accordingly, looking at service granularity,business process services are still rare, whereas mostorganizations simultaneously use fine-granular services,such as activity & task services or utility & entity services.Missing interoperability and pressure to offer multiple

integration mechanisms are most often cited as IS pressuresthat lead to the adoption of SOA, followed by legacy &monolithic applications as well as heterogeneous technol-ogies & platforms.

Comparing our empirical findings to SOA and IOSliterature

In case study research, it is generally recommended that,through pattern matching, researchers should compareempirically observed patterns with either predicted ones orpatterns established in previous studies and in differentcontexts (Dubé and Paré 2003; Gibbert et al. 2008; Yin2003). In order to assess whether our findings are consistentwith prior IOS and SOA research, we started by synthesiz-ing a set of propositions from prior literature (cf. Table 4).Empirical studies on IOS were the basis for derivinggeneral propositions on IOS adoption, and SOA publica-tions were used as source for propositions describing howSOA alters inter-organizational integration in businessnetworks. By analyzing the patterns that we find in ourdataset, we qualitatively evaluate whether our empiricalevidence supports or contradicts the propositions. Thisanalysis is a starting point for a further discussion andtheory-building related to SOAs’ specific contributions tobusiness networking.

Table 3 Assessing SOA adoption at the IS layer

Attributes Attribute Values

SOAScope

Entire business network [#8, 24%]

Different int. and ext. partners[#21, 64%]

Group[#4, 12%]

SOA Implem.Strategy*

Heterogeneous[#4, 12%]

Best-of-breed[#10, 30%]

Single Vendor[#11, 33%]

In-houseDevelopment

[#7, 21%]na [#2, 6%]

SOA affected Application*

ERP/Core App.

[#14, 42%]

CoreBank. &

Insur. [#7, 21%]

e-Com. & CRM

[#18, 55%]

CMS/DMS

[#9, 27%]

Groupw. &Comm

[#3, 9%]

MIS &BI

[#2, 6%]

e-Proc. & SCM/PLM[#5, 15%]

HR[#1, 3%]

Finance(Acc.)

[#1, 3%]

CouplingIntensity

Stable (previous agreements)[#26, 79%]

Dynamic (on-the-fly) [#4, 12%]

na [#3, 9%]

CouplingApproach

Centralized (1 instance)[#25, 76%]

Distributed (multiple instances)[#7, 21%]

na [#1, 3%]

Communic.Type*

Human-to-machine[#24, 73%]

Machine-to-machine[#31, 94%]

Info. ExchangeStyle*

Message-oriented[#28, 85%]

File-&Data-oriented

[#12, 36%]

Remotemethod inv.[#7, 21%]

Pipe-and-filter[#8, 24%]

Portal[#22, 67%]

na [#2, 6%]

IntegrationApproach*

Business Process Layer [#15, 45%]

Presentation Layer[#22, 67%]

Function & Data Layer[#31, 94%]

na [#2, 6%]

ServiceGranularity*

Business Process[#17, 52%]

Activity & Task[#28, 85%]

Utility & Entity[#23, 70%] na [#3, 9%]

Standardiz.Scope*

Pragmatic[#9, 27%]

Semantic, Syntactic[#27, 82%]

Technical[#29, 88%]

na [#4, 12%]

IS-relatedPressure*

MissingInteroperability

[#22, 67%]

Heterogenity[#17, 52%]

Redundancy[#3, 9%]

Security[#2, 6%]

Legacy &Monolithic[#19, 58%]

Costs[#7, 21%]

Missing &Imperfect SW

[#7, 21%]

No IS Press.[#0, 0%]

Legend: [# of occurrence in cases, percentage of cases] ; attribute value “na” denotes case(s) without categorization* non-exclusive attribute values (multiple counting of cases).

188 J. Löhe, C. Legner

Page 9: SOA adoption in business networks: do service-oriented ...aws.iwi.uni-leipzig.de/em/fileadmin/user_upload/doc/Issues/Volume_20/Issue_03-04/V20I3...understanding of how SOA is applied

Empirical studies emphasize that prior forms of IOS –notably EDI – are mostly used in stable and long-termconnected relationships (Daniel and White 2005; Gosain etal. 2004; Malone 1987; Reimers 2001). In contrast, SOAliterature proposes (P1) flexible SOA infrastructure anddynamically discoverable business services support dynam-ic business network configurations (Iyer et al. 2003; VanHeck and Vervest 2007). However, this proposition (P1) isnot supported by our empirical data. Not only highoccurrences of stable network configurations argue againstit. The use of stable, centralized couplings which rely onpre-defined arrangements also questions this proposition.

Numerous studies have shown that IOS often evolvearound a focal company with hierarchical and centralcontrol (Christiaanse and Venkatraman 2002; Clemons etal. 1993; Malone et al. 1987; Robey et al. 2008). They alsounderpin the role of the focal company in imposingelectronic linkages on customers or suppliers and setting

up the required infrastructure. In contrast, our secondproposition (P2) holds that SOA is expected to fosternetwork orchestration with distributed control and decision-making through interoperable platforms based on openstandards (Legner and Vogel 2008; Papazoglou andGeorgakopoulos 2003). However, our empirical data doesnot fully support P2. Focal network governance, verticalvalue chain integration, and 1:n network topology militatein favour of central and hierarchical instead of distributednetwork structures. At the same time, only some SOAimplementations realize polycentric governance or involvedifferent tiers of business partners, resulting in a m:ntopology. We conclude that SOA implementations are notinherently different from prior IOS as they still fosterhierarchical, centralized network structures. They extendhowever the scope of electronic collaboration to includea larger number of business partners from different tiersof the business network, such as customers and whole-

Table 4 Comparing empirical findings to propositions from IOS and SOA literature

LayerFindings from IOS

Literature Propositions from SOA Literature

Findings derived from Empirical Data

Stable, long-term, connected relationships (Daniel and White 2005; Malone et al. 1987)

(P1): Dynamic, quick connect and disconnect relationships (Iyer et al. 2003; Van Heck and Vervest 2007)

(P1) is not supported: most SOA-based business networks are stable and rely on predefined arrangements

(D1)Strategy

Business networks evolve around focal company with hierarchical and central control (Christiaanse and Venkatraman 2002; Clemons et al. 1993; Malone et al. 1987)

(P2): SOA will foster network orchestration with distributed control and decision-making through interoperable platforms, based on open standards (Legner and Vogel 2008; Papazoglou and Georgakopoulos 2003)

(P2) is only partly supported: focal arrangements still predominate and only few examples of polycentric networks have been found. But SOA extends the scope of focal arrangements by including a larger number of business partners from different tiers of the business network

Tight supply chain process integration with existing partners (Daniel and White 2005; Gosain et al. 2004; Son et al. 2008; Zhu et al. 2006)

(P3): SOA maintains customer-facing demand chain integration, through business services that can be called up and executed on demand (Zhao et al. 2007)

(P3) is supported: focus on customer-facing channel extension networks and information-intense processes

(D2)Processes Very rigid document-centric

integration (Banerjee and Golhar 1994; Damsgaard and Truex 2000; Kuan and Chau 2001; Reimers 2001)

(P4): SOA introduces a process-centric approach to partner integration, thereby replacing the document-centric one (Bussler 2002; Dorn et al. 2009; Legner and Vogel 2008)

(P4) is only partly supported: process-centric integration still lags behind document-centric; little use of standards (semantics and pragmatics)

Tight coupling and pre-engaged customization (Christiaanse et al. 2004; Damsgaard and Truex 2000; Daniel and White 2005; Reimers 2001)

(P5): Loosely coupled architecture based on modular, reusable, dynamically discoverable and complementary business and technical services (Demirkan et al. 2008; Iyer et al. 2003)

(P5) is only partly supported: first steps into the loosely coupled network platform direction through process activity and data services in larger network settings (D3)

InformationSystems

Silo applications and customized systems are a hindering factor in external integration (Zhu et al. 2006)

(P6): SOA can expose an organi-zation’s private computing assets as services for inter-enterprise collaboration (Demirkan et al. 2008; Hagel and Brown 2001)

(P6) is supported: companies expose services to provide external partners access to internal applications and data

SOA adoption in business networks 189

Page 10: SOA adoption in business networks: do service-oriented ...aws.iwi.uni-leipzig.de/em/fileadmin/user_upload/doc/Issues/Volume_20/Issue_03-04/V20I3...understanding of how SOA is applied

salers and sales agents as members of distributionchannel.

Given that IOS can affect the efficiency of a channel(Robey et al. 2008), former IOS were more often used tointegrate supply chain processes with existing partners(Daniel and White 2005; Gosain et al. 2004; Son et al.2008; Zhu et al. 2006). However, as a new networkingapproach (P3), SOA can facilitate customer-facing demandnetworks through business services that can be called upand executed on demand (Zhao et al. 2007). In our data set,we find strong empirical evidence that SOA adoption isfocussing on customer-facing or end-to-end processes,whereas we have not seen typical supply chain scenariosin our sample. P3 is also supported by high occurrences offocal companies collaborating electronically with custom-ers, wholesalers, sales agents, or other distribution channelmembers to provide virtual products and services.

Among the difficulties that have been reported withusing EDI is, among others, a general lack of a universallyacceptable standard for data transmission due to the veryrigid document-centric communication (Banerjee and Gol-har 1994; Damsgaard and Truex 2000; Kuan and Chau2001; Reimers 2001). The fourth proposition (P4) holdsthat SOA might overcome those difficulties by introducinga process-centric approach to integrate partners based onopen Internet standards (Bussler 2002; Dorn et al. 2009;Legner and Vogel 2008). High occurrences of machine-to-machine integration as well as pragmatic, semantic andsyntactic standards would be the preconditions for theprocess-centric partner integration. However, our empiricaldata provides only partial support for business process layerintegration, business process services, and process stand-ards. It seems that SOA implementations, like their IOSpredecessors, strongly rely on document-centric integrationand make little use of the standards required for realizingfully integrated and dynamic business processes. Hence, weargue that proposition four is only partly supported.

Likewise, prior forms of IOS often required considerablerelationship-specific investments as they needed pre-definedagreements and partner-specific customization (Christiaanseet al. 2004; Damsgaard and Truex 2000; Daniel and White2005; Reimers 2001). Consequently, they were often foundto be too costly and complex, especially for small andmedium-sized businesses (Kuan and Chau 2001). In contrast,the service-oriented paradigm is introducing more flexibleforms of electronic integration. The SOA proposition (P5)holds that loosely coupled network platforms can be realizedthrough the combination of modular, reusable, dynamicallydiscoverable, and complementary services (Demirkan et al.2008; Iyer et al. 2003). Our case material reveals that manySOA implementations still heavily rely on stable, pre-definedarrangements, whereas dynamic coupling is still rare.WhereasSOA implementations hardly realize the dynamic discovery

and composition of services in the way described by SOAliterature, we see some empirical evidence of more modulararchitectures based on reusable services. The occurrence ofbusiness-related services encapsulating tasks & activities orutilities & entities, which are used for both human-to-machineand machine-to-machine integration, tend to be steps into thedirection. Thus, we have partial support for (P5).

Finally, silo applications and customized systems in thecontext of IOS are a hindering factor in external integration(Zhu et al. 2006). Oftentimes companies have not been ableto seamlessly integrate electronic channels with theirinternal systems. SOA can now expose an organization’sprivate computing assets as services that support inter-organizational collaboration (P6) (Demirkan et al. 2008;Hagel and Brown 2001). Our findings support propositionsix: First, we have a high number of cases that seek tostrategically position a new channel and strive for opera-tional effectiveness based on tightly integrated cooperationprocesses within a larger network of (smaller) partners.Second, a large number of organizations provide directaccess to internal information via portal technology basedon an underlying SOA. Third, nearly all organizationsexpose activity & taks and entity & utility services, andpursue a function and data layer integration approach.

Five clusters of SOA adoption

While the previous section discussed SOA adoption in thecontext of the scientific literature, we further analyzed thepatterns in our dataset. By means of a cluster analysis, weidentified five different groups of SOA implementations (cf.Table 5):

Cluster 1 comprises SOA implementations, where oneorganization offers shared services to external parties. Inthis cluster (shared services), a large as well as heteroge-neous group of customers and business partners accessesthe SOA infrastructure provided by one organization andcalls Web services, mostly via a portal. One example is INGCard (Coenen 2006), which performs the credit scoringfunction centrally and makes it available as service to allother card systems within the ING Group. As mentioned inproposition six, this cluster exposes private computingassets as reusable services for inter-enterprise collabora-tion (Demirkan et al. 2008; Hagel and Brown 2001). SuchSOA adoption affects information-intensive processes, andthereby extends the scope of electronic collaborationbeyond transaction processes that were at the heart ofprior IOS.

Cluster 2 represents SOA implementations with a strongfocus on technical integration. It differs from the other clustersas it aims at solving interoperability issues due to heteroge-neous technologies and platforms as well as legacy andmonolithic systems. It comprises stable, machine-to-machine

190 J. Löhe, C. Legner

Page 11: SOA adoption in business networks: do service-oriented ...aws.iwi.uni-leipzig.de/em/fileadmin/user_upload/doc/Issues/Volume_20/Issue_03-04/V20I3...understanding of how SOA is applied

couplings and is the only cluster that makes substantial use ofstandardized semantics. Those SOA implementations seek torealize efficiency and effectiveness gains in processing high-volume data in distributed administrative processes, such aselectronic vehicle registration (Heffner 2005).

Cluster 3 represents customer-facing networks in theservice sector where SOA supports channel extension. Theservice-based coordination of distribution partners andproduct providers allows organizations (e.g., in the finan-cial or telecom industries) to move into new channels andenhance their product offerings. This is the only clusterwhere comprehensive business process integration isrealized and where legacy and monolithic systems as wellas heterogeneous technologies and platforms are connectedby semantically and pragmatically standardized servicecommunication. Overall, we summarize this group aschannel extension in customer-facing service networks.

There are several similarities between Cluster 4 and Cluster3, with exception of the industry and product focus. Cluster 4focuses on industrial distribution networks and concentrateson end-to-end commerce and supply chain processes forphysical products. The Volvo case (Holmqvist and Pessi2004) represents this cluster, with a multi-tier processintegration in an after-market scenario involving suppliers,manufacturers, dealers, and customers. Compared to Cluster3, the levels of process integration and standardization arelower. Compared to traditional supplier-facing process

integration, this cluster represents a developing and earlyform of end-to-end integration in industrial networks.

Cluster 5 is the smallest cluster and differs substantiallyfrom the other four. This markets / e-hubs group character-izes large, polycentric networks that realize dynamic m:nintegration. This approach is mainly used by onlineplatforms such as Amazon or eBay that support commercialtransactions and offer Web services that are integrated intotheir partners’ platforms (Wilkes 2004). This clusterrepresents distributed SOA adoption and supports mostlydynamic end-to-end couplings through function integrationvia task or entity and utility services as well as presentationintegration via portal technology. This group most expan-sively adopts SOA for inter-organizational integration, inaccordance with proposition two, and is closest to thevision of SOA enthusiasts.

While our cluster analysis reveals characteristic patternsof SOA adoption in the inter-organizational domain, it alsounderpins the fact that service-oriented coordination differsfrom traditional forms of IOS (notably, EDI-based supplychain coordination). Most importantly, SOA addressescoordination and improves interoperability in complex,multi-level networks, whereas prior forms of IOS whereoften limited to dyadic customer–supplier relationships.While commercial and supply chain transactions have beenat the core of many EDI and e-commerce applications,SOA extends the scope from typical order-to-delivery

Table 5 Cluster analysis

Brief description Cluster 1 Shared Services

Cluster 2Technical SOA

Integration

Cluster 3 Channel Extensionin Customer-facingService Networks

Cluster 4 End-to-End

Integration inIndustrial Networks

Cluster 5 Markets / e-Hubs

n = 6 4 8 8 3

Partner Heterogenity Heterogeneous Homogeneous Homogeneous Homogeneous Heterogeneous

Network Topology 1:n 1:n 1:n 1:n m:nNumber of Partners > 100 > 10 > 100 > 100 > 100

Network Governance &Stability

Stable, Focal Stable, Focal Stable, Focal Stable, Focal Dynamic,

Polycentric

Network Purpose Efficiency Efficiency,

Channel ExtensionChannel Extension Efficiency Venture

Cooperation ProcessContent &

Community,Finance

Finance,Commerce,

Mainten. & Repair Commerce, Finance

Commerce, SupplyChain

Commerce

Process Output Virtual Virtual Virtual Physical Physical / Virtual

Cooperation SpanEnd-to-end,

Customer facingEnd-to-end,

Customer facingCustomer Facing

End-to-end, Customer facing

End-to-end

SOA Scope External Interfaces External Interfaces. External Interfaces External Interfaces External Interfaces

SOA Affected Apps ERP & Core ERP & Core eCommerce & CRM, Core Bank. & Insur.

eCommerce & CRM, ERP & Core

eCommerce & CRM,CMS & DMS

Communication Type Human-to-machine,Machine-to-machine

Machine-to-machineMachine-to-machine,Human-to-machine

Machine-to-machine,Human-to-machine

Machine-to-machine,Human-to-machine

Coupling Intensity &Approach

Stable, Centralized Stable, Centralized Stable, Centralized Stable, Centralized Dynamic,

DistributedInformation Exchange

StylePortal, (Message-

oriented)Message-oriented,

pipe-and-filtersMessage-oriented,

(Portal)Message-oriented,

(Portal)Message-oriented,

Portal

Service Granularity Activity/Task,

Entity/Data/Util.Activity/Task

Business Process,Activity/Task,

Entity/Data/Util.Activity/Task

Activity/Task, Entity/Data/Util.

Standardization TransportSemantics, Transport

Semantics, Transport

TransportSemantics, Transport

SOA adoption in business networks 191

Page 12: SOA adoption in business networks: do service-oriented ...aws.iwi.uni-leipzig.de/em/fileadmin/user_upload/doc/Issues/Volume_20/Issue_03-04/V20I3...understanding of how SOA is applied

processes to also cover other transactional as well asinformation-intensive processes. Service-based coordina-tion thereby intensifies electronic coordination to coverthe entire customer and product life-cycles, but also fostersthe pooling of resources and strategic partnerships. In thisregard, SOA has a particular role to play in coordinatingcustomer-oriented networks with heterogeneous customerand channel partners. Another interesting finding, and onethat contradicts a popular assumption, is that we identifiedonly a few scenarios of polycentric business networksapplying SOA for dynamic, m:n coordination.

Limitations

While this is an early study on SOA adoption in businessnetworks, its results must be viewed in light of somelimitations. Most importantly, our research is based onsecondary data of successful real-world SOA implementa-tions. While this provides us with a rich case database andallows us to study phenomena in their context, our data setmay involve only positive SOA cases, which ignore imple-mentation failures. We argue that this survivor bias carries lessweight because we only observe characteristics and drivers ofSOA implementations and do not intend to measure SOAimplementation success. The use of secondary data does notonly have implications in respect of the data quality, but alsoimposes the problem that the data was originally collected fora different purpose. This may limit the external validity of ourfindings. We tried to address this challenge by not onlyanalyzing the case for single attributes, but also took the entirecontext description (e.g., examples, figures, and attributerelationships) into account. This allowed for an understandingof why and how SOA was adopted and for the correctinterpretation of whether or not an attribute was present.During our analysis, multiple iteration steps and discussions toagree on a classification further improved validity.We ensuredinter-rater reliability through a research framework that isexplicitly derived from the literature, to provide an objectiveand common understanding. Moreover, we verified constructvalidity by explaining the data collection circumstances anddata analysis procedure (Gibbert et al. 2008).

Furthermore, the small number of valid cases (33) doesnot allow for inferring statistically valid conclusions onSOA adoption. Given the early stage of research, our goalwas not to take a representative sample, but to analyze adiverse set of SOA implementations using replication logic(rather than a sampling logic) to provide valuable insightsin a first exploratory research phase. As mentioned earlier, amultiple-case study research design allows for analyticalgeneralizability. The knowledge gained from replicatedcase studies is usually considered to increase the externalvalidity of the findings (Yin 2003).

Conclusion

The objective of our research was to improve ourunderstanding of SOA’s role and implementation in inter-organizational settings. Given the fact that no classificationscheme or taxonomy of IOS or SOA adoption in businessnetworks exists, a first contribution to research is the multi-dimensional framework that covers three dimensionsdescribing the alignment of strategic and organizationalcapabilities with the IS/IT architecture in a given businessnetwork. Since the derived classification scheme builds onprior conceptualizations from business networking and IOSliterature, it allows for systematically assessing SOAimplementations in the inter-organizational domain andcomparing them to prior forms of IOS.

Through our analysis of 33 real-world SOA implemen-tations that sought to improve their external integrationcapabilities, we were able to analyze focus areas of SOAadoption in business networks and derive relevant charac-teristics and drivers at the strategic, process, and IS/ITdimensions (RQ1). Moreover, we used our empiricalevidence to show how SOA is leveraged for inter-organizational coordination and how SOA implementationsdiffer from prior forms of IOS. In doing so, our researchcontributes to anchoring the technology-oriented stream ofresearch on SOA in prior IOS literature. Based on ourempirical data, we could not confirm all the promisingpropositions related to SOA’s impact on business networks.Most importantly – and in contrast to SOA literature – ourempirical analysis suggests that most SOA-based businessnetworks are still stable and rely on predefined arrange-ments. Although we found little empirical evidence ofpolycentric business networks applying SOA for dynamic,m:n coordination, our study also clearly reveals that SOAadds new aspects to IOS. Our cluster analysis reveals fivepatterns of SOA adoption in the inter-organizational domain(RQ2). It demonstrates that SOA allows additional groups(e.g., smaller businesses, public administration, and individ-uals) to participate in electronic collaboration and supportsB2B, B2C and G2B/G2B/G2C relationships. At the sametime, SOA extends the scope of IOS from dyadic supply chainrelationships to improve coordination in customer-facingnetworks while emphasizing end-to-end integration.

As SOA provides organizations with a more standard-ized, scalable, and modular platform for internal andexternal integration, it eliminates some severe shortcomingsof tightly-linked, proprietary machine-to-machine commu-nication and forms the basis for supporting complexelectronic interactions in multi-level business networks.

Our conceptual and exploratory work can be helpful tofuture research by providing systematic insights into SOAdeployments in the inter-organizational context and ad-vancing the understanding of how SOA should be

192 J. Löhe, C. Legner

Page 13: SOA adoption in business networks: do service-oriented ...aws.iwi.uni-leipzig.de/em/fileadmin/user_upload/doc/Issues/Volume_20/Issue_03-04/V20I3...understanding of how SOA is applied

leveraged in business networks. First, future empiricalstudies can build on our multi-dimensional framework andpropositions to validate our findings. Second, the frame-work might serve to study and assess alternative approachesto IOS and SOA implementation in business networks.Third, another interesting field of research is the develop-ment of design guidelines for service-based architectures

that increase networkability and take into account differentpartner segments and levels of process integration.Fourth, we recommend further studies on SOA benefitsand value in business networks as well as on the variousfactors that impact SOA adoption so as to help ensureefficient and effective SOA applications in businessnetworks networks.

Appendix

Table 6 Overview of SOA case studies

Case Description

Amazon.com Virtual shopping mall that enables a wide range of companies to use the Amazon platform totrade on-line.

Arid Lands Infor- mation Center (ALIC) Library-extension collaboration by delivering a calendar of events from the AgricultureNetwork Information

Center portal to County Extension Web pages.

Australian Telecom Process integration of a telecommunication wholesaler’s order management system to morethan 150 different retailers.

British Telecom Coupling of new offerings with those of third parties and integration with the internal,mainframe-based billing, provisioning, and other support systems.

Cendant Travel Distribution Services Small and larger services to perform specific tasks (shop and book a flight, car and hotel,and show pictures) for different travel products, business units and customers.

Deutsche Telekom SOA for the order fulfillment process with the expansion of indirect distribution channelsand an extensive enlargement of the product and service portfolio.

Dorma-Glas Integration between the customers’ inventory management system and Dorma’s SAP, whereWeb services enable electronic ordering.

Dutch Information Bureau SOA to link the files of social security organizations, to streamline data exchange betweenmunicipalities and third parties, as well as to implement social security laws and automatethe verification and issuing of social security benefits.

eBay Web services-based access to core applications for large and small businesses trading largevolumes of goods.

Emergency Medical Services Scenario Provisioning of time-critical services to exchange and dispatch emergency medicalinformation between different organizations.

Endress + Hauser Portal-based, Web-enabled asset management solution with Web services for end-to-endprocess integration with customers.

Experian Services for providing clients with greater visibility into the mainframe-based process and theresulting marketing data, thereby increasing speed and interactivity.

Bundesamt für Migration und Flüchtlinge Integration of a large amount of external authorities and training providers into theauthorities’ predefined business processes.

Fidelity National Financial Migration of the flagship loan origination software to a SOA environment in order to addresslarger markets and support a range of Internet-based front ends, Web services, and externalservices.

H&R Block Financial Advisors A SOAP-based data access framework for tax and mortgage planning, recruiting, andproductivity reporting to improve access to a wide range of multiple divisions’ enterprisedata, as well as process orchestration for sales lead management.

Helvetia Patria Integration of e-business applications of multiple autonomously working national subsidiarieswith their own IT infrastructure and data for streamlining and simplifying the insuranceproducts’ sales processes.

IBM Corporation SOA based on RosettaNet PIP and XML message exchange to automate the ordering ofmidrange computing systems for independent distributors in several countries bystandardizing and improving business processes.

ING Group ING Card built an application for its customer base that enables it to link to new Web sites,implement new product features, and maintain credit scoring rules easily and quickly.

SOA adoption in business networks 193

Page 14: SOA adoption in business networks: do service-oriented ...aws.iwi.uni-leipzig.de/em/fileadmin/user_upload/doc/Issues/Volume_20/Issue_03-04/V20I3...understanding of how SOA is applied

References

Al-Naeem, T., Rabhi, F. A., Benatallah, B., & Ray, P. K. (2004).Systematic approaches for designing B2B applications. Interna-tional Journal of Electronic Commerce, 9(2), 41–70.

Alonso, G., Casati, F., Kuno, H., & Machiraju, V. (2003). Webservices: Concepts, architectures and applications. Springer.

Alt, R., & Österle, H. (2004). Real-time business: Lösungen,Bausteine und Potentiale des Business Networking. Springer.

Alt, R., Legner, C., & Österle, H. (2005). Virtuelle Organisation—Konzept, Realität und Umsetzung. HMD- Praxis der Wirtschaft-sinformatik, (242), 7–20.

Baida, Z., Rukanova, B., Liu, J., & Tan, Y.-H. (2008). Preservingcontrol in trade procedure redesign—the beer living lab.Electronic Markets, 18(1), 53–64.

Banerjee, S., & Golhar, D. Y. (1994). Electronic data interchange:characteristics of users and nonusers. Information Management,26(2), 65–74.

Bauer, S., & Stickel, E. (1998). Auswirkungen der Informationstech-nologie auf die Entstehung kooperativer Netzwerkorganisationen.Wirtschaftsinformatik, 40(5), 434–442.

Benbasat, I., Goldstein, D. K., & Mead, M. (1987). The case researchstrategy in studies of information systems. MIS Quarterly, 11(3),369–386.

Buhl, H. U., Heinrich, B., Henneberger, M., & Krammer, A. (2008).Service science. Wirtschaftsinformatik, 50(1), 60–65.

Bussler, C. (2002). The role of B2B engines in B2B integrationarchitectures. SIGMOD Record, 31(1), 67–72.

Bussler, C. (2003). B2B-integration: Concepts and architecture.Berlin: Springer.

Chen, M., Zhang, D., & Zhou, L. (2007). Empowering collaborativecommerce with Web services enabled business process manage-ment systems. Decision Support Systems, 43, 530–546.

Christiaanse, E., & Venkatraman, N. (2002). Beyond sabre: anempirical test of expertise exploitation in electronic channels.MIS Quarterly, 26(1), 15–38.

Christiaanse, E., Van Diepen, T., & Damsgaard, J. (2004). Proprietaryversus internet technologies and the adoption and impact ofelectronic marketplaces. The Journal of Strategic InformationSystems, 13(2), 151–165.

Clarke, R. (2001).Towards a taxonomy of B2B e-Commerce schemes. Paperpresented at the 14th Bled Electronic Commerce Conference). Bled.

Table 6 (continued)

Case Description

Luxemburger Familienkasse SOA integrating existing mainframe and proprietary client-server applications to pay outsocial benefits to French frontier workers; integration of different data sources and a safedata exchange with French authorities.

NTUC Income (BigTrumpet.com) Web service-based portal that aggregates numerous services and allows users the full controlover and access to personal information anywhere and anytime.

US E-Government: Law enforcement SOA provides a message switching solution that supports information exchange betweenvarious US agencies that are engaged in crime prevention and internal security.

Provide Commerce In-house-developed e-Commerce solution rebuilt on an SOA where Web pages routecustomers’ order requests to business services. Then based on a custom-built rules enginethe business services may invoke other services behind them.

Queensland Transport Integration with vehicle dealerships’ applications to electronically initiate and track theregistration process based on XML data exchange.

Södra Cell SOA solution that allows smaller customers to manually administer their paper pulp needs viaa portal directly in Södra Cell’s business system; larger customers can choose to integratetheir business systems.

Sony SOA provides the flexibility to quickly add or change workflow processes associated with thedigital asset management systems.

Thomson Prometric Services provide necessary interfaces and connection points for white-labeling Prometric’sWeb site with registration to tests and managing test center capacity by pre-allocating ablock of seats to tests.

Transamerica Life Insurance Existing legacy systems as a set of core, reusable Web service building blocks that areassembled to create new processes and applications and to provide the business partnerswith real-time self-service access to insurance products via different distribution channels.

Unique (Zurich Airport) SOA-based portal solution to improve management insights drawing data from multiple coreapplications operated by different outsourced service providers and running on differentunderlying platforms and operating systems.

Volvo Implementation of an advanced Web service solution to improve after-market process integration.

Vontobel / Raiffeisenbanken Provisioning of a broad range of services related to investment fund products forRaiffeisenbanken. To be multi- client capable.

Wachovia (2 cases: front and back officeintegration)

Service-oriented solution framework for Internet banking (portals, process orchestration, Webservices, service aggregations, business rule engine, etc.); Enabling customer access viamultiple channels (Internet, ATM, etc.).

wine.com SOA rebuilding the warehouse management system and enhancing the customer-facing Web site.

194 J. Löhe, C. Legner

Page 15: SOA adoption in business networks: do service-oriented ...aws.iwi.uni-leipzig.de/em/fileadmin/user_upload/doc/Issues/Volume_20/Issue_03-04/V20I3...understanding of how SOA is applied

Clemons, E. K., Reddi, S. P., & Row, M. C. (1993). The impact ofinformation technology on the organization of economic activity:the “move to the middle” hypothesis. Journal of ManagementInformation Systems, 10(2), 9–35.

Coenen, A. (2006). An SOA case study: Agility in practice. Retrievedfrom http://www.soamag.com/I2/1106-1.pdf.

Damsgaard, J., & Truex, D. (2000). Binary trading relations and thelimits of EDI standards: the Procrustean bed of standards.European Journal of Information Systems, 9(3), 173–188.

Daniel, E. M., & White, A. (2005). The future of inter-organisationalsystem linkages: findings of an international Delphi study.European Journal of Information Systems, 14(2), 188–203.

Demirkan, H., Kauffman, R. J., Vayghan, J. A., Fill, H.-G.,Karagiannis, D., & Maglio, P. P. (2008). Service-orientedtechnology and management: perspectives on research andpractice for the coming decade. Electronic Commerce Researchand Applications, 7(4), 356–376.

Dorn, J., Grün, C., Werthner, H., & Zapletal, M. (2009). Frombusiness to software: a B2B survey. Information Systems and E-Business Management, 7(2), 123–142.

Dubé, L., & Paré, G. (2003). Rigor in information systems positivistcase research: current practices, trends, and recommendations.MIS Quarterly, 27(4), 597–635.

Erl, T. (2005). Service-oriented architecture. Concepts, technology,and design. Prentice Hall International.

Fettke, P., & Loos, P. (2003). Classification of reference models: amethodology and its application. Information Systems and E-Business Management, 1(1), 35–53.

Fleisch, E. (2001). Das Netzwerkunternehmen. Strategien undProzesse zur Steigerung der Wettbewerbsfähigkeit in der “Net-worked Economy”. Berlin: Springer.

Frick, N., & Schubert, P. (2009). An empirical study of the currentstate of B2B integration in practice. Paper presented at the 22ndBled eConference “eEnablement”). Bled, Slovenia.

Gibbert, M., Ruigrok, W., & Wicki, B. (2008). What passes as arigorous case study? Strategic Management Journal, 29(13),1465–1474.

Gosain, S., Malhotra, A., & El Sawy, O. A. (2004). Coordinating forflexibility in e-Business supply chains. Journal of ManagementInformation Systems, 21(3), 7–45.

Hagel, J., III, & Brown, J. S. (2001). Your next IT strategy. HarvardBusiness Review, 79, 105–113.

Heffner, R. (2005). Real-world SOA: SOA platform case studies.Forrester.

Heutschi, R. (2007). Serviceorientierte Architektur—Architekturprin-zipien und Umsetzung in der Praxis. Berlin: Springer.

Holmqvist, M., & Pessi, K. (2004). Process integration and webservices a case of evolutional development in a supply chain.Scandinavian Journal of Information Systems, 16, 117–144.

Huang, Y., & Chung, J.-Y. (2003). A Web services-based frameworkfor business integration solutions. Electronic Commerce Re-search and Applications, 2(1), 15–26.

Huhns, M. N. (2008). Services must become more agent-like.Wirtschaftsinformatik, 50(1), 74–75.

Iyer, B., Freedman, J., Gaynor, M., & Wyner, G. (2003). Webservices: enabling dynamic business networks. Communicationsof the AIS, 11, 525–554.

Jacobides, M. G., & Billinger, S. (2006). Designing the boundaries ofthe firm: from “Make, Buy, or Ally” to the dynamic benefits ofvertical architecture. Organization Science, 17(2), 249–261.

Kambil, A. (2008). Purposeful abstractions: thoughts on creatingbusiness network models. The Journal of Business Strategy, 29(1), 52–54.

Kambil, A., & Short, J. E. (1994). Electronic integration and businessnetwork redesign: a roles-linkage perspective. Journal of Man-agement Information Systems, 10(4), 59–83.

Kauffman, R. J., & Weill, P. (1989). An evaluative framework forresearch on the performance effects of IT investment. Paperpresented at the 10th International Conference on InformationSystems). Atlanta, GA.

Klein, S. (1996). Interorganisationssysteme und Unternehmensnetzwerke:Deutscher Universitätsverlag.

Kuan, K. K. Y., & Chau, P. Y. K. (2001). A perception-based modelfor EDI adoption in small businesses using a technology-organization-environment framework. Information Management,38(8), 507–521.

Legner, C., & Vogel, T. (2008). Leveraging web services forimplementing vertical industry standards: a model for service-based interoperability. Electronic Markets, 18(1), 39–52.

Li, Y., Lu, X., Kuo-Ming, C., Huang, Y., & Younas, M. (2006). Therealization of service-oriented e-Marketplaces. Information SystemsFrontiers, 8, 307–319.

Linthicum, D. S. (2001). B2B application integration: E-business-enable your enterprise. Boston: Addison-Wesley.

Löwer, U. M. (2005). Interorganisational standards: Managing webservices specifications for flexible supply chains. Physica-Verlag.

Luthria, H., & Rabhi, F. (2009). Service oriented computing in practice—an agenda for research into the factors influencing the organizationaladoption of service oriented architectures. Journal of Theoreticaland Applied Electronic Commerce Research, 4(1), 39–56.

Maglio, P. P., Srinivasan, S., Kreulen, J. T., & Spohrer, J. (2006).Service systems, service scientists, SSME, and innovation.Communications of the ACM, 49(7), 81–85.

Malone, T. W. (1987). Modeling coordination in organizations andmarkets. Management Science, 33(10), 1317–1332.

Malone, T. W., Yates, J., & Benjamin, R. I. (1987). Electronic marketsand electronic hierarchies. Communications of the ACM, 30(6),484–497.

Mayring, P. (2000). Qualitative content analysis. Forum: QualitativeSocial Research, 1(2).

McAfee, A. (2005). Will web services really transform collaboration?MIT Sloan Management Review, 46(2), 78–84.

Melville, N., Kraemer, K. L., & Gurbaxani, V. (2004). Review:information technology and organizational performance: anintegrative model of IT business value. MIS Quarterly, 28(2),283–322.

Merrifield, R., Calhoun, J., & Stevens, D. (2008). The next revolutionin productivity. Harvard Business Review, 86(6), 72–80.

Österle, H., Fleisch, E., & Alt, R. (2001). Business networking—shaping collaboration between enterprises (Second Revised andExtended Edition ed.) Springer.

Papazoglou, M. P., & Georgakopoulos, D. (2003). Service-orientedcomputing. Communications of the ACM, 46(10), 25–28.

Papazoglou, M. P., & van den Heuvel, W.-J. (2007). Service orientedarchitectures: approaches, technologies and research issues. TheInternational Journal on Very Large Data Bases, 16(3), 389–415.

Premkumar, G. (2000). Interorganization systems and supply chainmanagement: an information processing perspective. InformationSystems Management, 17(3), 56–69.

Reimers, K. (2001). Standardizing the new e-business platform:learning from the EDI experience. Electronic Markets, 11(4),231–237.

Reitbauer, S., Kohlmann, F., Eckert, C., Mansfeldt, K., & Alt, R.(2008). Redesigning Business Networks—Reference Process,Network and Service Map. Paper presented at the Symposiumon Applied Computing (pp. 540–547). Fortaleza, Ceará, Brazil.

Robertson, P. L., & Langlois, R. N. (1995). Innovation, networks, andvertical integration. Research Policy, 24(4), 543–562.

Robey, D., Im, G., & Wareham, J. D. (2008). Theoretical foundationsof empirical research on interorganizational systems: assessingpast contributions and guiding future directions. Journal of theAssociation for Information Systems, 9(9), 497–518.

SOA adoption in business networks 195

Page 16: SOA adoption in business networks: do service-oriented ...aws.iwi.uni-leipzig.de/em/fileadmin/user_upload/doc/Issues/Volume_20/Issue_03-04/V20I3...understanding of how SOA is applied

Rodon, J. (2006). A methodological and conceptual review of interorganizational information systems integration. Paper presentedat the Fourteenth European Conference on Information Systems(pp. 402–413). Goteborg.

Senger, E. (2004). Zum Stand der elektronischen Kooperation—Fallstudien, Muster und Handlungsoptionen. Unpublished Dis-sertation, Universität St. Gallen.

Siedersleben, J. (2007). SOA revisited: Komponentenorientierung beiSystemlandschaften. Wirtschaftsinformatik, 49(Sonderheft), 110–117.

Sinz, E. J. (2008). SOA und die bewährten methodischen Grundlagender Entwicklung betrieblicher IT-Systeme. Wirtschaftsinformatik,50(1), 70–72.

Son, J.-Y., Narasimhan, S., Riggins, F. J., & Kim, N. (2008).Understanding the development of IOS-based trading partnerrelationships: a structural model with empirical validation.Journal of Organizational Computing & Electronic Commerce,1(18), 34–60.

Spohrer, J., & Maglio, P. P. (2008). The emergence of service science:toward systematic service innovations to accelerate co-creation ofvalue. Production & Operations Management, 17(3), 238–246.

Tafti, A., Mithas, S., & Krishnan, M. S. (2008). The effects ofinformation technology and service-oriented architectures onjoint venture value. Paper presented at the ICIS 2008). Paris.

Van Heck, E., & Vervest, P. (2007). Smart business networks: how thenetwork wins. Communications of the ACM, 50(6), 29–37.

Viering, G., Legner, C., & Ahlemann, F. (2009). The (Lacking)business perspective on SOA—critical themes in SOA research.Paper presented at the 9. Internationale Tagung Wirtschaftsinfor-matik (pp. 45–54). Vienna, Austria.

Vogel, T., Schmidt, A., Lemm, A., & Österle, H. (2008). Service anddocument based interoperability for European eCustoms solu-tions. Journal of Theoretical and Applied Electronic CommerceResearch, 3(3), 17–37.

Wareham, J., Zheng, J. G., & Straub, D. (2005). Critical themes inelectronic commerce research: a meta-analysis. Journal ofInformation Technology, 20(1), 1–19.

Wilkes, L. (2004). Amazon and eBay web services—the newenterprise applications. CDBi Journal(October), 9–20.

Yin, R. K. (2003). Case study research—design and methods (3rd edn.Vol. 5). SAGE Publications.

Zhao, J. L., & Cheng, H. K. (2005). Web services and processmanagement: a union of convenience or a new area of research?Decision Support Systems, 40(1), 1–8.

Zhao, J. L., Tanniru, M., & Zhang, L. J. (2007). Services computing asthe foundation of enterprise agility: overview of recent advancesand introduction to the special issue. Information SystemsFrontiers, 9(1), 1–8.

Zhu, K., Kraemer, K. L., Gurbaxani, V., & Xin Xu, S. (2006).Migration to open-standard interorganizational systems: networkeffects, switching costs, and path dependency. MIS Quarterly, 30(Special Issue on Standards), 515–539.

196 J. Löhe, C. Legner