Changing Spaces

Changing Spaces for Social Learning in the Implementation Biography of an ERP System: a Longitudinal Case Study G.M. Campagnolo 1, S.Ducati2

Abstract The research presented in this paper focuses on four different periods of implementation choices in the history of an italian public sector organization as ERP software support shifts from being provided in-house to a market-based supply. The case illustrates a deeply contextual reflexivity between the varied compositions of supply and use space and the social learning leading to the technical knowledge that shapes members representations of the technical feature of the ERP system driving implementation choices. We explore the role played by the reconfiguration of actors and their interactions along the ERP system support chian across time in patterning the way project participants make sense of notions like customizations or standardization. Data on distinctive locational patterns of organizational resources concerning post-implementation enhancements of the system were gathered through interviews and observations of the system in use. Four different locational patterns of IT related expertise have been identified in the case across different time periods: the ‘implementation hum’ period (1998-2001), the ‘era of personalization’ (2002-2005), the move from the in-house software enhancement to external consultancy (2005-2008) and the more recent appointment of a different consortium of multiple consultants to implement additional software modules.

1

Department of Sociology and Social Research, University of Trento, Via Verdi 26, I – 38100 Trento, [email protected]. 2 Faculty of Sociology, University of Trento, Via Verdi 26, I – 38100 Trento, [email protected].

2

1. Introduction Considerable attention has been placed in recent years to the dilemma user organisations can face regarding how much to adapt and enhance the ERP package they have acquired, as this may prejudice maintainability and ability to take on board future enhancements and upgrades (Brady et al., 1992; Fincham et al. 1995). However, the connections between implementation choices and issues of postimplementation maintainability underline the need user organizations may have to address the whole package implementation lifecycle with technologies such as ERPs. The career of the package within the implementation arena of its organisational users (the ‘project’ life cycle) seems to represent only one moment in a bigger picture in the evolution of the technology itself (the ‘product’ lifecycle), alongside the further development of the software supply market. These observations point to important processes of social learning as an initially generic software suite is appropriated or domesticated within a particular organisational setting and the broader context of supply and use (Williams, 2005). Academic and practitioner literature have focused upon documenting the various forms extensive process of customization and adaptation of the ERP software systems could take (Pollock & Cornford 2004; Richmond & al. 2006; Liang & al., 2004; Kumar & al., 2003; Light et al. 2001). However, very little has been written on the evolution of implementation choices (and the way participants try to make sense of them) about the same ERP system in the same case study over time and on how these choices relate to each other. Some authors have begun to explore different “configurations” of spaces of social learning using data coming from different field studies (Stewart & Hyysalo, 2008). In this study, we build upon these initial explorations to find different configurations of social learning in different time periods on a single case. In this research paper we attempt to contribute to the development of theoretical approaches focused on long-term social learning processes in the appropriation of ERP systems in the broader context of supply and use of the public sector . By focusing on a single italian public sector organization case study - let us call it Dante Province -, we highlight how implementation choices depend upon the distribution of responsabilities for the realization of the software across diverse intermediary actors in the space between supply and use at different times. We draw on and extend contemporary thinking about ERP systems appropriation, and social learning in technological innovation by foregrounding the spatial aspects involved in knowledge formation about the appropriation and domestication of an ERP system within a particular organizational setting. Our aim is to understand the spatial aspects of knowledge formation about ERP implementation and after-implementation. We begin by providing some background on the dimensions of the dilemmas related with implementation choices and ERP systems appropriation. ERP vendors stress the capacity of their systems to provide best practice solutions for all firms and the advantages of ‘vanilla implementation’ of ERP packages (utilising

3

the functionality it incorporates without adaptation). Customisation has been portrayed by vendors and ERP consultants as liable to prejudice the success of ERP, bringing high costs when the system is implemented (Davenport 1998; Liang et al. 2004; Soh et al. 2000). It is striking therefore that the practices of user organisations have diverged sharply from these recommendations. Surveys of ERP implementations in the late 1990s revealed some instances (representing a minority of implementations) in which organisations, having identified gaps between the package and their organisation’s ways of doing things, decided to just live with these gaps or re-engineer their business processes to meet the requirements of the package. More frequently (and in the majority of implementations), there was an extensive process of customization and adaptation of the software systems. These adaptations could take various forms: configuring the package (Clausen & Koch, 1999; Pollock & Cornford 2004), customising the package (Davenport 1998; Soh & al., 2000; Brehm & al. 2001; Richmond & al. 2006), partial, selective implementation of package (Davenport 1998; Liang & al., 2004; Clemmons & Simon, 2001), add-ons, bolt-ons or ‘extension software’ (Kumar & al., 2003; Sprott, 2000) and ‘best of breed’ multi-vendor systems (Mabert & al., 2001; Light et al. 2001). Furthermore, for the user organisation, the adoption of ERP constitutes a major and long-term investment – one that shapes the user organisation’s information systems and practices for extended periods. The benefits of ERPs in terms of organisational performance typically take a long time to accrue. This can be attributed to an extended process of experimentation as the organisation grapples with the affordances of these enormously complex systems for their own purposes. Managerial guidelines for how to achieve successful outcomes from ERP implementation place increasing weight on the post-implementation phase (Somers & Nelson, 2004; Wei & al., 2005). Implementation choices becomes not so much a question of depth of technical expertise but more of identifying interactions between third party participants affecting/intermediating the user/developer relationship in the development, implementation and post-implementation of packaged software in extended periods. In many cases, the issue is not to reveal available technological choices and analyse the forces determining which designs are eventually adopted. Instead, the situation is often characterised by an apparent absence of choice, and the problem is to account for this ‘absence’. In doing so, attention is focused on the ‘real’ limiatations on choice which are located in the wider social system, and which bear in upon specific contexts in which technical change is taking place. Resolving these calls for extensive, and by implication shallower, knowledge of the enormously wide array of potential interactions between techological products and other elements of the broader socio technical infrastructure affecting implementation choices, it creates issues about how such knowledge can be represented and managed in extended periods. In this paper we respond by proposing a theoretical approach that develops a spatial focus upon social learning processes in ERP implementation and provides insights into the relationship between supply and use for implementation choices concerning ERP systems in extended periods. Theorizing the geographical dimension of social learning in ERP adoption is fundamental to identify how implementation choices depend upon the distribution of responsabilities for the realization

4

of the ERP software beyond the organizational unit of analysis across a diversity of actors of which software vendors are only one among others. Theorizing its historical dimension is crucial to understand its changes over time. We author four locational patterns of user/developer relationship identified during my analysis that highlight the relational and historical nature of technical knowledge formation guiding implementation choices. Our conclusions suggest that the way learning participants make sense of implementation choices cannot be understood starting from the view of a single organizational actor nor even from a single (or modular) organizational implementation project. Rather, they depend upon organizational moves taking place in an extended situation, involving constiutencies taking form in the wider inter and intra-organizational suppy/use interactional space. Furthermore, the creation of representations of ERP implementation and support needs continues throughout multiple generations of product development, where each generation tends to build representations of ERP uses in relation with what has been learned by the previous one. We suggest that this highlights the need that the scope of ERP projects and their inherent risks challenging IS researchers to inform the work of practitioners have to draw insights from the analysis of historically and geographically extended cases of ERP implementations. Through our spatial analysis we reveal the distributed (Pentland, 1992) and abductive (Almkov, 2008) nature of technical knowledge formation concerning implementation choices in organizations. The main theme that we explore centers around the question: what counts as a resources for social learning in ERP implementation? Locational patters we describe during a tenyears ERP project highlight that both the distribution of actors in space and their changes in interactions over time count as a resource for social learning. Our spatial turn within social shaping of technology and social learning provides insight into popular notions such as trajectories of development (Fleck et al., 1990), sociotechnical constituency (Molina1989a), supporting the idea that (i) implementation options are not given but they emerge as organizational moves for the distribution of responsabilities for the realization of software in a network of practices within an evolving supply-use interactional space and that (ii) organizational appropriation of the package can not be interpreted from the point of view of a single organizational implementation project. Rather, appropriation depend to a larger extent upon knowledge gathered from previous organizational implementation projects.

2. Theoretical Background In this section we outline the theoretical foundations of the paper, providing a brief overview of social learning in technological innovation (SLTI) in its conjunction with the theme of spatiality. We chose SLTI because we were studying a long standing ERP project at Dante Province were actors involved in the implementation were changing over time and SLTI combines social shaping of technology with the evolution of spatial configurations in organizations and it does so at a level of analysis that includes the broa-

5

der institutional and societal context, in an attempt to explain both particular instances of technology and the ‘general characteristics of a society’s technological ensemble’ (Russell & Williams, 1988, p. 11). Most government software development before 1998 was made in Dante Province by either an organization’s internal IS staff or by direct subcontract to a software house. Dante Province was the first Italian public administration implementing an ERP system. With ERPs, software development is increasingly being done by specialized software vendors and requires extensive post-purchase tailoring and draw eavily on software service market. The main focus of this market in Italy was to serve the fit between the existing government information system and the new ERP packages together. In Dante Province this involved long and drawn out third parties relationship building (vendors of interoperable add-ons, software service consultants) and stop-start processes of institutional learning and forgetting about how to build a software support chain across a constantly changing network of actors. In the sociotechnical usage, social learning denotes the reflexive yet often negotiated, complex and ‘political’ process in transforming environment, instrumentation and work that reach beyond the boundaries of a single organization. We adopt an interpretive, articulative and practice-based research stance (Gubrium, 1998; Nicolini, 2006; Nicolini et al., 2003). We shall thus endorse a type of social science which focuses on observing daily practices and understanding how members of situations assemble a reasonable understanding of the things and events that concern them. Looking at learning through the lens of practice shifts attention from production and delivery only - from structural or organizational perspectives - to include consuption and the perspective of participation (Brown & Duguid, 2001). We do not intend to use the lens of practice as an useful organizational subset. Rather, we use the concept of practice to demarcate the extent to which knowledge can spread and bind together extensive disciplines most of whose members will never know one another. The ERP system in Dante Province serves the management of 14 long-terms plans, each corresponding to a relevant organizational Unit. Organizational Units in the Trento Province are divided in Staff Services and Line Services. Staff Services are organizational units that provide services to other internal units. Line Services are organizational units that provide services to other parties (citizens, firms, etc...). Long-term plans have to be filled only by Line Services, while Staff services like the Planning and the Accounting Services just consult them as reports. Beyond that, participants of the Dante Province ERP implementation and post-implementation network of pratice are 38 developers (of which two thirds from Dante Province in-house IT company, and a third from third party programming suppliers), the IT in-house company general manager, general managers from 5 different Dante Province Departments, 4 different consulting companies or consortia changing over time, 340 users of the system for data entry, of which 60 accountants, 8 key users and 10 tutors. As the ERP project at Dante Province has been a partial, selective ERP implementation, with some of the 10 modules/applications being introduced at years of distance from each other, we also became interested in how the array of relationships tracked out from the organization contemplating ERP adoption was changing in extended times. SLTI supports an analysis of these issues, and through its particular mapping of inter-

6

mediaries between supply and use and sensitivity for constant changes of networks of actors, helps us better understand the implications that can be drawn from developments like ERP in the public sector, together with its frequent political agenda changes and continuous updates of legislative-driven goals (Rusaw, 2007). A brief overview of the basic tenets of SLTI may help those readers unfamiliar with it. SLTI is a relatively recent approach developed out of the tradition of “social shaping of technology” approach (Williams & Edge, 1996; MacKenzie and Wajcman, 1999) by combining it with the work from organizational studies on ‘organizational learning’ (Schon, 1983), signalling a focus not only upon analysing processes but also upon the possibilities for intervention, particularly in policy making (Sorensen & Williams, 2002). SLTI has been applied by a variety of IS process-oriented researchers to remind us that many contemporary technologies are not discrete self contained systems but “configurations” and that the learning dynamics associated with them not only shape technology but can have dramatic effect on the structure of the innovating network, the constitution of the organizations involved and the nature of technical knowledge (Hasu, 2001; Russel & Williams, 2002; Hyysalo, 2006). Altough alternative theoretical lenses are available to analyze the conjunction between social issues and space, SLTI stresses that technology emerges through a complex interaction between many diverse players and the importance of giving more detailed accounts of how these actors play key roles in innovation in the long term. The conjunction between social issues and space has been explored at different levels of analysis by the network and embeddedness perspective in economic and organizational sociology, by micro-sociological approaches and by the actornetwork (ANT) analysis in the sociology of scientific knowledge (Granovetter, 1985; Knorr-Cetina & Bruegger, 2002; Latour, 1992). However, in network approaches, networks are sparce social structures, and it is difficult to see how they can fully account for what we observe in the course of inter- and intraorganizational relations in terms of intense and dynamic conversational interactions, knowledge flows and temporal structures. Network approaches have been criticized for containing no sense of the specific processes and mechanisms of knowledge transfer and the consequences these may have for technology choices and their societal outcomes (Knorr-Cetina & Bruegger, 2002, p.910). A closer look is needed to capture the interactional means of structuration that are embedded in post-local forms of software service sourcing. However, the assumptions that have characterized much microsociological thinking in the past - that tacit knowledge requires spatial proximity while codified knowledge is ubiquitous or that knowledge externalities are spatially agglomerated - are seen as theoretically no longer adequate in a world in which interactions can also be dis-embedded from local settings, in which space may be separated from place (Giddens 1990, p.18). Situating knowledge in extended practices allows knowledge production to be viewed both as ‘local’ and as ‘distributed’ without privileging the former over the latter or, more generally, the micro over the macro (Harris 1998, p.289) . Places can thus be imagined as ‘articulated moments in networks of social relations and understandings’ (Massey, 1991, p.28). As noted by Quattrone & Hopper

7

(2006), ANT analysis in IS and elsewhere has concentrated on the emergence of IT as black boxes. By developing a critique of studies which are concerned only with social relations, ANT argues that such relations count little unless they are held together by durable and resistent materials (Quattrone & Hopper, 2006). As such, ANT emphasizes the quality of networks that endure and remain stable across space. To follow Murdoch distinction (Murdoch, 1998) these networks often demarcate ‘spaces of prescription’. Networks as spaces of prescription help clarify how IT systems acquire stability and become taken for granted, but ANT neglects what happens after systems achieve this status. This carries the danger of wrongly assuming that organizational worlds achieve order and stability once processes leading to black boxing are identified (Quattrone & Hopper, 2006). With Dante Province case, we want to illustrate how the ERP system changed continually and differently across various interactional spaces and time periods to meet emergent development-in-use and use-in-development demands and how in this ‘technological ferment’ was very difficult to ever achieve “closure”. The SLTI approach is not a narrowly cognitive, social or modelling process and, as noted by Hyssalo and Stewart (2008), the term “social learning” is used in a very different way to its usage in education and social psychology such as that of Bandura (1977). This usage also differs from more generic conceptions of social learning in evolutionary economics, where learning tends to be taken as an explanatory term for growth in learning economy (Lundvall & Johansson, 1994). The SLTI approach depicts a framework to position various intermediaries between primary supply organizations and primary user organizations of technology systems as well as in regards to developing and using them (Williams et al., 2005). The SLTI framework also illustrates differences in profiles and in consequent mediating capacities of intermediaries. The mediating capacity is described in terms of “reach” between supply and use and “width” of the content they cover.

Fig.1 (adapted from Williams et al., 2005)

The most studied type of intermediaries are supply-side industrial consultants (Hislop 2002, Newman & Westrup 2005). This is because they tend to be more nu-

8

merous, visible and formal than those close to the end-users. What the SLTI framework let me articulate with its sensitivity for micro-scale mechanisms and social dynamics and symmetrical treatment of supply and use spaces is that there are often less formal use side intermediaries that may perform their work as peer favours or sidejobs to their formal work. In Dante Province case a central role in ERP after-implementation support chain is played by key users. Key users have an important role in the implementation and post-implementation period of the biography of the ERP project in Dante Province. The key user is a typical intermediary user in terms of the SLTI approach. He has the role to represent employee requests of technical support to the IT maintenance group, asking required customizations. At the same time, key users have to know the system, be aware of what the consequences of customizations are and be able to communicate to their colleagues what the standard version of the system can do. Key users are selected among Staff structures. Differently from Operative services, they do not directly use the system: their role is to monitor what Operative service do with the system. Altough the role of key user is a sidejob in addition to the more institutional professional position, the appointment of this role can have great consequences on the individual carreer. After becoming the reference person of the Dante Province ERP project implementation, one of our informants has been moved from the IT Department to the Accounting Department, to be closer to the Accounting requirements and transfer them to the ERP developers and after 6 years she is still there. Space has been a concern in IS research particularly in inter-organizational networks (Jin & Robey, 2008), virtual organizations (Panteli, 2003) and geographically distributed work (Zolin & al., 2003). In these studies, the implications of time-space distantiation have been discussed in terms of interpersonal trust, creation of social and technical interfaces and different articulations of presence. But, in these beginnings of spatial theorization in IS, the inclination has been to use space in primarily territorial terms (to identify distant territorial placements) or as an artefact of co-location, rather than through its varied compositions and their enrolment into knowledge formation. Space has also been criticized for being seen in its absolute view as a ‘container’of human cognition (Boland, 2001) and organizational action (Tsoukas, 1992) or as having different conceptions in different cultures (Sahay, 1998). According to concerns with ‘time-space compression’ (Harvey, 1989) and ‘time-space distantiation’ (Giddens, 1990) which reflect a geographical focus upon economic and social processes and their related spatial configurations, this paper works with an extended definition of space, one that recognizes territorial placements but also other spatial arrangements (Amin & Cohendet, 2005). ERP systems are perceived as configurational technologies that cross occupational, task and even organizational boundaries and draw together operations performed by previously separate islands of information automation, with everything in between only concerned with aligning, converting or transmitting the diverse information contained within the previously separated units. We argue instead that what is happening with the modern organization implementing ERPs is that this “everything in between” is doing more than aligning dispersed databases, but it

9

counts as a formative space. As a methodological implication, we identified a singular SAP module - SAP is the ERP system used by Dante Province - as a starting point for our fieldwork in Dante Province: the Project System module (PS). The idea of selecting a single module as a starting point was to be able to identify relevant intermediary actors through their enrolment in the SAP support chain rather than from their organizational function. Representatives of each organizational unit concerned with SAP-PS module have been interviewed. Providers of technical support on the SAP-PS system have also been identified and interviewed. By putting a focus on organizational consuption of ERPs, we also propose that complex network formation and management devices are required to organize the many wishing to speak for ERP systems implementation and support. The organization of the software support chain is one of these management devices we identified in Dante Province. Along that chain, the right to speak for the system is taken by different actors at different times. The organization of these turns and their position along the chain affects the relative power each intermediary actor (key user, IT manager, consultant, management accountant) has to modify the system, directly or indirectly, by influencing the decisions of the software developers. In Dante Province, ERP developers have been from time to time: members of the in-house IT company, technology provider external consultants and other kind of consultants. The changing relative position of external consultancy and the different interactions taking place between key users and develpers had rather surprising implications on implementation choices. Proximity between the developer and the user is described by the ERP implementation project manager as a negative quality for effective interaction. The developers of the inhouse IT company, belonging to the same administrative organization, were doing everything the Accounting department required with little knowledge of its feasibility. The technology provider external consultants, not having any organic dependency from Dante Province, were instead perceived as more restrictive with the requirements, shortening the time of feasibility study. We start with the premise that spaces of social knowing and learning involve all manner of spatial mobilization of which the SLTI approach emphasizes the following: facilitation, configuration, brokering . They all involve creating, increasing mantaining and defending spaces of different types: social (communities, networks), knowledge (skills and know-how resources), cultural (positive images), physical (a place or equipment), economic (providing funds), and regulatory (creating rules to guide activities and reduce uncertainty) spaces. Facilitation is providing opportunities to others, by distributing resources, influencing regulations and setting local rules. Configuring is more an act of infuencing perceptions and goals that involves the intermediaries in active processes of configuration of technical and symbolic means. Brokering is an act of bringing together spaces of different types: e.g. supply and use space. Through our spatial analysis we want to reveal how the varied compositions of supply and use space at different times enrol into knowledge formation. We want to illustrate the distributed (Pentland, 1992) and abductive (Almkov, 2008) nature of technical knowledge formation in ERP implementation and after implementation. A template we find useful to conceptualize the distribution of competences in

10

ERP implementation and subsequent search comes from Brian Pentland research on ‘telephone hotlines’ (1992). When faced with complicated calls, support workers embarked on what, drawing on Goffman, Pentland call ‘organizing moves’: they often moved the problem through giving a call away to someone else in the same office. In terms of the latter Pentland writes: ‘there were four generic kinds of moves through which this was accomplished: assign, refer, transfer and escalate. These moves [...] involved assigning the responsibility for a call to someone else. The responsibility could shift back to the originator at a later time, but for the time being, the call would be someone else’s problem’ (1992: 539). ‘To assign’ is a strategy of spatial mobilization consisting in the assignement of a problem to a functional area or team that seems appropriate based on the user’s description of the problem. In Dante Province case ‘tickets’ are the formal identification of a technical support call (or e-mail) from a user. ‘To refer’ is a further stage in the software support chain. When a ‘ticket’ cannot be solved in the first place by the generic support team, it is referred to a specific functional analyst within the in-house IT company. ‘Transfer’ consist in an horizontal circulation of the problem among functional analysts. ‘Technical escalation’ concerns the transfer of a problem motivated by the need of greater technical skill (typically from internal functional analysts to an external consultant). Since this involves greated expenditures, technical escalation is often coupled with a political escalation, involving greater managerial autority (typlically from the IT Department). Escalation implies also a different packaging of the problem: the ordinary ‘ticket’ becomes an evolutionary ‘project’ and it is ruled and resourced as such. In our view, and where we differentiate our perspective from Pentland’s, is that we think the ability to distribute problems is not a capacity simply residing at the level of situated, face-to-face interactions, but it takes shape across a global network of practices connected up by the ERP support chain in extended times. When making decisions on how to assign, refer and escalate the problems, participants to this network of practice are drawing on experience and reasoning from more or less similar episodes in the past. Both Charles S. Pierce (1958: 90), Misak (2004: 16ff.) and Bateson (2002[1979]: 133ff.) note that ‘abduction’ is an important aspect of all scientific work (also see Harries-Jones, 1995: 177-83). Simply put, abduction is distinguished from induction and deduction, and involves reasoning through analogy: a search of resemblances from which to construct heuristic rules of thumb, and eventually to formulate and test new hypotheses. Thus, the reader wouldn’t expect a detailed description of a single instance of organizational move. Rather, our case illustrates the broader picture of the changing strategies of spatial mobilization and technical knowledge flows in the company across different time periods and their implications on implementation choices. Some authors have begun to explore different “configurations” of spaces of social learning using data coming from different field studies (Stewart & Hyysalo, 2008). In this study, we build upon these initial explorations to find different configurations of social learning in different time periods on a single case. For the

11

purpose of analysis, the case of ERP implementation in Dante Province is described across four different time periods, from 1998 to 2008. Table1 The different configurations of social learning in different time periods the implementation ‘hum’ (2000-2002) the era of customizations (2002-2005) from the in-house IT company to SAP external consultancy (2005-2008) from the SAP consultant to a multi-partner consortium (2008-...)

Each period corresponds to a specific IT-related expertise locational pattern. In the first period that I term the ‘implementation hum’ period, the functional analysis expertise, the programming and the accounting competence were found together in the same place within the implementation team. There were no external consultants working in the team. The Dante Province at that time was the first Italian public administration implementing a SAP version for the public sector. A second period, described as the ‘era of personalizations’, describes the period following the implementation team quit. A product support chain was built including the appointment as key users of internal staff services employees. As mentioned before, internal Staff services employees were not direct users of the system. Staff Services role was to monitor the coherence and regularity of the data input in the system by the Operative Services. At that time, the IT in-house company started to outsource programming functions to small external consultants able to program the system, while maintaining feasibility and functional analysis in-house. In the third period, feasibility studies and analyses have been also outsourced to the technology supplier consultant. Other public administrations started to learn how to implement SAP. Consultants started to have experiences on the implementation of that system in the public sector. The fourth period concerns the issuing of a Consortium for the implementation of new system modules in the stead of the SAP consultants. Despite Dante Province wanted to continue with the SAP consultants, the in-house IT company decided to empanel a Consortium made by academic partners, other non-SAP consultants and members of the in-house IT company. The decision has been justified by the fact that the Consortium was already in place for other SAP implementation projects in other local public administrations (e.g. the University). The coverage of a decade of implementation choices in the public sector nicely illuminates the relative - historically determined - and relational - distributed - nature of the social learning process leading to technical knowledge formation. In the implementation ‘hum’ time period adopting a ‘standard’ version of the system meant giving responsibility for the data input and for the control to each single Operative service system user, for the purpose of decentering accounting services. In the subsequent period, the ‘vanilla implementation’ strategy started to be perceived as a strategy that did not fill the key user requirements enough. When consultancy was performed by the technology supplier, standardization becomed synonimous of ‘best practice’. And finally, with the Consortium the idea of the

12

diffusion of SAP coupled with the strategic goal of aligning a fragmented landscape of public sector organizations made of large organizations and smaller ones to common representation of organizational procedures. The representation of an identical system’s technical feature changed over time depending upon the relative position of intermedary actors and their goals in a constantly chainging (and growing) supply and use space. Each representation and the space in which it resides does not just react to the local network of actors interests. It is produced to a larger extent also as a consequences of former time periods representations of similar technical features.

3. Research Approach The paper presents the results of an interpretive, articulative and practice-based research project (Walsham, 1993; Nicolini, 2006; Nicolini et al., 2003) conducted between april 2006 and December 2008 at Dante Province. The lead author is an academic who began to study ERP systems usage in the public sector in the international context as part of his doctoral thesis (2003-2008). Drawing on personal connections, after that he began to build a network of contacts within the italian public sector, leading to a collaboration with his co-author. The co-hautor is a master student with ten years working experience of ERP usage in Dante Province - she manages the planning activities of her Unit by using SAP. She brings insights from professional practice, cumulative reading of technical reports and indepth knowledge of government policy documents. Conducting research with a “practical theorist” (Hoffman 2004) helped connect information systems theory with practice bringing specialist understanding to bear in what is widely acknowledged to be a sector under great evolution. The role of a practical theorist is similar to that of an ethnographer’s informant providing insights into the logic of the research setting, except that the practical theorist collaborates on a level playing field. During the analysis process the practical theorist is a collaborator making sense of the data. At practitioner/academic workshops the research becomes a polyphonic, “jointly told tale” (Van Maanen 1988:136) enabling the authors to speak to multiple levels of interest in the room. The field research was designed with the aim of providing a description of Dante Province ERP implementation choices evolution across the extended range of historical tempi and locales of the innovating market of software service supply in the public sector. We endorsed a type of social science which focuses on observing daily practices and understanding how members of situations assemble a reasonable understanding of the things and events that concern them. In particular, our methodological purpose is to describe sense-making activities as strategies of spatial mobilization. In order to capture this aspect, we did not focus only on the participants directly involved in local interactions. We also addressed local interactions and sensemaking as part of a strategy of spatial mobilization played out by an extended network of practice. The learning reproduced in such spaces is seen as reflecting back on actors sense making activities in situations and having a role in shaping

13

further strategies of spatial mobilization. Addressing this range of timeframes and levels of generalization that surrond social action in ERP implementation required different modes of construction of the field. Our fieldwork practice relied on “polymorphous engagement” (Guterson 1997:116) and multi-sited ethnography (Marcus 1995:96). Polymorphous engagement means interacting with informants across a number of dispersed sites, not just in local communities, and sometimes in virtual form (Guterson 1997:116). Multi-sited ethnography identifies a new mode of ethnographic research that “moves out from the single sites and local situations of conventional ethnographic research designs to examine the circulation of cultural meanings, objects, and identities in diffuse time-space” (Marcus 1995:96). Our field in Dante Province has been initailly constructed by following different kind of trajectories: − − −

the development of the practice supported by the ERP system; the development of the ERP implementation strategy; the development of the people involved in ERP implementation and after implementation;

In particular, a single SAP module has been initially identified as a starting point of our strategy of field construction: the Project System (PS) module. The choice of the PS module derives from its representativeness of Dante Province implementation strategy of introducing a single module on all organizational units before moving to the next. The PS has been introduced in all 14 Dante Province organizational units dealing with the “Pluriennal Sectorial Plan”. The practice supported by the ERP system consists in the management of Dante Province programs, plans and objectives by the Council and by the accounting management. PS in Dante Province counts 14 plans managed by about 340 users as content providers. As with the approval of the “Pluriennal Sectorial Plan” (Dante Province main strategic plan), the PS module is used in conjunction with the DDP module (the module that hosts the written administrative documents) and the COP module (the accounting module). Table 2 identifies the organizational structures involved in Dante Province planning activities at different stages and the role of the SAP ERP system for each stage. The second strategy adopted for the construction of our field consists in following the development of the ERP implementation and after-implementation. As well as examining formal intermediaries, our strategy aims to reveal those intermediaries that arise at the interstices between organisations as an informal role and gives greater attention to broader social relations, so to explain the finer structure of ERP implementation and post-implementation.

14

Fig.2 The activities supported by the SAP system

The key actors to further interview and observe have been identified starting from their role on the PS module implementation and post-implementation. Starting from there, we identified other software modules and organizational actors having a connection with the PS. The third and related strategy of field construction consists in following the development of the people involved in ERP implementation and post implementation in their work-setting, by interviewing and observing them. Table 3: interviews and observations table The development of the people involved in ERP implementation and postimplementation Functional Analyst of Interviewed: 5 March 2008 the in-house IT company Project manager of Dante Province Interviewed: 21 November 2007; SAP implementation project 3 March 2008. Key User for the Programming Service Interviewed: 10 January 2008 Observed: August 2008 Key User for the Accounting Service Interviewed: 3 March 2008 Observed: August 2008 Operative Service user Observed: August 2008

According to a ‘snowball’ effect (Czarniawska, 1998), this third strategy allowed us to further extend the research field and finally complement our map with the overall picture of the network of practice involved in the SAP implementation pro-

15

ject in Dante Province. The data collection methods employed to gather the empirical material to refine the researchers understanding of Dante Province implementation strategy have been biographic interview, participant observation and documents reading and discussion. Interviews have been conducted in a biographic form starting from a first invitation to analyse the professional life of the research participants concerning their connection with SAP in Dante Province. The biographic interview provides to the researcher a discourse in which “ the beliefs, the attitudes, the values, the representations of the biographic trajectories are expressed with a particular emotional colouring, and are inscribed within an argumentative structure that dermines their sequence and manifests their connections”(Cardano, 2007:74). All the interviews were tape recorded and verbatim transcripts were produced (Riessman, 1993). The second data collection method has been participant observation (Van Maanen, 1979; Rosen, 1991). The observation centered around ‘hot moments’. In particular, the participant observation took place during a major change of the “Pluriennal Investments Plan” with subsequent evolutive maintenance of the SAP PS module. The observation concerned three organizational units: the Unit that proposed the change of the Plan and the Planning and Accounting Service Staff Units, whose activity was to monitor the changes provided by the proponent Service. In addition to fieldwork interviews and observations, we also included the Italian public administration norms between 1996 and 2007 concerning the investment planning activities and the way they are distributed between political organs and administrative management, together with public reports concerning Dante Province IT strategy. The first pass through the data has been centered on basic sorting and inductive techniques information by an adaptation of grounded theory (Strauss and Corbin 1998). After discussion, both authors agree upon the following tematic data clusters: aim and scope of SAP implementation; implementation of SAP; characteristics of the activities supported by SAP; role of the organizational structures adopting SAP; customizations; relationship between standard SAP and customized version; key users role; transversality; responsibility; criticalities; advantages of SAP; outsourcing of SAP implementation and post-implementation tasks; iter of post-implementation development. The final empirical material covered a ten years time period (between 1998 and 2008) and concerned in total the implementation of 10 between SAP modules and applications, involving 38 developers from the IT in-house company and third party suppliers, 8 key users from as many Dante Province organizational units, 4 different consultancy firms and 340 users. The empirical material has been clustered around for time periods. Each historical era points to the reflexive nature of technical knowledge about implementation choices (such as those concerning to what extent to allow customization or to keep a standard version of the system module) and the contextual frame within which Dante Province administration members had to make sense of the changes of implementation choices taking place, connected to a description of the supply and use space that emerged (the enacted supply/use space). The time periods are only one representation of our longitudinal corpus of data, however they provide the foundation from which we claim

16

to identify four primary forms of space of social learning undertaken by Dante Province ERP project: the technological experiment, the intermediary and proxy user centred design; innofusion and generification. Although our empirical evidence speaks to all these forms of social learning, and we regard them as interlinked, we will focus our discussion here on how each different strategy of knowledge mobilization enacted different learning spaces in each period and how representations of implementation choices (customization vs. standardization) emerging from each historical period are interlinked with each other as a resource for abductive sense making. In the next session the empirical material is presented in the form of a case study. We choose a case study strategy because our research question is embedded in a specific organizational context within the living memory of our interviewees (see Yin 2003). The description of the empirical material will serve the analysis by describing the supply-use space of each time period.

4. Results The case of ERP post-implementation activities arrangements in the case of Dante Province is described across four different periods, covering the time span of a decade. Each period corresponds to a specific IT-related expertise locational patter. In the first period that I provisionally term the ‘Implementation Team’ period, the functional analysis expertise, the programming and the accounting competence were found together in the same place within the implementation team. There were no external consultants working in the team. The Dante Province at that time was the first Italian public administration implementing a SAP version for the public sector. A second period, described as the ‘era of personalizations’ , describes the period following the implementation team quit. A product support chain was built including the appointment as key users of internal staff services employees. As mentioned before, internal Staff Services employees were not direct users of the system. Staff Services role was to monitor the coherence and regularity of the data input in the system by the Operative Services. At that time, the IT in-house company started to outsource programming functions to small external consultants able to program the system, while maintaining feasibility and functional analysis in-house. In the third period, feasibility studies and analyses have been also outsourced to the technology supplier consultant. Other public administrations started to learn how to implement SAP. Consultants started to have experiences on the implementation of that system in the public sector. The fourth period presented concerns the issuing of a Consortium for the implementation of new system modules in the stead of the SAP consultants. Despite Dante Province wanted to continue with the SAP consultants, the in-house IT company decided to empanel a Consortium made by academic partners, other non-SAP consultants and members of the in-house IT company. The decision has been justified by the fact that the Consortium was already in place for other SAP implementation projects in other local public administrations (e.g. the University).

17

4.1 The implementation hum (1998-2001) It was the 1998 when SAP was first introduced in the Dante Province. SAP italian version of the verticalization for the public sector had just come out the year before. At that time, SAP in the public sector was implemented in Spain and Austria, but there were neither italian public administrations implementing SAP nor consultant knowing how to implement SAP in the italian public sector. A temporary ad-hoc implementation team has been empanelled by Dante Province for the purpose of first introducing SAP. It was formed by a project reference person, internal employees and by members of the in-house company, at that time performing the fasibility study, the functional analysis and the programming of new pieces of software. One of the most relevant aims for introducing SAP in the Dante Province was to decenter accounting services. Accounting services previously performed by a centralized group of 30 accountants were going to be distributed over 400 people. As it was first introduced, the SAP modules were in a standard version that is without customizations: “Each single user, each single unit was responsible for the data input and for the controls. We structured the system in a way that would have solved the information requirements of the Staff units, in order for them to be able to monitor. But essentially, the system was intended to serve the Operative services users” (Dante Province ERP Project manager). The modules served the management of 14 long-terms plans, each corresponding to a relevant organizational Unit. Organizational Units in the Trento Province are divided in Staff Services and Line Services. Staff Services are organizational units that provide services to other internal units. Line Services are organizational units that provide services to other parties (citizens, firms, etc...). Long-term plans have to be filled only by Line Services, while Staff services like the Planning and the Accounting Services just consult them as reports. The idea, in the implementation period, was that SAP should have served to manage the common data, while the sector-specific data should have been left with other applications, autonomously choosen by the Line Services. This would have avoided duplications of data and should have induced a clearer overall vision of the SAP functionalities: “Let’s say that the structural data, the sectorial data, should not be inputted in SAP, because it is useless to input in SAP information that does not have an added value for all. One must say: ‘ok, the data in SAP are the official data’. From there, beneath, or as a complement, one has her excel table where the source data and the sectorial data are identified. In this way there is no duplication” (SAP project reference person).

4.2 The era of personalizations (2002-2005)

18

Lately, this team has been dissolved. The SAP project reference person has been moved from the IT Division to the Accounting Division. In addition, the programming tasks (e.g. writing with ABAP code) have been outsourced by the inhouse company to small external consulting companies. In the stead of the expertise found together in the ad-hoc implementation team, a process where Staff Service Planning and Accounting key users were in contact with a maintenance group located at the in-house company has been introduced. Starting from that period, key users participated in meetings with the in-house maintenance group to take decisions about software enhancements, provide observations and approve all analysis and design documents from the maintenance group and give feedbacks on the prototype version of the system. 4.2.1 Key users- developers The reference person of the ERP project implementation in the Dante Province describes the key user as an important role in the post-implementation period of the biography of the ERP project. The key user has the role to represent employee requests of technical support to the IT maintenance group, asking required personalizations. At the same time, key users have to know the system, be aware of what the consequences of personalization are and be able to communicate to their colleagues how they can better work with the standard version of it: “Key users have a fundamental relevance. In some modules there are key users that force themselves to learn about the system and a big deal about the processes. There are instead key users that are not always able to negotiate with the others. Some others find an agreement with everybody, perhaps because they do not follow their ideas to the end. And this is not good. (ERP Project reference person). Key users found that the SAP as it was first implemented before 2002 was not adequate: “The standard SAP PS module was not good, because the public administration has constraints about authorization, balance, allocations. The private sector does not have such rigid constraints. The standard version of the module did not have cross-controls; there was no integration, neither with administrative provisions, nor with the accounting. Then it was as having information in separate and totally unrelated boxes” (Key User Planning Division). A number of personalizations took place starting from that period. An important aspect of personalizations concerns the relevant decision process. In the Dante Province, all decisions about personalization were taken in meetings between the Planning Division and the Accounting Division key users with the maintenance group of the in-house IT company. Planning and Accounting are Staff divisions. They do not directly use the system. Their role is to monitor what Operative Service do with the system. Only when decisions taken by Staff Services key users required personalization costing more than 10 working days from the maintenance

19

group, they were submitted to the IT Division. Otherwise, the IT Division was not involved in the decision. Customizations taking place during that period (that we term the era of personalizations) are presented by the key users as required because “informatics must be coherent with the provisions” (Key User Planning Division). Each of the 14 different Operative Services long-term budget plans include management particularities mandated by the law according to the specific service they provide (highway administration, emergency management have specific management requirements that are not shared by other kind of Operative Services). Thus, in order to be fully compliant with the regulation, the overall customized version in use included all the possible exceptions that are found in each Operative Service: “There are some plans that can directly commit. They do not have to book funding. They are the highway administration, the local autonomies and the disaster management services. Then, for the ‘control of the budget commitment availability, plans working with this particular typology have to be foreseen. Then, in the registry information of the plan, a function has been added to specify: ‘Does this plan require to book funding?’ since the system has to identify what kind of control to apply. Then there are some plans that are divided in sections. The aggregated vision of the sections at the level of the plan has been added as a further function to the module.” (Key User Planning Service). A representative of the IT maintenance group describes this situation as follows: “ In SAP we did a very rich reporting, extremely rich and extremely sophisticated I would say [...]. So, in theory, if the SAP reports are not complete, I do not really know what can be more complete” (IT Analyst). When it comes to consider the technical point of view, the same interviewee tells, “the interventions on SAP are done on a system that is 99% personalized [...]. When there is to upgrade, when there are new releases, we have to put the hands on all the previous personalizations. Then every time is like it is a new implementation project” (IT Analyst). 4.2.2 Accounting key user - planning key user Conversations highlights that the amount of the personalization required presents an unbalancing towards the accounting functionalities. The reference person of the SAP project tells about the SAP implementation: “We wanted to see SAP from the point of view of the project or of the plan, not from the accounting point of view [...]. But the logic has been changed.” (SAP Project reference person, Dante Province). The key user of the Planning Service tells that during the analysis phase of the SAP module that her service was going to use (the Project System - PS - module) they often met with the Accounting Service key user, because many planning aspect required “the sharing and the agreement by the Accounting Service” (Key User Planning Service). The relation of key user with each others shapes the im-

20

plementation and the post-implementation aspects of the module in different ways. In the implementation phase, their decision implies the amount of cross-controls between the Accounting and the Planning modules. Typically, cross-controls are in place when at stake is to control the coherence and the regularity of the LineServices inputs. Automatic controls generate blocking messages on Operative Services operations if their are not compliant with both the restrictions set on the system by both the Planning and the Accounting key users. Controls are manual, that is they do not generate blocking messages, but only warning messages - when they concern cross-checking between the data on the Planning module and on the Accounting module. During the post-implementation period “changes in the system might have a domino effect: you change sosmething and something happens somewhere else in the system that you did not expect” (Accounting key user). This generates problems with the identification of who is responsible for what when it is to identify problems, try to solve them or report them to the maintenance group: “Since accounting and planning modules are two integrated modules, it is not always immediate to identify what can be a problem. When you notice a problem you are not able to understand: is this a problem of the planning module, or is this a problem of accouting...And then who to address for this” (Key User Accounting). 4.2.3 Key users - final users Where the interviewed key users fully agree is on the required customizations on SAP in that they represent useful features for the Line Services. According to both key users, the on-line controls and the reports provided by the system allow the Line Services to have respectively (i) the confidence to do not forget anything or make mistakes in the data input, and (ii) to have the update situation of their project available: “The controls and the reports are mainly made for the Operative structures. With the controls, the operative services are confident that they can not make mistakes. And reports allow them to have all information, like the availability to invest” (Key User Planning Service). However, from the observations the fieldworker accomplished during a major evolution of the system, it emerged that each Staff Structure (and their related key users) has a different monitoring styles. This requires the Line Service operators to duplicate their efforts to comply with the Staff structures monitoring requirements. They have to provide different types of document containing the same data to different Staff Structures: some of these documents even include the use of additional software extra-SAP. An illustration of this is that while the Planning key user is content with the report that is automatically generated by the system, the Accounting Key User does not like it: “ It is not an easy to read file! They had to have it all on a single page!”(Accounting Key User). As a consequence, in order to perform the relevant controls, she asks the Line Services to provide a spreadsheet in the replacement of the SAP automatically generated reports, with an half-page ta-

21

ble summary of the total budget chapter per year. Line Services themselves are also keeping external data records on spreadsheets, due to a lack of project management functionalities of the SAP module. The PS format requires that once data are inserted for each budget item, the total should immediately correspond to the financial coverage. According to our observations, employee simulate the plan in a spreadsheet before including it into the PS system: “In excel we made a simulation of the plan, we made various attempts to best allot the available resources,moving and compensating the vaious costs and after finding the best solution, we said: ‘ok, from the accounting point of view this is ok’. In SAP we can not do this simulation work because it is rigid: we insert only the final version ...” (Line Service Employee using PS).

4.3 From the in-house IT company to SAP external consultancy (20052008) The third locational pattern that has been identified corresponds to a strong presence of the technology supplier external consultancy. From June 2005 in, Dante Province feasibility studies about the introduction of new SAP modules are made by SAP consultants. According to the project responsible, the presence of an external consultant allowed Dante Province to learn from others. At the time the ERP implementation project begun, Dante Province was the first Italian Public Administration carrying on such a project. In 2005, there were many other P.A. implementing their SAP ERP system. Identifying an external consultant became then an opportunity to look outside her own organization to how other organizational contexts acted about similar projects. Looking backwards to the previous period, she told that by doing everything in-house, they also did some errors. The in-house company was strongly conditioned by the Accounting Department. Her role as a project manager at that time was to tell IT people from the in-house company to insist when an adaptation was not feasible. But when the Head of the Accounting Department raised her voice, the IT people did everything she needed, increasing the number of ‘personalizations’. According to her view, the personalization where only good for change management strategic purposes, not for functional reasons. “Instead of encountering a rejection of the system, I can do some personalizations in the spirit that after the user ‘jumps the river’, I will be the rule-maker...” But the in-house company never contributed to this strategic point of view. Her role as project manager was not relevant enough to modify the current state of affairs. IT people from the in-house company were willing to please the users from the Dante Province “those who will pay the bill at the end” and a special role in this respect was that of the Head of the Accounting Department. Contrasts on issues like standardization or personalization of the system between the in-house company and the Head of the Accounting Department were constantly avoided. In order to modify the current situation - that we termed the ‘era of personalizations’

22

- the project manager then decided that the partnership between the in-house company and the technology supplier was to be better exploited by involving technology suppliers consultants. The reason was just that in 1998 there was neither a public administration implementing SAP nor a consultant knowing the public sector as an implementation context. After some years, other administration started to implement SAP. As a consequence, consultants were around having experiences of SAP implementation in the public sector. While direct reuse (e.g. duplication) of software was unlikely, since each local administration had its own regulations, the sharing of experiences (and even of names of reliable consultants) started to be a value for the public administration having implemented SAP. In 2004, a Public Sector User Group has been held at the national level to share implementation experiences. The User Group experiment did not last for long, since the effectiveness to share experience on the information system ws highly undermined by the degree of difference between each local government regulatory framework. However, local experiment of reuse between different organization took place. In the Dante Province case, a consultant that already worked for a neighbour local administration in the implementation of the Material Management module (MM) has been hired for the same purpose. He made an industrial version of the customized module, setting some parameters to be adapted to each single user integration or regulatory requrements. The same consultant worked for two neighbour regional administrations, providing a standard version of the SAP module and the consultancy for adaptation and maintenance in addition to it. At the same time the key users of the two different admnistrations were talking to each other, trying to do the same interpretations of the integration and regulatory requirements and triangulating with a common consultant serving them both.

4.4 From the SAP consultant to the a multi-partner Consortium (2009-) Despite the different advise from the ERP diffusion project manager, in the case of the Real Estate Management module, a consortium made by academic partners, consultants and the in-house company has been appointed to provide consultancy. Consultancy provided directly by the technology supplier is not provided any more. The ERP project manager describes the situation as a pity. According to her advise, the best choice would have been instead to go for the consultant knowing more about the system. Knowing less about the system and not having experience about implementation of that specific module in the public sector, the consortium is described by the interviewee as having made Dante Province loose time during the feasibility study. She describes the situation as a ‘going back’ to the 20022005 period (the ‘era of personalizations’), where many personalizations were performed with the allowance of the in-house IT company. The reason of the choice to go for a consortium instead that for the SAP consultant was made by Dante Province in-house company. The IT in-house company, being the leader of the local IT market, wanted to exploit the same consortium that was built to implement the SAP system in the local University.

23

5. Discussion In this session I discuss the different ages in the implementation biography of the SAP system in Dante Province in term of the locational patterns of the interactions taking place in the design/use space, putting a special focus on the structure of the innovating network, the consitution of the organizations involved and the identities of the actors. The Social Learning in Technological Innovation (SLTI) framework will be applied for data interpretation (Stewart and Hyysalo, 2008). My effort will be to contribute towards the depiction of the evolution of the space of social learning in the biography of ERP projects in Dante Province, presenting a synoptic chart of the locational patterns that took place at different times.

5.1 The technological experiment The first locational pattern (Fig.2) is charaterized by a unity between the project management needs, the technical point of view and the organizational needs. The technical point of view is represented by both the in-house IT consultants and by representatives of the IT department as well. All are represented and work together in the implementation team. Feasibility, functional analysis, programming and implementation are performed in close connection by a group of people working in team. In this pattern, the emerging feature of the technological solution is one of a standardized SAP that represents the “official” source of data. The tasks concerning the particular work of each organizational unit is proposed to be carried on by the operators by using other legacy applications. According to the Social Learning in Technological Innovation (SLTI) framework, this is the typical design/use space configuration of a technological experiment, where users, developers and suppliers (Jaeger et al., 2000; Brown et al.,2003), often deliberately construct into a constituency by certain key players to provide a framework of ideas and resources to shape innovation (Molina, 1995).

24

Fig.2 Technological experiment

The role of the Dante Province as an intermediate organization in this space is a crucial one. At the time were the implementation team was working in Dante Province, there were only private companies or foreign public administration adopting SAP. According to the ERP implementation project manager, learning from others was not an opportunity at that time. Anyhow, it would have been easier to learn from the italian private companies (“Either for companies or public administrations, accounting is a standard package.”) than from foreing public administrations (“...they have an Aglo-Saxon accounting model. We have a latin model. In the Aglo-Saxon model controls are all final. Ours are all estimate”). The space where interactions between design and use take place is smaller here, while other intermediary organizations populate remote areas of the extended countour, with very few interactions with Dante Province.

5.2 The intermediary and proxy user centred design The period that I identified as the ‘era of personalizations’ corresponds to a locational pattern that concerns the activities of a range of users in actually getting the SAP to work after-implementation. The configuration of the interactional space between design and use here is not a temporary ad-hoc configuration, like that of a task force or of the constituency we found at work for the implementation of SAP in Dante Province. It is more like a cross-units institutionalized division of labour.

25

In this period, the project management task, the task to deal with technical aspects (feasibility, functional analysis, design and implementation) and that of gathering the organizational requirement are formally distributed across units in the organization. A relevant role in this newly patterned intermediate space is that of the key user. The key user is an intermediary role carried by an individual that refers to a specific unit in the organization. She has to bridge links between the functional area she is part of and the technical people of the in-house IT company. According to our framework, key users are a kind of “proxy user” and “intermediate user” as well. Key users are “proxy users” in that, while working in close connection with the in-house IT developers, they have to represent both their functional unit requirements and the final users - the Operative Services users - requirements. But key users are also “intermediary users” in that, since developers are not member of the same organization (they are part of an in-house company), they act like an interface of the entire organization (the Dante Province) to an external partner.

Fig.3 Intermediary and proxy user design

In their role of “intermediary users”, key users interact with in-house IT developers by raising issues of regulation deriving from their administrative background and contrast them with issues of technical feasibility coming from the in-house IT developers. Whenever a regulation mandates an exception in the procedure, key user strived to get the exception implemented in the system. There is not an overall IT Department control over key users decisions. This control intervenes only for software enhancement projects that are larger than 10 working days. In the ordinary management, there is not any formal decision-making process that foresees

26

a check of the IT department over key user decisions. Since contrasts between key user requirements and technical feasibility issues are constantly avoided by the IT developers in their role of in-house suppliers, the absence of an internal intermediation by a competent IT department representative resulted in the straightforward approval of any requirement key user would have presented in that period. This is especially relevant as key users do not take into account the system development drawbacks of their requests. A condition that intensifies a tendency of this pattern to produce ‘personalizations’ is the move of the ERP project manager from the IT department to the Accounting Department. The ERP project manager is no longer perceived by the key users she has to negotiate with as an IT person in staff to them. Rather, she becomes perceived as holding the stake of the Accounting Department. Key users will be resistant to accept the mediation of a project manager coming from a sector-specific functional division. The in-house IT company to its turn has to negotiate the realization of the system with the Accounting Department instead of with the IT Department, with consequent priorities assigned to accounting aspects of the system. As mentioned before, key users are also “proxy users” in that while working in close connection with the developers they have to represent both their requirements (Staff services requirements) and the final users - the Operative Services users - requirements.

5.3 Innofusion The third locational pattern of the implementation biography of SAP in Dante Province can be interpreted as a space of innofusion (Fleck, 1988). The innovation does not come any more from a single identifiable co-located source, e.g. from the interaction between Dante Province and its in-house IT company through a small number of key users. Rather, innovation takes place in a multi-sited and long lasting intermitted interaction between global producers and users from different public organizations. The ERP software companies in this space make initial customizations to systems built for other sectors or for other user organizations. Modifications made in-site by the consultants with the user IT staff become later incorporated as part of supplier’s generic package, while some others become discarded and kept up only locally. In this space, consultants become the key intermediary, not only between production and use, but also, and most importantly, between different user sites. Being closer to the production side, consultants are perceived by the Dante Province ERP project manager as providing more reliable feasibility studies than the in-house IT company people that do not have a comparable expertise on the specific topic, and are prone to accept any requirement from their customer. Being able to travel from an user site to another, consultants also act as intermediaries between different user organizations, in ways that overtake internal struggles between competing ”proxy users”, allowing re-use of solutions. But together with the major role of the consultants as intermediaries, user organizations innovate their intermediating space as well. Public organizations can employ more indirect forms in order to impose their standards over other public administrations through the consultants. Being first first adopters of new modules becomes to be perceived as a value instead of a problem, like it was at the begin-

27

ning. It allows to have consultants build in the software modules their procedures as a best practice for the public sector at large. Being these the major features of production space, from the use side of the space this locational pattern corresponds to a transition of collaboration practices from a direct form to a more indirect one, both in term of contents and of means: key users of different public administrations start to exchange references of consultants they like and talk to each about the consultancy taking place, in order to try to do similar interpretations and to minimize the initial costs of consultancy. The content of the collaboration is both the reuse of the code and the comparison of performace of the consultant, in term of contracts as well as of less formal aspects of the relation. Public Sector user groups take place not only to exchange experiences of technological procurement but also of consultancy procurement on technology procurement, adaptation and maintenance.

5.4 Generification The fourth pattern corresponds in Dante Province case to the strategic objective to enhance the systemic relations between different local public administrations, both from the production side and from the use side, in order to allow organizational information systems as well as organizational best practices concerning accounting, procurement, human resource management to travel from an organization to another in the region. From the production side, the idea to move from consultancy provided by the technology supplier to a Consortium that includes the local inhouse IT company and other, more generic, consultancy companies corresponds to the idea that best practices have to be formed, exchanged and reused locally by different public administrations in different domains (health care, education, energy, etc...). From the use side, the idea is to bring the diffusion of SAP within the strategic goal of aligning a fragmented landscape of public sector organizations made of large organizations and smaller ones to common interpretations of organizational procedures.

6. Conclusions My interpretation of the data allows to depict spaces of intermediation both from the production side and from the appropriation end. Each space affords access to specific resources and prevent access to some others and give centrality to different intermediations. Each period description enphasizes different layers of granularity of the problem. In the first space, the unity of project management, technical and organizational point of view allowed an interpretation of SAP as a standard system, with fewer personalizations. At the same time, learning was limited to a small group of people, without the opportunity to design the system for specific uses. Design for specific uses was not even perceived in the production space as an opportunity leading to best practices formulation and reuse. In the second timespace configuration, that I described as a “proxy/intermediary user centred design” the presence of more intermediaries from the appropriation end allowed for a more

28

personal interpretation of the SAP system, although never leading to an effective inclusion of Operative Service user requirements. At the same time, a shallower intermediary configuration was emerging in the production end in which the distance of the IT Department from the resources for decision making on SAP implementation - delegating IT decisions to the in-house IT company in connection with few sector-specific key users; moving the ERP project manager to the accounting department - set the conditions for a loss of control on the technical consequences of the personalizations. The third locational pattern has been identified with a more distributed provision of innovation, where the consultants started to play a central role in the space of interaction between design and use of the SAP system. In the adoption end, an escalation to indirectness has been noticed. The content key users relations moved from issues related to technology procurement to issues of procurement of consultancy for technology procurement. Intermediaries of the adoption end surrendered their space to intermediaries from the production end. With this escalation to indirect decision making, the role of top management becomes as relevant as never before: controlling the expenditures on consultancy on technology procurement is an issue of their own. In continuity with this move to IT decision as strategic decisions - with political visibility as well, in the fourth period a Consortium of consultants becomes the key actor shaping the space of the network and the nature of the technological solution. The SAP system is enacted in this space as a resource for local development. In each space domain the different granularity of the problem description produces an umbalance towards some aspects of the problems connected with the adoption of the technological solution. In the first space, the focus has been on the micro level of the implementation team interactions, focussing on the technical design and on the project management needs to role the system out, without too much attention for the involvement of the production end and of the use end intermediaries. In the second time, descriptive terms employed by interviewee allow to segment in detail the use end, while the production end becomes peripheral and out of control. The third space stresses the innovation of the production end, with the outsourcing to a SAP consultant being perceived as the solution to the loss of control on the production space. Management started to be involved, since the issue of external consultancy started to raise as political issues. As a consequence, the fourth space focuses on the benefit the region can profit from the adoption of complex organizational systems like ERP systems. While keeping the evolution of the overall social learning taking place is a difficult task to be managed, focussing on a single corner of the design/use interactional space can mislead implementation and post-implementation strategies. The analysis provided here wants to represent an intial contribution towards the depiction of the evolution of the space of social learning in the biography of ERP projects, presenting a synoptic chart of the locational patterns that took place at different times.

References 1. Almkov, P.G., (2008), ‘Standardized Data and Singular Situations’, Social

29

Studies of Science, 38, 6: 873-897. 2. Amin, A. & Cohendet, P. (2005). Geographies of Knowledge Formation in Firms. Industry and Innovation, 12(4), 465-486. 3. Bandura, A. (1977). Social Learning Theory. Englewood Cliffs, N.J: PrenticeHall. 4. Bateson, G. (2002 [1979]). Mind and Nature: A Necessary Unity. Cresskill, NJ: Hampton Press. 5. Boland, R. J. Jr. (2001). The tiranny of space in organizational analysis. Information and Organization, 11, 3-23. 6. Brady, T., Tierney, M., and Williams, R. (1992) ‘The Commodification of Industry-application Software’, Industrial and Corporate Change, 1, 3: 489-514. 7. Brehm, L., Heinzl, A., Markus, L. (2001) ‘Tailoring ERP Systems: A Spectrum of Choices and their Implications’, 34th Annual Hawaii International Conference of System Sciences (HICSS-34), 8: 8017. 8. Brown, J., S., & Duguid P. (2001). Knowledge and organization: A socialpractice perspective. Organization Science, 12(2), 198-213. 9. Brown, H.S., Vergragt P., Green K. and Berchicci L. (2003). Learning for sustainability transition through bounded socio-technical experiments in personal mobility. Technology Analysis & Strategic Management, 15(3), 291–315. 10. Bruegger U. and Knorr-Cetina K. (2002) Global microstructures: the virtual societies of financial markets. American Journal of Sociology vol. 107 (4) pp. 905-950. 11. Cardano, M. (2007). Tecniche di ricerca qualitativa. Percorsi di ricerca nelle scienze sociali. Carocci Editore, Roma. 12. Clausen. C. and Koch, C. (1999) ‘The Role of Spaces and Occasions in the Transformation of Information Technologies: Lessons from the Social Shaping of IT Systems for Manufacturing in a Danish Context’, Technology Analysis and Strategic Management, 11, 3: 463-82. 13. Clemmons, S. and Simon, S. (2001) ‘ Control and Coordination in Global ERP Configuration’, Business Process Management Journal, 7, 3: 205-15. 14. Czarniawska, B. (1998). A narrative approach to organization studies. London: Sage. 15. Davenport, T. (1998) ‘Putting the Enterprise into the Enterprise System’, Harvard Business Review, 76, 4: 121-32. 16. Di Maggio, P. (2001) "Introduction: Making Sense of the Contemporary Firm and Prefiguring Its Future." Pp. 3-30 in The 21st Century Firm, edited by Paul DiMaggio. Princeton, N.J.: Princeton University Press. 17. Fincham, R., Fleck, J., Procter, R., Scarbrough, H., Tierney, M., Williams, R. (1994) Expertise and Innovation, Oxford,: Clarendon Press. 18. Fleck, J., Webster, J., Williams, R. (1990), ‘Dynamics of Information Technology Implementation: a Reassessment of paradigms and trajectoris of development ’, Futures, July-August 1990. 19. Fleck, J. (1988). Innofusion or diffusation? The nature of technological development in robotics, Edinburgh PICT Working Paper No. 7, Edinburgh University. 20. Fligstein, N. (1990). The Transformation of Corporate Control. Cambridge,

30

Mass.: Harvard University Press. 21. Giddens, A. (1990) The consequences of modernity. Cambridge University Press, Cambridge. 22. Granovetter, M. (1985) Economic action and social structure: The problem of embeddedness. American Journal of Sociology, 91(3), 481-510. 23. Guterson J. (1997) Studying Up Revisited. Political and Legal Anthropology Review (PoLAR) 21: 114-9. 24. Harries-Jones, Peter (1995). A Recursive Vision: Ecological Understanding and Gregory Bateson. Toronto: University of Toronto Press. 25. Harris, S. (1998). Long-Distance Corporations, Big Sciences, and the Geography of Knowledge. Configurations, 6(2), 269-304. 26. Harvey, D. (1989) The condition of postmodernity. Blackwell, Oxford. 27. Hasu, M. (2001). Critical Transition from Developers to Users. Academic Dissertation. Helsinki: University of Helsinki, Department of Education. 28. Hislop D. (2002). The client role in consultancy relations during the appropriation of technological innovations. Research Policy vol. 31 (5) pp. 657-671. 29. Hoffan, A. (2004) Reconsidering the Role of the Practical Theorist: On (re-) connecting theory to practice in Organizational Theory. Strategic Organization 2: 213-222. 30. Hyysalo, S (2006). The role of learning-by-using in the design of healthcare technologies: a case study. The Information Society, 22(2), 89–100. 31. Jaeger, B, Slack R. and Williams R. (2000). Europe experiments with multimedia: an overview of social experiments and trials. The Information Society, 16(4), 277–302. 32. Jin, L., Robey, D. (2008). Bridging social and technical interfaces in organizations: An interpretive analysis of time-space distanciation. Information & Organization, 18, 177-204. 33. Kumar, K., Maheshwari, U. and Kumar, U. (2003) ‘An Investigation of Critical Management Issues in ERP Implementation: Empirical Evidence from Canadian Organizations’ Technovation, 23: 793-807. 34. Latour, B. (1992) Where are the missing masses? The sociology of a few mundane artifacts. In Shaping technology/building society: studies in sociotechnical change, eds. W.Bijker and J. Law, pp. 225-264. MIT Press, London. 35. Leyshon, A. (1995) ‘Annihilating space?: the speed-up of communcations. In A shrinking world?’ Global uneveness and inequality, eds. J. Allen and C.Hamnett, pp.11-54. Open University Press/Oxford University Press, Oxford. 36. Liang, H., Xue, Y., Boulton, W. and Byrd, T. (2004) ‘Why Wstern vendors Don’t Dominate China’s ERP Market: Examining Cases of Failed ERP System Implementation in China and Explaining the Unique Circumstances’, Communication of the ACM, 47, 7: 69-72. 37. Light, B., Holland, C., Wills, K. (2001) ‘ERP and Best of Breed: A Comparative Analysis’, Business Process Management Journal, 7, 3: 216-24. 38. Lundvall, B-Å. & Johanson, B. (1994). The learning economy. Journal of Industry Studies, 1(2), 23–42. 39. Mabert, V., Soni, A., Venkataramanan, M. (2001) ‘Enterprise Resource

31

40. 41. 42. 43. 44.

45. 46. 47.

48. 49. 50. 51. 52. 53. 54.

55.

56. 57.

58.

Planning: Common Myths Versus Evolving Reality’, Business Horizons, May-June: 69-76. MacKenzie, D., & Wajcman, J. (eds.) (1999). The Social Shaping of Technology, 2nd Ed. Buckingham: Open University Press. Marcus GE. (1995) Ethnography in/of the World System: The Emergence of Multi-sited Ethnography. Annual Review of Anthropology, 24:95-117. Massey, D. (1991) A global sense of place. Marxism Today, June 24(9). Misak, C. (2004). The Cambridge Companion to Peirce. Cambridge: Cambridge University Press. Molina, A. (1995). Sociotechnical constituencies as processes of alignment: the rise of a large-scale European information technology initiative. Technology and Society, 17(4). Molina, A. (1989a), The Social Basis of the Microelectronics Revolution, Edimburgh University Press, Edinburgh. Murdoch, J. (1998). The Spaces of Actor-Network Theory. Geoforum, 29(4), 357-374. Newman M. and Westrup C. (2005) Making ERPs work: accountants and the introduction of ERP systems. European Journal of Information Systems vol. 14 (3) pp. 258-272 Nicolini, D. The work to make telemedicine work: A social and articulative view. Social Science & Medicine, 2006, 62, 2754–67. Nicolini, D., Gherardi, S. & Yanow, D. (Eds) Knowing in practice. Armonk, NY: ME Sharpe, 2003. Peirce, C., S. (1958) Collected Papers, vol. 5, Cambridge, MA: Harvard University Press. Panteli, N. (2004). Discoursive articulations of presence in virtual organizing. Information and Organization, 14, 59-81. Pentland, B. (1992), ‘Organizing Moves in Software Support Hot Lines’, Administrative Science Quarterly, 37, 4: 527-48. Pollock, N., Cornford, J. (2004) ‘ERP Systems and the University as an ‘Unique’ Organization’’, Information Technology and People, 17: 1:31-52. Quattrone P. and Hopper T. (2006) What is IT? SAP, accounting, and visibility in a multinational organisation. Information and Organizationvol. 16 (3) pp. 212-250 Richmond, W.B., Nelson, P., and Misra, S. (2006) ‘An Empirical Analysis of Software Life Spans to Determine the Planning Horizon for New Software’, Information Technology and Management, 7, 2: 131-49. Rusaw, A., C., (2007). Changing Public Organizations: Four Approaches. International Journal of Public Administration, 30(3), 347-361. Russell, S. and Williams R. (2002). Concepts, spaces and tools for action? Exploring the policy potential of the social shaping perspective. In Shaping Technology, Guiding Policy: Concepts, Spaces and Tools, K Sørensen and R Williams (eds.), pp. 133–154. Russell, S. and Williams, R. (1988). Opening the black-box and closing it behind you: on micro-sociology in the social analysis of technology. Edinburgh PICT Working Paper No. 3, Edinburgh University.

32

59. Sahay, S. (1998). Implementing GIS technology in India: some issues of time and space. Accounting, Management and Information Technology, 8, 147-189. 60. Sawyer S.(2000). Packaged software: implications of the differences from custom approaches to software development. European Journal of Information Systems vol. 9 (1) pp. 47-58 61. Schön, D., A., (1983) The reflective practitioner - how professionals think in action. Basic Books, 1983. 62. Soh, C., Kien, S.S. and Tay-Yap, J. (2000) ‘Cultural Fits and Misfits: Is ERP a Universal Solution?’ Communications of the ACM, 43, 4: 47-51. 63. Somers, T. and Nelson, K. (2004) ‘A Taxonomy of Players and Activities Across the ERP Project Life Cycle’, Information and Management, 41: 257-78. 64. Sørensen, K.H., and Williams, R., (edited by) (2002). Shaping Technology, Guiding Policy: Concepts, Spaces, and Tools. Cheltenham, UK and Northhampton, MA (USA): Edward Elgar. 65. Sprott, D. (2000) ‘Enterprise Resource Planning: Componentizing the Enterprise Application Packages Tomorrow’s Customers Will Demand the Ability to Buy, Reusem and Build their Competitive-edge Solutions: Can the Package Vendors Adapt in Time?’, Communications of the ACM, 43, 4: 63-9. 66. Stewart J. and Hyysalo S. (2008) Intermediaries, Users And Social Learning In Technological Innovation. International Journal of Innovation Management, vol. 12 (3) pp. 295-325. 67. Strauss, A.L. and Corbin, J. (1998) Basics of Qualitative Research: Techniques and procedures for developing Grounded Theory. Thousand Oaks, CA: Sage. 68. Tsoukas, H. (1992). Wyas of sseing: Topographic and Network Representations in Organization Theory. Systems Practice, 5(4), 441-456. 69. Van Maanen, J. (1998) Tales of the Field: On writing ethnography. Chicago: University of Chicago Press. 70. Wagner E. and Newell S. (2004) “Best for whom”: the tension between ‘best practice’ ERP packages and diverse epistemic cultures in a university context. The Journal of Strategic Information Systems, vol. 13 (4), 305-328. 71. Walsham, G. (1993). Interpreting information systems in organizations. Chichester, UK: Wiley. 72. Wei, H., Li, C., Wang, E., Li, C., Ju, P. (2005) ‘Understanding Misalignment and Cascading Change of ERP Implementation: A Stage View of Process Analysis’, European Journal of Information Systems, 14, 4: 324-34. 73. Williams, R, Slack R. and Stewart J. (2005). Social Learning in Technological Innovation — Experimenting with Information and Communication Technologies. Cheltenham: Edgar Elgar Publishing. 74. Williams, R. & Edge, D. (1996). The social shaping of technology. Research Policy, 25, 856–899. 75. Yin, R.K. (2003). Case study research: design and methods. Thousand Oaks, CA, Sage Publications. 76. Zolin, R., Hinds, P., J., Fruchter, R., Levitt, R., E. (2004). Interpersonal trust in cross-functional, geographically distributed work: A longitudinal study. Information and Organization 14, 1-26.

33