How Sap Conquered The World

Draft of Introduction from: Neil Pollock and Robin Williams (2009) Software and Organizations: The Biography of the Enterprise Solution Or How SAP Conquered the World, London, Routledge.

i

Introduction: The Reshaping of the Modern Enterprise Solution November 1990: The Hague. The UK Science and Engineering Research Council (SERC) sent one of its leading experts to an international workshop organised to discuss the future of the computer systems used to run industrial enterprises. The workshop was one of a number organised in Europe and the USA that year to assess the prospects for these technologies that were seen as constituting ‘best practice’ in manufacturing organisations and crucial for industrial competitiveness.

The workshop, organised by The Eindhoven Group, widely regarded as the ‘leading research group in Europe’ on these technologies, attracted a strong and interdisciplinary turnout, with over sixty consultants, technology vendors, users and academics, coming together to discuss its provocative ‘rationale document’. Gerry Waterlow (consultant to SERC’s Application of Computers and Manufacturing and Engineering Directorate) circulated a report, drawing attention to the consensus that appeared to have been reached around the central argument advanced by this document. He suggested that these conclusions could probably be regarded as a ‘reasonable snapshot’ of the direction in which the particular technology they were all there to discuss was moving. It was even suggested that – such was the consensus - the workshop itself might help to underwrite this future direction since many of the actors central to its shaping were present in the room.

The technology under the microscope was the state of the art of what we today would call Enterprise Resource Planning solutions – known then as Manufacturing Resource Planning (MRP) systems. In the late 1980s, MRP technology had been heavily promoted as a solution suitable for a wide range of organisations. However the title of the workshop - ‘Beyond MRP: MRP and the Future of Standard Software for Production Planning and Control’ – made it clear something was afoot. One did not have to read too far into the workshop rationale document to see the sting:

1

The development of MRP (I and II) has led […] to a specific production control philosophy [as well] as to standard software for production control. Control philosophy and standard software are heavily intertwined. Having standard software for production control is very important in practice, as well with respect to the whole implementation and training process as with respect to maintenance. On the other hand MRP (philosophy and software) seems not to fit well everywhere (Workshop Rationale Document, no page no.).

The workshop had been motivated by growing concerns that this latest breed of enterprise system was proving problematic. Users, it seems, found these systems ‘difficult to apply’ and as a result, they were ‘not widely adopted’. Some of the difficulties experienced concerned their ‘generic’ nature and it was generally perceived that the processes embedded in the software were ‘too rigid’ for most adopting organisations. Indeed, The Eindhoven Group saw the workshop as a means to ‘debate the reasons’ as to why this was, as well as to ‘identify ways forwards’. They concluded that it was ‘…time to discuss the future of standard software in general and more specifically the future of MRP’. The workshop put forward ‘three scenarios for MRP development:

i)

gradual evolution of generalised MRP with the existing software suppliers remaining the major vendors;

ii)

increase in user-driven special versions of MRP for particular industries, leading to partnerships between users and smaller suppliers concentrating on vertical markets;

iii)

decline in significance of MRP [to be replaced by] Factory management systems, supplied by system integrators with a broad range of skills (systems, software, communications, automation) [which] will take over MRP2 functions’ (Waterlow 1990: 2).

2

The conference background paper (Wijngaard 1990: 5) described the latter as the 'more radical scenario, and one for which there is substantial evidence that new ideas are emerging from outside the MRP world’. Concepts such as Just-In-Time and Computer-Integrated-Manufacture would, it argued, be ‘captured better by Factory Management Systems rather than MRP’. Moreover:

[t]he special needs of industry sectors cannot be met by a generic MRP system, and different methods will emerge. These developments are being made today partly by a new group of ‘systems integrators’ who have stronger technical skills in systems, communication, automation, and new software technologies. In this scenario the structure of the software industry is likely to change as new suppliers appear (ibid.: 5).

During the workshop, there was the general feeling that the current direction of the enterprise system was not sustainable. ‘The majority [present at the conference] considered MRP2 in the form of standard software as an unworkable concept’ (ibid.: 4). The future as they saw it was not with generic software packages but instead there was an ‘urgent need’ for alternative more ‘context specific software packages’ (ibid.: 6). Intriguingly, the workshop deliberations showed little awareness that waiting just around the corner was a new breed of software supplier that would herald in a very different future for the enterprise system…

The modern enterprise-wide information system has become a generic software package. A small number of software suppliers, it seems, of which the German-based software company SAP is the clear leader, have succeeded in deploying their Enterprise Resource Planning (ERP) systems within and across many different organisations, sectors and countries around the globe. Large corporations and organisations throughout the world now appear to be dominated by a new breed of standardised packaged solution. How has this happened? Indeed the fact it has happened at all is remarkable when one considers only two decades ago leading experts and practitioners agreed that the future for organisational information systems was not with generic IT solutions. Back then, and based on experience with the state3

of-the-art integrated enterprise planning solutions of the day, many experts considered it highly unlikely that a small number of generic information systems could meet the needs of organisations within and across sectors (Waterlow 1990). These systems were seen as too standardised for the complex and diverse needs of adopting organisations (which, as they saw it, required alternative and more flexible, locally specific kinds of solutions). Thus, the future for technology supply was seen to lie with vendors developing varieties of sector specific offerings that could be locally adapted to the various particular user organisations seeking to apply them. Discussions of sectoral difference and organisational uniqueness were the order of the day and ‘semi-generic’ and highly tailorable packages were seen to be the way forward.

However, whilst the ‘structure of the software industry’ has changed this is not in the way predicted by The Eindhoven Group. Despite their assessment at the 1990s workshop, a new breed of software package supplier has emerged which has managed to reuse and recycle highly standardised systems into thousands of different organisations. These packaged solutions now make up a substantial part, perhaps the majority of organisational IT expenditure (Jakovljevic 2001) and include, as well as ERP, Customer Relationship Management (CRM), Supply Chain Management (SCM) and other financial and administrative systems. The extension of the generic package into organisations worldwide is all the more remarkable when one considers that not only was the phenomenal success of suppliers like SAP ‘not on the radar’ during the workshop but the vision for the direction of these systems was far removed from what we have today. However, before the end of that decade, SAP’s now famous ‘R3’ package, followed by other suppliers of similar generic software solutions that have became known as Enterprise Resource Planning systems, would have swept through major corporations in Europe, the USA and beyond, moving out from manufacturing into services and the public sector. This poses the following questions: How has this happened? How has this new breed of supplier been able to extend their systems into organisations worldwide? Moreover, what does this mean for the character of the organisational information system and wider arena in which they are situated?

4

What is clear from the rise of these kinds of solutions is that the nature of the modern enterprise system is changing. Not only have these new suppliers recycled their technologies into many different places but, arguably, in doing so, they have heralded a shift in the conception of the organisational information system. What is at stake is a profound change in ideas about the very notion of the modern corporate solution: this encompasses how they should be developed and implemented as well as the extent to which they should address particular sectoral and organisational requirements. Clearly, these new kinds of systems have important implications for researchers interested in the technology and organisation relationship. How are we to respond to the rise of this new breed of software supplier and the extension of the generic enterprise system in a sensible and comprehensive way (i.e. without either inflating or reducing these changes)?

However, despite the fact these kinds of standardised packaged solutions account for the bulk of systems used today, we cannot in a conceptually and empirically robust manner explain their rise to prominence. We do not know precisely how the modern corporate system became a generic package. Though practitioners may advance wellrehearsed ‘potted-histories’ of this artefact there are very few studies of the origination and design of these artefacts, let alone research which addresses the evolution of this technology along its protracted lifecycle. These kinds of IT systems have had nowhere near the kind of sociological attention they deserve. Why is this?

One reason is that the received wisdom amongst many scholars interested in the social study of technology would be that generic solutions only have limited applicability: for some, there is no such thing as a ‘universal’ or ‘one size fits all’ solution (Star & Ruhleder 1996; Hanseth & Braa 2001). Standard systems only work to the extent they are adapted by user organisations through messy localisation processes. Thus, according to many sociologists it is ‘users’ and ‘adopting organisations’ that should be studied. Whilst, on the one hand, we share their interests in implementation it has also meant that, on the other, recent research on information systems has become somewhat unbalanced. In focusing principally upon user organisations, social scientists have not adequately conceptualised and analysed standardised information solutions. There is not, for instance, a comprehensive understanding of the inner workings of the leading software supplier organisations. Nor do we have a 5

sophisticated appreciation of the wider information system industry dynamics that surrounds software producers. This is reflected by the lack of frameworks that explain the extension of these systems across sectors and this wider shift from specialised to generic software. Let us briefly look at some of the dominant ways these technologies are researched.

The current social science research on packaged enterprise systems is broadly gathered around two opposing poles. The first, typified by more managerially focused kinds of analysis, views ERP systems and the like as more or less ‘transformatory’ technologies containing ‘universal logics’. They imply that because of the nature of their design these systems can be applied extensively across all kinds of corporations and bring about widespread change (see for instance O’Leary [2000] and Bendoly and Jacobs [2005]). Not surprisingly, this view has been seen as problematic by critical social scientists. Thus, a second pole has subjected these discourses of transformation and universalism to critical assessment. In what might broadly be characterised as the ‘Social Study of Information Systems’ (McLaughlin et al 1999; Ciborra 2000; Walsham 2001; Sawyer 2000; Avgerou 2002) scholars have advanced alternative accounts of the spread of these solutions. Many have produced what might be described as ‘situated’ and ‘localist’ explanations, often drawing on the groundbreaking work of Suchman (1987) as well as ethnographic study (see for instance Knox et al. [2005]). These accounts typically contrast the uniqueness in structure and practices of user organisations with the standardisation of packaged solutions, and have tended to emphasise the ‘contingency’ surrounding the implementation of these systems (Hanseth et al. 2001).

However, whilst this literature is highly informative, it also tells us rather little about what we regard as one of the most important developments in the short history of corporate information systems: the shift from locally specific to generic systems. Scholars in the Social Study of Information Systems, for instance, have thus focussed selectively upon certain aspects and moments of the software package life cycle and as a result, they offer what have now become well-rehearsed but also highly partial accounts. Critical social science should be able to give a more comprehensive analysis of the reshaping of the modern corporate system. Not to do so has risks - handing the

6

terrain to other disciplines.1 This has meant that the debate around enterprise systems has been unevenly developed and unhelpfully fragmented between rather narrow (e.g. managerial or technical) perspectives. Of course, every failing is also an opportunity, and the gap that exists in our current understanding is also one we hope to fill (at least in part) with this book.

RHETORICS OF TECHNOLOGY SUPPLY Today, few can deny that packaged highly standardised forms of enterprise solutions have become an important feature of our organisational landscape. In this respect, Management scholars have been prolific in celebrating their various features and characteristics. Daniel O’Leary (2000), for instance, goes as far as describing systems like ERP as nothing less than a ‘corporate marvel’. They have undoubtedly had an enormous influence on the business and information system worlds, he argues, affecting each of the following dimensions. They have experienced a huge market growth, being taken-up by most of the major corporations around the world. They are also now increasingly being rolled out within small and medium-sized enterprises. Moreover, within corporations they have been used as one of the primary tools for reengineering the organisation as well as the diffusion of many best practices.

Added to this, there is also the (mostly implicit) assumption that they have heralded in a new class of computer solution (Klaus et al. 2000). This, firstly, is the suggestion that the ‘generic-ness’ of these solutions is an achievable design issue (Carey 1998). In addition, that these solutions can be ‘recycled’ across similar classes of organisations (Deifel 1999). This can be within the same or related industrial sector or as is now increasingly common across different and unrelated sectors and organisational forms. Secondly, and in stark contrast to the organisational information systems that went before, these systems are now generally thought to behave like ‘products’ that can be selected and purchased as with other kinds of commodities (Deifel 1999; Heiskanen et al. 2000; Regnell et al. 2001; Xu and Brinkkemper 2005).

There are of course aspects of this brief account that deserve to be challenged. One of which is that enterprise systems were not simply borne ‘software packages’ or ‘generic solutions’. Rather, something had to be done to them to achieve their ‘generic-ness’ and ‘commodity’ status. It is notable that Management research 7

provides very little, whether in terms of empirical findings or conceptual frameworks, that will afford an adequate understanding of this ‘something’. This is true amongst even the more highly regarded of works – such as Davenport (2000b). In terms of the first point (‘generic-ness’) this literature tells us almost nothing about how the suppliers design and develop these systems and products or on what they base the design of generic solutions? We understand very little about how different suppliers manage the tension between designing systems for a specific user and for a wider market. This is important whether a package is being redesigned from a generic to a niche specific solution or whether it is being recycled from one sector to another or ‘upgraded’ from custom built software to a generic system. In terms of the second point (‘commodity status’) we have little understanding of just how software packages are typically presented to potential adopters. This includes the different strategies and decision-making processes of those adopting software packages (the process by which users assess and make sense of the wide range of alternatives and options available). This is whether to procure one of the more generic of systems on offer or a more flexible ERP alternative. On one hand, it is acknowledged that organisations find it difficult to critically assess and evaluate the range of packages on offer. Moreover, there is a growing awareness of the costs of ending up with the ‘wrong’ solution and this is provoking uncertainty among user organisations. Yet, on the other, this appears to have done little to deter the uptake of packages. Other considerations and actors are obviously at work here.

One of the other problems of the Management literature (including Davenport) is that it tends to be based on a particularly weak empirical base. Rather than study actual technologies, these writers tend simply to align themselves with the statements and rhetorics of technology supply. We therefore turn to other disciplines within the social sciences, where these criticisms apply less, and where there are numerous frameworks available to trace this development (although none appear fined tuned enough to analyse the ‘biography’ of the generic solution in the way we think necessary). And which over recent years they have amassed an enormous amount of qualitative and particularly ethnographic research data.

8

REACTION AGAINST PACKAGED SOLUTIONS The Social Study of Information Systems is made up of work from scholars within Science & Technology Studies (STS), Information Systems (IS) research and Organisation Studies.2 In the face of often quite deterministic and supplier dominated debates, researchers from these approaches were among the first to characterise the complexities and difficulties associated with modern packaged information systems (see Lucas et al. 1988; Webster and Williams 1993; Salzman and Rosenthal 1994). Much of this ‘critical project’ has grown up in opposition to the more dominant supply side accounts. Thus, it is no surprise this work predominately focuses on the struggles adopting organisations engage in whilst attempting to make generic and standard systems work within their user settings. There have been many studies now showing how packaged systems seldom translate (or translate easily) across boundaries, whether these are between organisations within the same sector, between industrial sectors, or between public and private sector ones (Pollock & Cornford 2004). The difficulties in developing solutions that can be widely applied result, it is commonly argued, from the diversity of organisational settings and the resultant gulf that exists between the system and the specific contexts, practices and requirements of particular user organisations. Indeed if generic packages do work across settings, this would, under these perspectives, be seen to be only at great expense to the adopting organisation (in terms of adapting the package and prejudicing the benefits of standard solutions or imposing unwanted organisational change in order to meet presumptions built into the package). Indeed there is now a large (and rather interesting) literature on cases of failure, implementation difficulties, and on the costs and risks associated with adopting these systems (Scott and Vessey 2002; Newman and Westrup 2005; Wagner and Newell 2006).

Yet, if we are to view this literature a little more critically, when reading some of these studies it is a wonder these systems extend at all. The latter sets of arguments are pursued with such vigour – there is such a desire it seems to demonstrate the complex organisational and technical reworkings necessary to sustain packaged software – that there appears to be an entrenched scepticism with regard to their wider applicability (Soh et al. 2000; Scott and Wagner 2003; Soh and Sia 2004). For many social scientists, especially those informed by sociology and anthropology, the large software suppliers like SAP should not be successful. Sociological/anthropological 9

theory tells them that organisations are too diverse to deploy these highly generic kinds of systems. Many studies therefore end up suggesting - based on difficulties and complications witnessed during fieldwork - that ERP systems and the like have no more than limited potential for extension. How could the same or similar (or even slightly adapted) organisational IT system be applied across many different types of organisations with all the diversity and heterogeneity found there (Soh et al. 2000; Soh and Sia 2004)? In addition, to support these assumptions, we are introduced to various explanations (and a growing vocabulary) as to why these systems should not work across settings.3 Related to this, many accounts of package implementation describe how these solutions if not completely failing appear constantly on the brink of failure (in this respect see Constant’s [1999] critique of tendency within STS to over emphasise technological breakdowns). In short, it appears that within the Social Study of Information Systems there are (implicitly) a powerful set of objections advanced towards generic enterprise solutions. Sociologists interested in the technology and organisation relationship have looked in one direction only, through what we would argue is an inappropriate theoretical lens, studying reworkings rather than extensions, with the upshot that many of their assessments now sound rather reminiscent of The Eindhoven Group.

Although this work was a useful corrective to the more dominant supply side view, today, it now appears incomplete. It is advancing a fairly reductive analytical schema (and one it must be said that tends to produce rather familiar sounding stories). If we are to answer the questions set out above, which as scholars interested in the social analysis of technology we must, then the current approach by itself is no longer sufficient. We ourselves have been ‘guilty’, if this is the right word, of advancing such arguments. Williams, for instance, strongly concurred with the consensus back in the 1990s that the future was not with generic packages. Likewise Pollock problematised the transfer of ERP within the public sector suggesting, based on observations during ethnographic fieldwork, that these systems would be ineffective for these very different kinds of organisations. We both presented what were accurate ‘snapshots’ at that point in time but also ones that some years later, and with the luxury of resources/time to revisit fieldwork settings, we realise were very much only a ‘partial picture’.

10

One of the reasons why the social analysis of technology generally and the Social Study of Information Systems specifically has not adequately conceptualised and studied these technologies is intellectually deep rooted. Scholars interested in the technology and organisation relationship have been highly influenced by constructivist, interactionist and, especially more recently, Actor Network Theory (ANT) accounts. However, we would argue these are not adequate to study packaged systems - especially in light of their increased commodification, globalisation and ‘generification’. Our main point of contention is the emphasis these approaches bring towards local ethnographic studies and micro-sociological concepts (but see KnorrCetina and Bruegger [2002] who develop micro-sociological concepts for studying global phenomena). Whilst highly effective in producing rich local pictures, that is to say capturing the various struggles and choices around the design or (more frequently) the implementation of new technologies, they also tend to provide a rather reductionist form of analysis.

Certain kinds of study and situation have become the norm in our discipline (and given undue emphasis). This is the ‘ERP implementation study’ and within this attention is given to immediate action and ‘heroic’ local actors, for instance, who appear able to create and recreate their organisational world almost from scratch (a form of work that often includes the large-scale reworking of the newly implemented information system – see for instance Scott and Wagner [2003]). However we are not wholly convinced that most useful way to study these artefacts is solely at the place where the user encounters them (Kallinikos 2004a,b). This view is inherently problematic when one considers that with software packages we are dealing with a technology that was developed at some distance from it place of use. In this respect, the emphasis on local studies of adoption offers an inadequate lens for exploring the development and influence of complex organisational technologies like ERP (that exhibit very different dynamics to traditional software systems or, for that matter, other kinds of organisational technologies). We do not think it sufficient to analyse the extension of the software package by reducing explanations to simply local action and contingency or in terms of the victory of the local over the global – or vice versa. Existing studies both downplay the influence of technology supply and often overlook the influence of the broader historical setting on the unfolding of the technology.

11

Thus, we argue that we need different approaches to explain the rise of the modern enterprise system.

Indeed the lack of research around the topic is one of the reasons we have chosen to subtitle this book in this provocative manner (‘How SAP conquered the world’). Let us briefly explain. Firstly, this is not a book specifically about the large global supplier SAP (it is not the ‘history of SAP’ though such a study is long overdue) but a more general account of the new breed of software produced that has recently emerged. SAP, and a number of its competitors, notably Oracle, have been the principle actors heralding in the new kind class of software package known as the ERP system.4 Secondly, there are many who might take issue with the word ‘conquer’, which suggests the winning of a battle or victory of some sort. This is not what we intend here. We do not believe this new breed of software supplier has simply waged a war and emerged successfully. This is far from the case. Rather we use the term primarily as offering a counter to current biases within the Social Study of Information Systems towards localisation arguments, to encourage social scientists to offer alternative explanations for the rise of this new class of technology. It is in this sense we hope the title and the book more generally is read.

AIM OF THE BOOK To this effect, we see the book as a means to redress the imbalance that has developed in the social analysis of technology through encouraging a shift beyond the implementation study to study the ‘career’ of artefacts in their historical context and along the full length of their life cycle. This includes studying enterprise-wide software applications across their ‘software’ and ‘product’ life cycle. It is only through tracing what might be called the ‘biography’ of the modern enterprise system and observing this biography from multiple viewpoints and timescales that we can begin to understand how this kind of software package has emerged. It is interesting to note that, despite the fact that the packaged enterprise system has been around for a few decades now, that there are still far too few specific sociological concepts to capture the various interdisciplinary issues surrounding generic and commodified solutions, even though some scholars have acknowledged the stark difference between this and bespoke software (Sawyer 2000, 2001). There is no sociology of the packaged software solution. Much discussion seem content to borrow or recycle terms 12

from other fields: from the discipline of Information Systems itself whose terms and concepts have emerged from the study of different kinds of technology (for instance bespoke software development); or from anthropology and cultural sociology both of which exhibit particular ‘disciplinary biases’ constricting our view of these systems. If we are to be serious about studying the specific dynamics and lifecycle of generic packages as both software and product then we arguably require a more specific framework to do this. This should build on as well as provide a critique of existing social science accounts of technology. It is this that we modestly attempt here through offering an approach that can loosely be described as the ‘Biography of Artefacts Framework’.

The Biography of Artefacts Framework It has become axiomatic in Science and Technology Studies (STS), the sub-discipline in which we are located, that we need to analyse technologies not just as material objects but also as ‘heterogeneous assemblages’, which means taking into account the visions and beliefs, the techniques and practices, as well as the various actors involved in the development, implementation, use and governance of an innovation. The creation and implementation of new technologies thus involves a complex interplay between diverse ‘social’ and ‘technical’ factors. As there is no clear boundary between what is social and what is technical, we should refer to these more precisely as socio-technical factors (though the latter term is something of a mouthful and for ease of expression we have not always used it (Hughes [1983])).

One of the ongoing debates within STS is that there has been insufficient attention to theorising how these socio-technical assemblages and their development are shaped by their historical context. There are important exceptions of course. Bijker (1995), for example, drew attention to the configuration of the Technological Frame(s) that surrounds the development and use of an artefact. More particularly, scholars from the Social Shaping of Technology (SST) perspective (MacKenzie and Wajcman 1985) have sought to characterise how the pathways of technological innovation are patterned by their history and context. That is, how innovation is shaped by an array of existing social relationships: the knowledges and commitments of various actors involved; the complementary technologies available, and in particular by the ways these elements are all configured together (Sørensen and Williams 2002). One of the 13

new concepts that have arisen within SST to capture the dynamics of innovation processes is that of the ‘biography of artefacts’.

Brady, Tierney and Williams (1992) suggested that packaged software artefacts had biographies – highlighting how custom applications became the basis of commodified niche-specific solutions. Williams (1997) further applied this concept to analysing the evolution of Computer-Aided Production Management (CAPM) systems. The notion was used to describe how new industrial IT applications often emerged through the enhancement of existing applications. When supplier offerings were implemented, they inevitably had to be adapted to fit the technical and operational circumstances of adopting organisations. This process often threw up further useful innovations that could feed into future technology supply. Industrial automation artefacts thus evolved through successive cycles of technical development and industrial implementation and use, a ‘spiral of innovation’ if you like, oscillating between moments of development, implementation and use. These short-term cycles were phases in a longer-term biography; and longitudinal studies showed how the CAPM and MRP II systems of the 1980s and 1990s, widely seen as the precursors for today’s ERP systems, themselves emerged from stock and production control systems developed in the 1960s in Vehicle and Aerospace sectors. Later Pollock (Pollock et al. 2003; Pollock & Cornford 2004) expanded the concept of biography to study how ERP systems were able to move within and across industrial sectors (most notably from the private to the public). And in so doing they highlighted how these systems often carried with them large amounts of ‘accumulated functionality’ and how this ‘history’ had important implications of the reshaping of adopting organisations (public organisations and specifically universities). Overall then our early usage of the concept of biography drew our attention to the way in which the development and evolution of artefacts was shaped by its social (or more precisely socio-technical) context.

What we want to do now with our emerging biographies framework is to redirect the analytical lens, if you like; to broaden the scope of enquiry away from the (mainly implementation) stages about which we already have a reasonable body of knowledge and towards those locales and moments of innovation where much less is known (of which there are many of interest and relevance). We also suggest that to have a 14

complete understanding of the biography of the packaged enterprise solution we must not simply study single systems but trace and compare the career of a number of solutions. On top of this, we must also broader our field of view. We must shift our lens outwards at times to investigate the wider context in which these systems are located, to capture the other actors who play a role in constituting these systems and the market in which they are located. Some of the suggestions we will put forward in the book include the arguments of a move away from ‘flat ethnographies’ or simple methodological nostrums such as ‘following the actor’ (Latour 1987) to more theoretically informed, longitudinal selections of different sites and moments for study. We will argue that there is the need for a different type of qualitative study, a more ‘strategic ethnography’, which addresses the technology/society relationship at multiple levels and timeframes. In lieu of what we might call ‘atomistic studies’ we want to focus on the need for biographies that track the trajectory of a group of artefacts and their associated practices over time; and for better spatial metaphors addressing how these generic and global technologies are instantiated at multiple sites and across distributed contexts (see Burawoy et al. [2000] who have discussed this latter point albeit not in the context of software packages). Rather than study ERP, for instance, in particular socially/temporally bounded locales we argue for a ‘variable research geometry’ that can be applied to diverse issues and in differing contexts, depending on the issue(s) being addressed and entities being tracked. More concretely, we develop the concept of biographies as a means of analysing the emergence and evolution of software packages. And here we found useful to follow Hyysalo’s (2004) multi-level framework and distinguish at a least three different levels:

(i)

the development of particular enterprise systems (as well as the organizations and people connected to them);

(ii)

the development of an overall class of artefact; and

(iii)

the coupling of a technological field and a societal practice.

15

To this aim we seek to provide tools for analysing the influence of the social setting at different levels of generality (from the immediate micro-context of involved actors, to the broader institutional macro-setting) as well as the multiple historical timescales (short, medium and long-term) at which analysis may be undertaken. We identify the social spaces in which innovation occurs, including the specific arenas in which technologies are developed and implemented, and broader linkages across this heterogeneous community. Here, we highlight the emergence of new kinds of intermediaries, such as industry analysts, who help shape expectations about the development of technological fields and constitute markets for constantly changing supplier offerings.

RESEARCH SCOPE This book is based on a long-term research project where we have been able to assemble what we would argue is a comprehensive and in-depth picture of the evolution of particular enterprise-wide solutions for the greater part of their lifecycle, from their early stages of conception to today, including projections of future developments. We have had unique access to several software providers, including a global software giant whom we describe throughout using the pseudonym ‘SoftCo’, and a number of user organisations and user fora. Importantly we have been able to view the work of SoftCo from a number of distinctive viewpoints. Firstly, this was from ‘inside’ where we observed how they managed their packages as well as the users attached to them at one particular point in their lifecycle (the technical support process for their Customer Relationship Management (CRM) system). Secondly, again from the inside, we witnessed how SoftCo interfaced with various sets of users during the development of one of their products (this was the design and requirements gathering stages of a new ERP module that we describe throughout as ‘Campus’). Thirdly, we continued to study this particular ERP module along a number of different phases in its life cycle, from inception through to ‘maturity’. Fourthly, we also studied the module at the supplier-user nexus through long-term participation in a particular SoftCo User Group where we observed the user community attached to the module and wider ERP system. We can thus claim to have a comprehensive knowledge of this particular ERP module (having followed its career for nearly a decade now).

16

As well as discussions of ERP, we also analysed other solutions - although our knowledge of them is based on much shorter studies. We researched the design of a small software package, which is being used for accommodation management, relatively early in its development. We studied this system particularly from the point of view of its history and future projections (i.e. regarding which markets the supplier hoped to take it into) and this was useful as it allowed use compare the strategies of various vendors and at different stages of maturity. We also conducted a study of a CRM system procurement within a large public organisation. This gave us the opportunity to observe how suppliers present their solutions to potential users and how adopters make technical assessments of the various vendor offerings and ‘promises’. Finally we conducted a study of a group of industry analysts called ‘The Gartner Group’ which, whilst not directly involved in the production or use of software, play an important role in constituting various aspects of the wider packaged enterprise system arena.

In terms of our emerging framework, the approach we have adopted is a comparative one that analyses the biography of a number of packages as they move across similar organisations, from one national context to another, and from the private to public sector. The selected packages were at different stages in their biography and were characterised by different levels of product maturity and standardisation. This rich combination of data collection methods enabled detailed current and longitudinal analysis and comparisons between cases in different sectors and stages in the package maturation. Whilst we discuss our framework in more detail in Chapter Three, we briefly mention our overarching methodology and research design. Most of the insights presented here were gathered during ethnographic research and observation. At times, however, we also supplemented this research with interviews. We chose our sites based on a combination of opportunism and through theoretically informed choices about which sites and nexuses might be interesting and, according to our view of the state of the field, in need of further study. In other words, we studied those places where we could negotiate access (and a difficulty with access is one reason for the relative paucity of studies of packaged software design) but also sought out particular sites. These choices, of course, were constantly being modified to address new phenomena and particularly ‘surprises’. This was the case, for instance, when we chose to study the supplier/user nexus and the complex web of relations that exist 17

between them, which, in turn, alerted us to the important role of industry analysts. This group of experts has since become a major focus of our subsequent work.

The biography approach is a novel and ambitious one. There are two further aspects worth mentioning that have meant we have been able to conduct such a study. First, this is our historical perspective, facilitated by the fact we have been able to revisit findings and material from research conducted over the last two decades ago now as well as studies carried out more recently. In terms of the former, this was the 19871991 study of CAPM conducted by Fleck, Webster and Williams. This was research funded under the UK Economic and Social Research Council (ESRC) Programme in Information and Communication Technology (PICT) where we were amongst the first social scientists to investigate the shaping of earlier types of packaged enterprise solutions and their effects on organisations. In terms of the latter, Pollock and Cornford conducted a 1998-2001 a further ESRC study of an ERP implementation where they followed the system rollout over a three period in one particular organisation. This has meant in our current study – our 2003-2007 ESRC research grant on the Biography and Evolution of Standard Software Packages conducted by Pollock and Williams - we have been able to contrast our research and assessment at the time with the state of the field as it has panned out today. Importantly, this has allowed us to highlight the interesting linkages that current day technologies share with their older predecessors, and to analyse current developments in light of this historical knowledge. One of the unusual things this historical insight has permitted us to do, for instance, is to be able to criticise the now well-rehearsed historical account of the emergence of ERP systems. As the conclusions of the 1990 Eindhoven Group workshop demonstrate, the future direction of the technology was far from clear at the time. We have thus been able to trace (some aspects of) the complicated rise of the ERP system.

Second, in conducting our analysis, we have also drawn upon parallel research by our Edinburgh colleague D’Adderio (2004), doctoral research by Grimm (2008) and Wang (2007a), who through ethnographic research have looked at the birth and evolution of major software packages through contemporary and retrospective study. Grimm conducted a participant observation of one of the worlds leading software producers (some of which we include in Chapter Eight). Wang traced the birth and 18

evolution of the Chinese national champion selling organisational Product Data Management (PDM) technologies in the Construction sector. We might suggest that it is only because we can draw upon this distinctive historical and interdisciplinary knowledge base brought together in our wider research team that we can write this book. This suggests that studying the biography of the enterprise solution is increasingly a ‘team task’ and not something that can be done alone or though a single study (cf. Burawoy et al. 2000; Koch 2007).

Two other final points merit attention. Firstly, is that we do not include in the book an implementation study; nor do we look at the effects of enterprise-wide systems on adopting organisations and users. We do not see this omission as a limitation. Instead, it represents one of the choices we have made in the current study. Much has already been much written about ERP implementation in the sociological and information system literature; some of which is reviewed in Chapters One through to Three. We have taken this study and this book as an opportunity to explore other aspects of the career of the packaged enterprise systems, which we think are of equal importance but have had nowhere near similar attention. Secondly, we have chosen to mask the identity of the organisations we have studied, except in one case where we thought it necessary to talk specifically about the work of one particular actor. This is ‘The Gartner Group’ where we saw them to be of such significance, as well uniquely identifiable, that it would have made little sense to attempt to anonymize them.

Structure of this book We begin in Chapter One by introducing the recent history of the software package industry. Here longstanding concerns about the availability, price and quality of software surfaced in an episode which became known as the ‘software crisis’ and which led to the reorganisation of software production and ultimately the rise of packaged software. We also review the main phases of the software package lifecycle - from design through to procurement, implementation, use and post-implementation. Whilst this first chapter is primarily descriptive, the following two are conceptual in nature. Chapter Two engages in detail with debates within STS about some of the major frameworks through which technologies and innovation have been conceptualised, including ideas about how to theorise the relationship between

19

technology and society. Chapter Three is where we set out the Biography of Artefacts Framework. This is followed by the various empirical chapters.

There are five empirical chapters in all, which are organised around interrelated themes. The first two, which closely build on each other, study the tension between software as a solution to specific user problems and as a generic product. This is, to use the framework we are developing, the biography of a particular innovation and of a wider product. As we will show these are two different views on the same technology, highlighting often-contrasting needs and demands from the different actors involved. More specifically, Chapter Four describes the development of the new ERP module we call ‘Campus’ built by one of the largest software package suppliers in the world. We describe how many users agreed to act as pilot sites for the new software predicated on the belief that they could influence the shaping of the package (through allowing the software to be designed around their organisation). Once the supplier attempted to make their product more generic, however, the user organisations experienced a loss of control as their specific features were ‘designed out’ of the system. Unsurprisingly, this provided something of a strain on the relationships between the suppliers and user organisations (which is the theme of the chapter).

Chapter Five continues to analyse how package suppliers’ manage the tension between designing a module for a group of specific users and, at the same time, a set of wider yet unknown users (the potential global market). Taking the view of the software producer this time, we compare two software producers targeting a similar sector. We investigate how both take decisions about product design and markets as well as how these influence the uptake and eventual fit of the package. From a more conceptual point of view the chapter attempts to describe a set of revealed strategies by which suppliers produce software that embodies characteristics common across many users; what we term ‘generification’ and ‘generification work’.

The next two chapters, which also closely build on each other, shift our focus away from specific innovations and products and towards the wider ‘technological field’ that constitutes software packages. Through this term, we address how certain ideas

20

about software packages achieve currency and become resources for others (and also how these ideas can change over time). In this sense, the broader technological field can be thought to possess a biography as ideas about software packages and their organisational benefits change (but evolve much more slowly than that of particular artefacts within that field). What interests us in these two chapters is how the technological field surrounding generic software comes to be constituted by certain key players and are also sustained by the activities of the wider communities of organisational users. Chapter Six considers the issue of ‘procurement’. Procurement is interesting because it is the process by which a potential adopting organisation becomes bound up with the biography of a particular artefact. This process moreover turns out to involve high levels of ambiguity, as there is often a lack of reliable information about the capacities and performance of packages and their ‘fit’ with the particular requirements of the would-be adopter. Those procuring solutions are often faced with intensive marketing efforts and may see more detailed and comparable information by organising ‘beauty contests’ from different package suppliers. We show how the procurement team within one large organisation laboriously attempt to analyse and ‘compare’ various offerings so that an effective and ‘accountable’ choice can be made.

What is interesting about procurement, as we explore more fully in the next chapter, is that it takes place on a highly complex terrain, involving various actors (organisation members, expert intermediaries and suppliers) who influence the practice of choosing between packages, and thus help constitute the market of technology artefacts and the field of technological practice. Chapter Seven vividly demonstrates how there are new kinds of intermediaries emerging and the work thy do in shaping expectations about the nature of software packages as well as constituting the markets for constantly changing supplier offerings. In particular, we look at the work of ‘industry analysts’ and the construction of one of the most infamous of market analysis tools - the ‘Magic Quadrant’. This device is widely circulated amongst the IT community so as to compare and rank vendors according to a number of highly contested evaluative criteria. These assessments include intangible properties such as supplier ‘competence’ and ‘vision’. Given that potential adopters are drawn to assess the reputation of vendors and products during procurement, we find these tools play an important role in mediating choices. Their assessments appear to play a role in 21

allowing user organisations to make comparisons between the proliferations of offerings. In other words, what was once a highly uncertain terrain is now becoming more organised (and we highlight the role of industry analysts in drawing and redrawing boundaries around the technological field).

Chapter Eight is the final empirical discussion and takes us inside the offices of ‘SoftCo’. Here we view the software through the lens of a supplier attempting to manage its wide family of generic products through one particular moment in the software package lifecycle - the support function (the process by which the supplier resolves the technical problems its users experience when installing solutions). The supplier is faced with the difficult problem of supporting the systems of its massively large and highly distributed user base. It has recently moved from what might be described as a ‘territorial’ model of support to one that is a predominately online (what we describe as a ‘globalised face-to-portal’ form of support). We describe the novel organisational form that has been put in place, which includes a global network of labs and sophisticated ICTs, and show the complex workings of this network as well as the various strategies developed by the supplier to lift out technical problems from out of their local context and bring them back to its labs.

Chapter Nine attempts to both bring together the main concepts developed in the book as well as build towards a more general discussion of how we might develop our conception of software packages. The reader should be aware that each of the arguments developed in the empirical chapters gradually build upon each other. They reflect the (roughly chronological) development of both our empirical and conceptual analysis. Each chapter develops diverse themes most of which (but not all) contribute to the overall discussion of the biography of the enterprise system. In this sense, the book is best read from beginning to end. The empirical chapters are to some extent self-contained in that each discusses its own theoretical starting point, empirical setting and detailed methodology and presents conclusions. In this respect, the reader should be aware that the final conclusions presented in Chapter Nine do not attempt to recapitulate all the arguments developed in the empirical chapters. The various chapters, particularly the empirical ones, reflect our own intellectual journey and process of discovery. The careful reader will notice how our analytical lens develops

22

– changing its focus and (hopefully) growing in sophistication and acuity, as we progress and trace out the biography of the enterprise system.

1 To some extent, this has already happened. As we will show many of the more influential accounts of enterprise-wide systems now stem from American Business Schools. 2 This broad area of work might in some places also come under alternative designations like ‘Social Informatics’ (see for instance Kling [2007]). 3 We might call the seemingly impossible project of developing the generic solution as akin to the ‘flying bumblebee problem’. According to calculations and presumptions from theoretical physics, bumblebees should not be able to fly (this is despite the frequent observation of flying bumblebees). 4 SAP is the biggest ERP supplier, followed by Oracle (which in 2003 took over Peoplesoft which in turn had just acquired JD Edwards – the next two largest players in the ERP market). Between them, they account for over 60% of the market, though there are a number of other challengers (Brunellii 2006).

23