Opening the black box of ERP: The implementation of an ERP system and the effect on management control

(1)

Opening the black box of ERP:

The implementation of an ERP system and the effect on

management control

Master’s thesis

By

Henk-Jan van Roekel

University of Groningen Faculty of Economics and Business

MSc Business Administration - Organizational & Management Control

Jozef Israëlsstraat 51 9718 GD Groningen The Netherlands hjvanroekel@gmail.com +31630415030 Student number: 1690590

Supervisor: dr. M.P. (Martijn) van der Steen Co-supervisor: prof. dr. ir. P.M.G. (Paula) van Veen-Dirks

(2)

(3)

3

Abstract

Over the last three decades, enterprise resource planning (ERP) systems have become widely used by all sorts of organisations. Despite their enormous potential, many companies still struggle with the implementation and use of ERP systems. Many authors point out the unstable and complex nature of ERP systems. These systems are argued to be hard to change, both in and after the implementation phase. This thesis draws on actor-network theory (ANT) and interprets the implementation of an ERP system to increase our understanding of the effect of such implementations on management control. It shows that an ERP implementation can be seen as the opening of a black box. Moreover, it lets us see that the punctualisation of actor-networks can be very helpful (though precarious) in managing complex situations. Furthermore, it extends the work of studies of ERP and management control by illustrating how organisations can fail to address management control issues early on in the implementation, potentially rendering management control hard to change.

(5)

5

Preface

“A man is not idle because he is absorbed in thought. There is a visible labour, and there is an invisible labour.”

Victor Hugo

With this thesis, my Master’s study Business Administration: Organizational & Management Control at the University of Groningen will come to an end. After a slow start, with a lot of invisible labour and lots of extracurricular activities like singing in my band(s) and being an active member of my student association, I eventually have reached this point in time. This thesis was conducted between October 2012 and June 2013. Hopefully, this research can be used as an inspiration for future research, for the pursuit of successful information systems implementations, and for an understanding of the relationship between information systems and management control.

I have many people to thank, and in this paragraph I will do those who I mention scant justice. None of the work below would have been possible without the help of my parents. Therefore, I would like to thank them for their unconditional support, not only in the period of this study, but throughout my whole education. Also, I am grateful to my family, friends (especially my roommates) and all others who supported me or had to endure my mood swings. Further, I would like to express my appreciation to my supervisor, dr. Martijn van der Steen, for his valuable feedback and inspiration during the writing of my thesis. I also wish to thank my second supervisor, prof. dr. ir. Paula van Veen-Dirks, for taking the time to read and assess this thesis. Finally, I would like to thank all the employees from both companies in my case study that aided me in my research.

By completing this thesis, my student career has come to an end. While I enjoyed this period in my life, I am glad it is finished and I am satisfied with the results.

Hopefully you will enjoy reading this thesis,

(6)

6

Introduction

Over the last three decades, many organisations, both large and small, have implemented enterprise resource planning (ERP) systems. Although the implementation of these systems can be very time-consuming and costly, ERP systems have become widely used (Cooper & Kaplan, 1998; Scapens & Jazayeri, 2003). Companies store vast amounts of data, which often is spread across many separate computer systems, functions and business units. Enterprise resource planning (ERP) systems have the potential to integrate all corporate information in a single comprehensive database. In principle, ERP systems allow every organisational object to be visible at every organisational position (Dechow & Mouritsen, 2005). ERP systems are considered “the most important development in the corporate use of information technology in the 1990s” (Davenport, 1998, p. 2).

Prior to the introduction of the enterprise resource planning system (sometimes also referred to as enterprise system), business functions had their own information system which operated separately from the information systems of other functions within the same organisation (Davenport, 1998). Enterprise resource planning systems (e.g. SAP, Oracle, and Microsoft Dynamics NAV) aim to integrate these functions. These systems are defined as “enterprise-wide package[s] that tightly integrate all necessary business functions into a single system with a shared database” (Lee & Lee, 2000, p. 281). They are packaged software applications that “connect and manage information flows within and across complex organisations” (Davenport, Harris, & Cantrell, 2004, p. 16), enabling managers to make decisions based on ‘real-time’ information.

Despite the enormous potential of ERP systems, many companies continue to struggle with its implementation and use. While the easy use of an acronym suggests otherwise, ERP has an incredible technological complexity and knows many diverse motivations for implementation (Chapman & Chua, 2003). Many authors point out this unstable and complex nature of ERP systems (e.g. Davenport, 1998; Quattrone & Hopper, 2001; Chapman & Chua, 2003; Scapens & Jazayeri, 2003; Chapman, 2005; Dechow & Mouritsen, 2005; Hyvönen, Järvinen, & Pellinen, 2008).

ERP systems are argued to be hard to change, both in and after the implementation phase. They comprise a set of modules, which span different business functions. These modules access business processes, based on industry best practice (the vendor's interpretation of the most effective way to perform a business process). This urges organisations to adapt to the system and reengineer their business processes; as opposed to adapting the ERP system to the organisation (Davenport, 1998; Davenport et al., 2004; Scapens & Jazayeri, 2003). Because of the frequent use of copy-past techniques in ERP implementations, (Davenport, 1998) ERP systems may easily become unquestionable and unchallenged. Under such conditions, even small changes may become difficult to implement (Granlund, 2009).

(7)

7

keeping with Hopwood’s (1983) earlier research findings that management accounting practices can only be understood in context. However, the relationship between ERP and management control is still an underdeveloped area; empirically, theoretically as well as practically (Granlund & Mouritsen, 2003). The studies that have attempted to analyse the impact of integrated technology on management control are said to have merely ‘opened the discussion’ (Granlund, 2011). Therefore, “ERP holds out fascinating possibilities for future studies that might significantly further our understanding of the nature of management control” (Chapman, 2005, p. 688).

The configuration process that occurs while implementing the system dictates what can be done - and what not - with the system in practice. Information system configuration may thus define certain aspects of management control (Dechow & Mouritsen, 2005). Furthermore, decision-makers can be unaware of the consequences of the choices they are asked to make during system implementation (Granlund, 2011). Moreover, when configured, ERP systems are hard to change, which means that companies might have to reconcile with the resulting inadequacies (Davenport, 1998; Dechow & Mouritsen, 2005).

Burns and Scapens (2000, p. 4) observe that “little research attention has been given to understanding the processes through which […] new management accounting systems and practices have emerged (or failed to emerge) through time”. The aim of this thesis is thus to follow the process of an ERP system implementation and show how the system emerges (i.e. becomes visible) trough time. I will draw on aspects of Latour’s Actor-Network Theory (ANT) to study the implementation process of an ERP system within HEALTHCO, a healthcare organisation, which is led by FANDOR, a software vendor. Chapman, Cooper and Miller (2010, p. 1) emphasise the “urgent need for research which can provide a basis for seeing accounting as both a social and organizational phenomenon”, and state ANT as a means to accomplish this. Moreover, an actor-network approach combined with a case study can reach beyond more or less simple assumptions of cause and effect (Granlund, 2011, p.15).

This thesis illustrates that an ERP system which is not yet implemented can act as a black box. It also shows that the black box of the ERP system, like all black boxes, is “leaky” (Callon & Latour, 1981, p. 285). Moreover, it shows that the creation of alliance networks can render the ERP system visible, hereby opening the black box. Additionally, this thesis addresses how the notion of punctualisation, i.e. how an entire network can be reduced into a single actor in another network (Callon, 1991), can be applied in the setting of an ERP implementation. The case study also demonstrates how management control issues are stalled in the beginning phase of the implementation, and are only wished to be acted upon once the operational process runs smoothly. Since initial configurations are hard to change, (Davenport, 1998; Dechow & Mouritsen, 2005) changes aimed at achieving an effective management control system may be difficult to implement.

(8)

8

Background Literature

Actor-network theory

In this thesis I use actor-network theory1_{(or ANT) as a framework to investigate the beginning}

phase of the implementation of an ERP system. ANT, which is also known as the sociology of translation, derives from the work of Latour and Callon (e.g. Callon; 1986; Callon & Latour, 1981; Latour, 1987; Latour, 1996; Latour, 1999a,b; Latour, 2005). It is a constructivist approach, understanding change as an outcome of historical contingent processes, as opposed to linear rational improvements or functional adaptions.

ANT suggests that individuals form alliances and enrol other actors, and use artefacts to strengthen such alliances, creating networks of human and non-human actors which either facilitate or constrain the emergence and diffusion of innovations (Chua, 1995). “An actor in ANT is a semiotic definition – an actant2 – that is something that acts or to which activity is granted by another…an

actant can literally be anything provided it is granted to be the source of action” (Latour, 1996, p.373, emphasis added) An actor can be defined as “any element which bends space around itself, makes other elements dependent upon itself and translates their will into the language of its own” (Callon & Latour, 1981, p. 286).

Alliances of human and non-human elements are called sociotechnical networks or heterogeneous networks. ANT was developed to analyse situations where separation of these elements is difficult. For example, it is difficult to differentiate an ERP system’s technical aspects from the influence exerted by the socio-cultural background of the developers (cf. Tatnall & Gilding, 1999). What appears to be social is partly technical, and what may seem to be only technical is partly social. ANT therefore denies a distinction between the two. (Hyvönen et al., 2008)

According to ANT, everything happens within networks; innovations would not exist without the networks they are embedded in. Latour suggest that the success or failure of a technology “cannot be predicted by correlating with a list of ‘social factors’, [but rather] lies in the hands of those who come after the ‘inventor’” (Briers & Chua, 2001, p. 240). Accounting technologies are not seen as inert, but are considered to be shaped by an interplay of actors who actively ‘push and pull’ the technology (Alcouffe, Berland, & Levant, 2008; Lodh & Gaffikin, 2003).

The concept of translation

ANT analyses change as a process of translation. “Translation […] refers to the process of creating a temporary social order, or the movement from one order to another, through changes in the alignment of interests in a network” (Sarker, Sarker, & Sidorova ,2006, p. 54). One of the definitions given by Latour (1999b, p.179) is that translation means “displacement, drift, invention, mediation,

1_{At first, Bruno Latour criticized the term ‘actor-network theory’ as poorly designed, faulty and, like a}

malfunctioning product, badly in need of recall (1999a, p. 15). However, by 2005 he accepted the term, stating it is “so awkward, so confusing [and] so meaningless that it deserves to be kept” (2005, p. 9).

2_{Bruno Latour (e.g. 1996) sometimes uses this word instead of actor, an approach to differentiate between}

(9)

9

the creation of a link that did not exist before.” There are four processes of translation, which are: Problematisation, interessement, enrolment and mobilisation (Callon, 1986).

Problematisation refers to an actor’s effort to convince other actors to subscribe to their own view by showing they have the demonstrating that they have the right solution - in this case, the ERP system - to the others’ problems (Alcouffe et al., 2008; Lowe, 2000). Interessement involves creating an interface between the interests of heterogeneous actors and strengthening the link between these interests (Lowe, 2000). The process of interessement often involves negotiations among actors. However, Callon (1986) stresses that the actors do not always participate in such negotiations themselves. Spokespeople or representatives may negotiate on their behalf. For example, a CEO can speak on behalf of an employee or an IT vendor can speak on behalf of an ERP system. This does not necessarily mean that actors will abide by the agreements signed off by their representatives (Sarker et al., 2006). Actors can fail to act as promised by their representatives, a phenomenon referred to as betrayal. If and when interessement is successful, subsequent enrolment is said to occur. Enrolment involves the creation of alliance networks, aiming to build up agreement among the relevant actors concerning their interests (Alcouffe et al., 2008) Since any enrolment is necessarily temporary, betrayal by an enrolled actor (i.e. acting in contradiction to the interests it has agreed to support) is always a possibility (Sarker et al., 2006). Finally, mobilisation refers to monitoring the different interests; so that they remain more or less stable (Alcouffe et al., 2008). Note that “these moments [emphasis added] are not meant to be regarded as entirely distinct, nor do they imply temporal differentiation.” (Lowe, 2000, p. 86). The different moments of translation are summarised in Table 1.

Table 1 – Different moments of translation

Moment Definition

Problematisation actor’s effort to convince other actors to subscribe to their own view by _{demonstrating that they have the right solutions}

Interessement creating an interface between the interests of heterogeneous actors and _{strengthening the link between these interests}

Enrolment creation of alliance networks, aiming to build up agreement among the _{relevant actors concerning their interests}

Mobilisation monitoring the different interests, so that they remain more or less stable

Actor-network theory in accounting research

(10)

10

medium.” (Roth & McGinn, 1998) Accounting numbers (see Robson, 1992), graphs, lists, photographs, spread sheets and equations can all be classified as inscriptions. Building inscriptions requires “techniques of ‘marking’ an object or event that is to be known – writing, recording, drawing, tabulating”, where ‘knowledge’ is the outcome (Robson, 1992, p. 689).

Inscriptions are important because they are able to “convert local events into textual forms that are mobile and can stay immutable through their displacements” (Latour, 1986). Their mobility and immutability allows inscriptions to be easily sent to most places while remaining the same, e.g. by means of computer networks, mail, or even cell phones. They thus enable controllers of a system to accumulate knowledge ‘at the centre’ in order to ‘act at a distance’ (Lowe, 2001; Robson, 1992). An analogy given by Latour (1987) describes how a map can make it possible for one to act in relation to a distant place that one may never have set foot upon. This ‘action at a distance’ is explained by Chua (1995, p. 116) as “enabling people far away from the scene of activity to ostensibly have a window on those activities and intervene in the name of better management.”

The second use of ANT addresses the diffusion of accounting innovations and track how they are “produced, modified and accepted (Alcouffe et al., 2008, pp. 3-4).” (e.g. Chua, 1995; Preston, Cooper, & Coombs , 1992; Quattrone & Hopper, 2005). It should be noted that accounting technologies are “never merely diffused, adopted or implemented; they are adapted and translated and, at the same time, they are enrolled in a network that reconfigures other actors’ interests.”(Justesen & Mouritsen, 2011, p.176). This thesis follows this second use, investigating the translation of an ERP system.

ANT has several specific benefits for management accounting innovation diffusion (Chua, 1995; Justesen & Mouritsen, 2011). First, the concept of translation gives a novel perspective on accounting change, rejecting the assumption of linearity. Second, accounting technologies are granted a more prominent role, because they are given actorhood through their relation with other actors. Third, ANT accounting research rejects traditional sociological dichotomies (e.g. macro/micro, structure/agency) but uses detailed descriptions of processes and actions at the empirical level (Justesen & Mouritsen, 2011).

ERP and management control3

Three decades since the first use of ERP systems, there is still a lack of understanding of the relation between ERP systems and management control (Granlund & Mouritsen, 2003; Rom & Rohde, 2007; Beaubien, 2013). There is much literature concerning ERP systems in accounting information system (AIS) journals, but considerably less attention has been paid to this phenomenon in accounting research journals (Scapens & Jazayeri, 2003; Granlund, 2009) although the number of studies in the field is seemingly increasing. This seems surprising since ERP has been argued to have significant implications for management accounting and control.

3_{See Rom & Rohde (2007) for a more detailed review of research regarding management accounting and}

(11)

11

The earlier empirical studies on the implications of ERP systems for management control suggest that ERP systems’ impact on management accounting and control remains very moderate. (e.g. Granlund & Malmi, 20024). However, these studies adopt a relatively static approach; they do not

explore the processes of change and the opportunities created by ERP implementation (Scapens & Jazayeri, 2003).

The cross-sectional field study by Granlund & Malmi (2002) studied a unidirectional relationship between ERP systems and management accounting and control, arguing that organisational practices “are typically changed to fit the new technology, not vice versa” (Granlund & Malmi, 2002, p.305). They found that ERP systems have had little impact on management control because the new technologies were perceived as cumbersome. They did not observe any changes in methods of control and only minor changes in the use of existing control systems, partly because ERP was often only loosely coupled to these control systems.

More recently, a second strand of literature has emerged, that says we need to study processes of change5_{, rather than seek to identify the 'impacts of ERP systems' (e.g. Scapens & Jazayeri, 2003,}

Dechow & Mouritsen, 2005; Quattrone & Hopper, 2005), While the first strand of literature struggles to find effects of ERP to firm performance, this set of authors ‘identify noteworthy and […] huge effects of ERP systems’, both in design and use (Dechow & Mouritsen, 2005, p. 692). These authors question the assumption that change is linear and dismiss simple cause and effect relations. Their studies show how organisational actors both influence and are influenced by ERP systems, trying to unravel the complexity of these systems. This strand of literature mostly uses case research, and explains why surveys are unable to capture ERP systems’ effects.

Scapens and Jazayeri (2003), following the implementation of an ERP system, did not find any fundamental changes in the accounting information used, but did find changes in the role of management accountants. They point out that ERP is not a standalone driver, but characteristics of ERP (i.e. integration, standardisation, routinisation and centralisation) opened up certain opportunities and facilitated changes already taking place within the company.

Granlund and Malmi (2002) and Quattrone and Hopper (2001) also found evidence of a wider role for management accountants in form of more business-oriented tasks. This could mean that management accountants are getting involved in general management by acting as business consultants (Rom & Rohde, 2007). Or to put it differently, the introduction of ERP systems leads to new, hybrid positions, where management accountants also carry out non-accounting tasks (Caglio, 2003; Hyvönen et al., 2009).

Dechow and Mouritsen (2005) differentiate between ERP as an integrated database and ERP as a broader system impacting management control by configuration decisions. They show how ERP

4_{In their research note on ERP systems and management accounting change, Scapens & Jazayeri (2003) also}

mention a series of conference papers: Fahy and Lynch (1999), Maccarone (2000), Booth, Matolcsy and Wieder (2000), and Beretta (2001).

5_{Quattrone and Hopper (2001) even suggest replacing the concept of organisational change by the notion of}

(12)

12

systems incur a ‘techno-logic’ that conditions how control can be performed. They argue that ‘control cannot be studied apart from technology […] because one will never get to understand the underlying “infrastructure”’ (Dechow & Mouritsen, 2005, pp. 730-731). Moreover, they argue that information system configuration can define certain aspects of management control and suggest that ERP creates visibilities; but with these visibilities, also blind spots. Decision-makers therefore cannot always oversee the consequences; not in the least because ERP does not simply enable or constrain control practices, but also new forms of management control can emerge. The actorhood of an ERP system is shown, as they argue that ERP systems are mouldable, but can fight back because choices for a specific item can make other items impossible.

ERP as a black box?

ANT states that every actor consists of a network, and every network is an actor (e.g. Callon, 1991). If this is the case, where draw the line? Law (1992, p. 384) uses the example of a television to illustrate this issue:

“For most of us most of the time a television is a single and coherent object with relatively few apparent parts. On the other hand when it breaks down, for that same user – and still more for the repair person – it rapidly turns into a network of electronic components and human interventions.”

Technologies ranging from seat belts (Latour, 1992) to nuclear weapons (MacKenzie 1996) have been depicted as black boxes. The presence or absence of networks has to do with simplification; it is dependent on the perspective of the observer. Although it is possible to endlessly delve deeper in actors, every time finding new networks, in practice we do not detect these complexities (Law, 1992). So, “when a network is strongly convergent and irreversibilised it can be [reduced] to a black box [emphasis added], whose behaviour is known and predicted independently of its context” (Callon, 1991, p. 152). An entire network then can become a single actor. Latour states that black boxing is:

“…the way scientific and technical work is made invisible by its own success. When a machine [or ERP system] runs efficiently, when a matter of fact is settled, one need focus only on its inputs and outputs and not on its internal complexity. Thus, paradoxically, the more science and technology succeed, the more opaque and obscure they become”. (Latour, 1999b, p. 304)

(13)

13

“The impossible task of opening the black box is made feasible (if not easy) by moving in time and space until one finds the controversial topic on which scientists and engineers are busy at work. This is the first decision we have to make: our entry into science and technology will be through the back door of science in the making, not through the more grandiose entrance of readymade science” (Latour 1987, p.4).

The opening of a black box thus is the very beginning of a process of translation, where actors will try to (once more) create a temporary social order.

Methodology

Method

This thesis reflects a case study of the beginning phase of the implementation of an ERP system within HEALTHCO6_{, a healthcare organisation. The research approach used is an interpretive case}

study.In particular, case studies are useful for in-depth research into a narrow issue within a specific organisation (Ahrens & Dent, 1998; Yin, 2011). Yin (2003) promotes the case study approach by stating the following:

A case study is an empirical inquiry that investigates a contemporary phenomenon within its real-life context, especially when the boundaries between phenomenon and context are not clearly evident. [...] The case study inquiry […] benefits from the prior development of theoretical propositions to guide data collection and analysis. (Yin, 2003, pp. 13–14)

In practice, interpretive management accounting research seeks to understand how social reality emerges from subjective interpretations and is objectified through human interactions (Ahrens, 2008; Ahrens and Chapman, 2006; Chua, 1986). Latour (1987) argues that we should study science “in action”, rather than ready-made science. Therefore, researchers should arrive before the facts and machines are black boxed or follow the controversies that reopen them (Lowe, 2001). The mantra of ANT is to ‘follow the actors’ (e.g. Latour, 1996). Latour (1996) compares it to a murder mystery novel in which directions and leads given by actors are followed up by the detective, i.e. the researcher. “To understand how a technology such as ERP acquires particular characteristics requires a fluid, adaptive mode of investigation that follows trails revealed in the field.” (Quattrone & Hopper, 2005, p. 744). Where possible, this mantra was followed. However, considering the scope of this thesis, always following the actors is time consuming and expensive. Thus, the spirit rather than the letter of ANT was followed (cf. Quattrone & Hopper, 2005).

Latour (1996, p. 2) calls ANT “as much an ontology or a metaphysics, as a sociology.” The aim of ANT is thus not to tell researchers what they should find, but rather how they should investigate. Lee and Hassard (1999, p. 392, emphasis in original) describe ANT as “ontologically relativist, in allowing the world to be organized differentially, yet empirically realist in providing `theory-laden' descriptions of organization.” They argue ANT has an antipathy to define itself, which results in a failure to forge its own boundaries. At the same time, this is the key to ANT’s success, making itself as ‘blank’ as it deems possible. By refusing to set boundaries or build discriminations, ANT is able to record the discriminations performed and boundaries constructed in the activity it studies (Lee &

(14)

14

Hassard) and let ANT concepts describe the case in which it is used, and vice versa. Also, without any preconception of what is worth looking at, we cannot know what data is worth gathering. This makes empirical observation inherently ‘theory-laden’. Therefore, like other ANT studies (e.g. Lowe, 2000, 2001; Dechow & Mouristen, 2005, Quattrone & Hopper, 2001, 2005), I use theory in my case analysis to guide my findings.

According to Latour, (2005) research of the social entails a number of challenges, or ‘sources of uncertainty’. On two of these, I will elaborate some more, in order to avoid a number of pitfalls in the analysis. First, there is ‘no group, only group formation’. “[T]here is no relevant group that can be said to make up social aggregates, no established component that can be used as an incontrovertible starting point” (Latour, 2005, p. 29). When group formation would stop, a group would cease to exist. Thus, I am interested in how groups are (re-)formed, what action binds them together.

Second; objects, too, have agency. “[N]o science of the social can even begin if the question of who and what participates in the action is not first of all thoroughly explored, even though it might mean letting elements in which, for lack of a better term, we would call non-humans.” (Latour, 2005, p. 72, emphasis in the original). For example, it is possible to create a social relationship with an ERP system, or other things that are not social by nature. Granting objects the same status as human actors may seem rather radical. Many critics have expressed disagreement with the theoretical position that non-human objects (e.g. ERP systems) can have their own interests (Sarker et al., 2006). However, I see it as a useful metaphor for analysing sociotechnical networks in detail. Also, one might argue that is unethical to reduce human actors to mere objects. On this subject, I embrace Law’s (1992, p. 383) argument that this “is an analytical stance, not an ethical position”.

Third, the number of actors should be limited in order to create meaningful research:

“How ridiculous is it to claim that inquirers should ‘follow the actors themselves’, when the actors to be followed swarm in all directions like a bee’s nest disturbed by a wayward child? […] And if each actor is made of another bee’s nest swarming in all directions and it goes on indefinitely, then when the hell are we supposed to stop? If there is something especially stupid, it is a method that prides itself in being so meticulous, so radical, so all encompassing, and so object-oriented as to be totally impractical.” (Latour, 2005, pp. 121-122)

Data collection and analysis

ANT does not prescribe which methods to use to investigate reality. In this thesis, semi-structured interviews comprise the largest part of data collected. Over the course of the study, respondents were interviewed formally. The interviewees came from both the healthcare company where the ERP system was implemented and the vendor that delivered the ERP system. I conducted three interviews with the CEO of the healthcare company, HEALTHCO. Two interviews were conducted with the lead consultant of the supplier, FANDOR. One interview was conducted with the project coordinator of HEALTHCO, who is also an employee at FANDOR7. This interview was included after

the name of this interviewee emerged in other interviews, as an attempt to ‘follow the actors’. The interview schedule can be found in Appendix 1.

(15)

15

The interviews were broadly structured around the topics of the study and spread over a month. The interview guide used for the interviews can be found in Appendix 2. The interview guide covered questions such as: ‘Does the information provided by the ERP system create possibilities for management control?’ and, ‘Which people participated in the implementation of the new ERP system?’ However, these questions were not followed up in a fixed order and were used primarily to start conversations. Interviewees were encouraged to speak freely and raise issues that they considered to be important; issues raised by subjects were pursued. By allowing the flow of the interviews to be directed by the interviewees, I strove to minimise the effect of my own conceptions on the structuring of the interview in order to capture the interviewees’ own interpretations as they organised their accounts of the implementation.

Since the CEO of HEALTHCO did not consent with recording, I wrote down his answers to the best of my ability; extensive notes were taken. With all his interviews, the notes taken during the interview were re-written in a Word document within two hours. The interviewees of the ERP system supplier agreed to the recording of the interviews. The interviews with people from this company were recorded and transcribed in full length. The interviews lasted between an hour and an hour and a half. All interviews were held and transcribed in Dutch, only to be translated when used as a quote in this thesis. Documentary material such as company histories, brochures, and the project proposal were collected and compared with interview material. Over the course of the study, I have kept a log, indicating all interviews, company visits and other events. Also, I structured seemingly important answers in a PowerPoint document, ordered by important themes within my study, such as ‘interessement’, ‘black box’ or ‘enrolment’.

(16)

16

In the following, I distinguish between the ERP technology and the ERP system. Whereas ERP technology in this case refers to the software of Microsoft Dynamics NAV, I use the term ERP system as a label for the broader notion of the specific system in this particular organisation (cf. Dechow & Mouritsen, 2005).

Case analysis

In the case analysis I will start by providing a contextual background, in order to show how the “current situation under investigation emerged” (Klein & Meyers, 1999, p. 73). After that, I will narrate how the implementation of the ERP system can be seen as the opening of a black box. This starts with the identification of the relevant actors, followed by the first chink in the black box. I will then go on by following the formation of two alliance networks, which further revealed the ERP system. This required the punctualisation of a certain group of actors, which I will elaborate about after that. Finally, I will show how management control issues were neglected in the early phase of the implementation.

Contextual background

The context of the case research was an implementation, in which two enterprises were involved: the client, who bought the system, and the supplier, who sold and implemented the system. To provide additional context, the ERP technology and the reasons for adopting this ERP system are also elaborated on.

The client

HEALTHCO is a healthcare organisation, consisting of two business units. One business unit supplies oxygen, oxygen therapy and devices to consumers. The other business unit, under a separate legal entity (SLEAP), provides treatment of sleep apnoea. HEALTHCO consists of 45 employees and is a subsidiary of the OXYGEN Group , a leading company in gases for industry, health and the environment. In an attempt to integrate the two business units and “create efficiency”, HEALTHCO had started with the implementation of a new ERP system in their organisation.

The supplier

FANDOR is an independent software vendor that develops, sells and implements ERP, CRM (Customer Relationship Management) and BI (Business Intelligence) software. FANDOR is specialized in creating industry specific software, based on the Microsoft Dynamics NAV technology. They operate in several industries (e.g. healthcare, logistics, retail).

The technology

(17)

17

testing, training, migrating over old data, and finally beginning active daily use. Many vendors (like FANDOR) develop additional customizations to the software to improve how it enables and supports certain businesses or whole industries.

Reasons for ERP

HEALTHCO’s CEO sought to “integrate and update [their] automation”, as he thought this would mean a great step towards their Target Business Model. The parent company, OXYGEN, required HEALTHCO to report to them in a specific way, something the previous accounting systems were not able to. This had led to conflicts in reporting.

Another reason for implementing the system was that HEALTHCO had acquired SLEAP in 2009, which was to be combined with HEALTHCO. Within a few month, the two legal entities were about to merge. It was considered a problem that these business units still worked with different databases. The HEALTHCO unit did not have an ERP system, but worked with Access, Microsoft’s database management system. SLEAP did have an ERP system, Microsoft Dynamics NAV 3.7, but this system was not able to provide the reports desired by the parent company. HEALTHCO’s CEO stated: “From now on, we want to treat all patients with the use of only one database”. This had to be done by the implementation of a new ERP system, the object of study in this thesis.

When asked why he had chosen this specific ERP technology, the CEO praised the flexibility of the system. “Within OXYGEN, our parent company, they work with SAP8_{. However, SAP is too heavy for}

their relatively small subsidiaries, [who have an annual turnover of] up to 20 million [euros], and is less flexible. For these reasons I chose Navision Dynamics, after which I had the important task to promote this system to the parent company”. Since SAP was used by and known to the parent company, for them it would be the obvious choice. However, after a presentation of the new Dynamics NAV by FANDOR at the end of August 2012, the CEO was convinced that NAV would be a better choice.

“[To convince the group] I had to prepare a Management Project Report, a document in which I described all the arguments for this specific choice. It contained the consequences, the required invest, the payback period, etcetera. I had to present the project in three steps. I presented my plan at the HEALTHCO Group, Group Finance OXYGEN and Group IT OXYGEN. Because we are part of such a large company, they have to look at the safety of the system and the reporting quality. A chain is only as strong as its weakest link. It takes quite some reporting work to convince everyone, but this is also a good thing, it means you have to think about it.” He showed the Management Project Report, coming from a large blue binder. The last page contained no less than five signatures for approval, indicating both the struggle and the victory of the NAV ERP technology. The parent company had consented with his plans; in the end of January 2013 the contract to implement the system could be signed.

Identification of the relevant actors

In order to begin the ERP system implementation, both parties had to decide which employees would be dedicated to this project. On the supplier side, they installed a project manager, who was

8_{The best-known ERP software product, by the corporation with the same name. A competitor of Microsoft in}

(18)

18

in charge of three consultants. One of these consultants was appointed as the lead consultant. Also, when necessary, he was able to employ several developers. On the client side, they also needed someone who would coordinate the project. However, the lead consultant explained that clients “often do not have the knowledge or the time to run a project, next to their daily activities”. This led to a situation where FANDOR supplied an employee to HEALTHCO, to act as their project coordinator. This project coordinator explained:

“The method we use with our projects is that we appoint a project manager from the supplier, FANDOR, who ensures the organisation of the project, seen from the supplier. He controls the planning of consultants, facilitates the workshops, does everything that needs to be done. But on the other hand, of course, you need an organisation from the client’s angle. You need, in fact, a first port of call to do business with. In the case of HEALTHCO, they had no one in their organisation of whom they said: ‘that person would be fit for that job and has enough time for it.’ Apparently, in the sales process, they thought that it might be helpful if someone from FANDOR was going to fulfil that role. […]I see great advantages of this for the HEALTHCO organisation, because I have much knowledge of the Navision product and I know how such projects progress, see, how that works and what should be done in order to make it work. In that sense, I think it’s a very logical choice, but it remains a bit ambiguous of course”

Later on, he explained what the potential consequences of this ambiguity might be:

“[..] it would be ridiculous if I would do things that would harm FANDOR, wearing my ‘HEALTHCO-cap’9_{, while}

I know that HEALTHCO might benefit from that. It won’t work that way and this is also clear to [HEALTHCO’s CEO]. When he would have wanted that, he should not have chosen me.”

This same ambiguity was also addressed by the lead consultant of FANDOR:

“[The project coordinator] falls under one of the legal entities of FANDOR […] so, yes, he is a colleague of mine, [laughs] so he has to work with two caps10_{. He has to, yeah, defend the interests of HEALTHCO and [at}

the same time] work together with us. So that’s a challenge for him. […] And yes, you could call it peculiar, but things like this happen more often.”

Introducing the system

As mentioned earlier, the ERP technology that was introduced to HEALTHCO was based on Dynamics NAV 2009 R2, a standard ERP software package from Microsoft. However, none of the interviewees used this term when talking about the software. Often it was simply called NAV, even more it was called Navision, a title Microsoft itself did not use anymore since 2005. FANDOR’s lead consultant explained the technology with the help of a sketch; Figure 1 is a reproduction of this sketch.

He explained that in addition to their standard package, Microsoft supplies a localisation per country, which contains language, legislation and tax settings. ‘On top of that’ FANDOR supplies an add-on, a piece of software meant to enhance and customise the ERP technology to the healthcare industry. This add-on is called the ‘Medical Appliances’ add-on. According to the strategy of the

9_{He is referring to a Dutch saying: ‘someone who wears a double cap/two caps’, which refers to someone}

who has two potentially conflicting roles. In this case that would be working for both HEALTHCO and FANDOR.

(19)

19

vendor, these standard products “should cover 90 to 95 per cent of the whole business process”. What remained had to be custom-made. In the words of the consultant:

“We have a standard system. We prefer to tinker with that as little as possible. As I said earlier, every custom-made piece takes us time, and thus costs the costumer money. We rather do not want that, […] we have a ready-made system. We only need some configuration.”

The FANDOR consultant knew the inner workings of his technology to a high extent. However, configuration was needed to create a working ERP system within the HEALTHCO organisation. Since “supplying [medical] shoes is completely different than supplying gas appliances, different steps are desired.”

In order to configure the system, both knowledge of the technology and knowledge of the processes were needed. Although “through [his] experience with other companies, [he knew] a little about how

such processes are, since there are many similar processes”, the consultant had little knowledge of the processes within HEALTHCO. On the other hand, actors within HEALTHCO had no knowledge of the workings of the ERP technology FANDOR offered and how to adjust this. They, however, did know how their processes worked, since they had worked with these processes for quite some time. In the words of the project coordinator: “Look, we, as a supplier, we arrive with a product, but the client is the one with knowledge of its own processes. Those two have to come together.”

The actor-network surrounding the ERP system had to overcome this knowledge barrier to become able to open the black box and adjust its inner workings to such an extent that the system would produce the right output:

“[To configure the system] I need to hear from them something like: ‘this is my process, from a to z’. The only thing I have to create is a way to get to the same outputs [as in the old situation]. The essential output in this case would be the invoice. […] That’s the only handhold I’ve got…”

Altogether, the relevant actors in the network surrounding the ERP system and the alliance networks they form at the start of the implementation are shown in Figure 2. While consultants were sometimes referred to as individual actors, they were mostly referred to as a group with convergent (if not the same) interests. The parent company, although consisting of several individual actors from different locations, appeared to be a tightly aligned network. For this reason, these networks are depicted as one punctualised actor.

(20)

20 The black box of the absent ERP system

In order to become a fully operational ERP system, the system needed configuration. Thus far in this story, this configuration did not yet take place. However, both the CEO of HEALTHCO and FANDOR’s consultants had already acted upon the system. They did so in absence of the system itself, since it was not yet there. The technology was there, the processes it needed to cover were there, we may even carefully presume that the knowledge to implement this ERP system was there. The ERP system, however, was not yet there. Despite the claims made by FANDOR’s lead consultant that he delivered a ready-made system, he did not. To become operative, NAV required considerable customisation.

Similarly, HEALTHCO’s CEO convinced the parent company of this specific system by showing them potential features and virtues of NAV but in its absence, for NAV is not yet a working technology. They were able to discuss it without knowing what exactly it constituted. The system could only be referred to as an ideal, and ideals are abstract by definition, “they do not become ‘real’ unless operationalized, enacted, and practiced. [However, this] allow[s] it to forge relations with potential users and other […] technologies [...].” (Quattrone & Hopper, 2006, p. 224).

“Presence, in short, depends upon absence (just as absence depends on presence). […]An object is a presence. It is present, here and now. But, whatever the form of its presence, this also implies a set of absences. The present object implies realities that are necessarily absent, that cannot be brought to presence; that are othered. So, to put it slightly differently, an object is a pattern of presences and absences.” (Law & Singleton, 2005, pp. 342-343)

(21)

21

To become an object, NAV required enactment. Since it was referred to, discussed, and acted upon, it became real. (cf. Quattrone & Hopper, 2005) However, it did not yet possess realised features; the actual functionality remained rather vague. While the object was visible, its content was not.

Since the features and inner workings of the system were not yet clear, the ERP system could be depicted as a black box. A black box is the evidence that the actor-network is stable and that the actors are working towards collective goals. In Latour’s (1987, pp. 130-131) words: “The assembly of disorderly and unreliable allies is […] slowly turned into something that resembles an organized whole.” This black box, and the actor-network it was part of, constitute the starting point of this case study.

The implementation of the ERP system at HEALTHCO can be treated as the process of enrolment of various actors within the organisation into the network of supporters of the ERP system. However, for the success of the implementation, the ERP system first had to be de-punctualised, in order to enrol other actors in its network. In its present form, the interests of the system and the interests of other actors within HEALTHCO were not aligned. If this would not change, the system could not be enrolled in the network representing the envisioned post-implementation state. Not being able to assemble the actor-network would imply failure of the implementation.

The creation of new alliance networks

Like all black boxes, the black box of the ERP system was leaky. The system had until this far been punctualised within a larger actor-network. Since the implementation now could take place, actors were about to interact with the system. While the actor-network as an ideal was taken for granted, the system would become ‘real’ by the actions of several actors.

The system began to translate due to the formation of two alliance networks. These networks were defined early on in the implementation process. By following the creation of these networks I will show how these formations gradually made the system more visible. It is important to note here that the definition of these two networks as a group is given by the actors themselves. Actors often wish to define groups and definitions by actors weigh more than those given by researchers (Hyvönen et al., 2008).

The interessement and enrolment of the key-users

FANDOR’s lead consultant explained that their implementations always started with an overview: “That’s sort of a kick-off, like: ‘Guys, look, we have a new system’.” Such an overview also took place within this implementation, in March, 2013. The overview was not a single session, but consisted of multiple sessions, spread over two days. At the overview, the key-users were present for the first time.

“[Key-users] are the people with whom you set the project in motion, who have to know the ins and outs and who also, in the last phase, have to deal with the training of the end-users.” (Project coordinator)

(22)

22

“[The overview] is like a training, based on our own experiences in the field, and our own test environment. We present the [technology], […] for example, there has been an overview [where a consultant] has shown how [the technology] works by using wheelchairs, because we just don’t know their processes yet. But his way they do get an insight in what the system does, what it can do. ” (Lead consultant FANDOR; emphasis added)

This overview included a problematisation moment, since the consultants showed the client how the ERP system would be the solution to their problems. More important, it also contained an interessement process where the first links between the interests of the different key-users and the ERP system were established.

As a non-human actor, the ERP system needed spokespeople. These were the vendor’s consultants, whose ‘experiences in the field’ indicate a strong tie with the ERP technology. By means of this overview, they attempted to establish a link between the key-users and this technology. The CEO of HEALTHCO, to the key-users an even more influential actor, considering his strong ties with these actors, also acted as a spokesperson for the system:

“[I]n fact, it is a presentation organised by the supplier, FANDOR, where… Well, no, it’s more a combined effort. The supplier explains how the project will work, what is expected from the other party and what they can expect from FANDOR. […] On the other side, […] you see that, in this case [the CEO] also tells his own people: ‘Hey’, just to give it a boost, ‘guys, we are going to start, now it is really going to happen. In the period ahead, instead of a 100 per cent, we have to give it 150 per cent or 200 per cent, so to speak.’ You know, that kind of pep talk.” (Project coordinator)

In this overview, the output of the system was introduced and explained, as well as how the system was supposed to be going to work. The explanation by the consultants made the system more real, but only to a certain extent. They but slightly lifted the lid of the black box, which slowly began to leak, which allowed the key-users to look inside. Nevertheless, the black box was only but a chink open, since a great deal of the system remained obscure, or even hidden.

After this overview they started with what they called the ‘workshop method’. In these workshops, consultants not only trained key-users to use the system, also the ERP system’s configuration took place here.This method consisted of a number of sessions, which each dealt with different parts of the processes that needed to be covered by the system. It depended on the subject discussed in a specific workshop, which key-users would be present. For every workshop, the project coordinator decided who would join in, i.e. he was an obligatory passage point11_{(Callon, 1986) for actors joining}

the actor-network of the key-users.

“For every part of the system we say: ‘OK, these key users are needed for this, it is important that they get the knowledge needed’. We are then going to educate them, take them through the system, give them a sort of training. In those workshops we follow a hands-on approach, so they also get the feeling: ‘OK, the system works this way’ and ‘I should think about these things’. […] That’s our way, that’s how we work” (lead consultant FANDOR)

(23)

23

This also meant that every workshop began with a new problematisation phase (and triggered a new translation). Whenever a new subject was dealt with, they in fact dealt with a sub-problem of the bigger problem, each time looking for a solution to overcome this problem. However, the supplier’s consultants did not always have a clear-cut solution for every sub-problem. This solution had to be found by the interaction of multiple actors, being the ERP system, the consultants and the key-users. In this process, these actors negotiated with each other and translated each other’s interests. The process of configuring the system thus can be seen as a great number of small translations, all dealing with specific problems.

“What we, consultants12_{, do most of the time is, we set up a basic system in which we can test, so we can at}

least show something, achieve something. At that moment, we do not know how the processes within such a company are. Actually, we assume […] what a standard company would do. You’ll see that throughout the workshop the processes of the company are brought to the surface, and the key-users know how these processes work, and recognise certain matters. Like: ‘Ah, right, with the financial configuration I should pay attention to this, and this’. Then we can advise them about that configuration…” (Project coordinator)

“Naturally, [the key-users] have to be guided in [the configuration] process. Sometimes... No, not sometimes… Often, they’re sent in a certain direction, with our knowledge of the add-on, to make certain choices. But in the end, the client is responsible for how [the system] is configured and how it will work. […]” (Project coordinator)

Since every workshop again dealt with different problems, the spokespeople of the ERP system had to convince key-users to align their interests with those of the ERP system, workshop after workshop.

“[Convincing key-users] is something you try in all workshops. It starts at the overview, but in every workshop you try to convince them to use [the system]. Like: ‘When you do it this way, you have one integrated system, customers only have to be inputted once, insurance company gets retrieved, you have no worries about how you go through the process, because you are being led [by the system]’. So, [I am] continually promoting and selling to let them work with it as good and as quickly as possible.” (Lead consultant FANDOR)

However, this was not such an easy task as his quote implies, since the ERP system did not want to be ‘tinkered with’, it did not easily consent to being translated:

“[W]e want to stay as close as possible to the standard solution [...]. Every, um, notion of customisation should stay out of the presentation, we have to be very strict about this. We should have an attitude like: ‘This is what the product is, this is standard, and something else is not really possible’. As a consultant, you're used to provide assistance […] so we have to work a bit against our nature […]” (Lead consultant FANDOR)

“Intuitively, it will never be as pleasant to work with as something that is custom-made. […] Such things can create resistance. […] Almost by definition, working with a standard system will be more cumbersome.” (Project coordinator)

12_{As explained earlier on, in this particular implementation, this project coordinator does not have the role}

(24)

24

In order to “create key-user commitment” (i.e. align the interests of the key-users with that of the system) and mobilise the aligned interests of both the ERP system and the key-users, the key-users produced inscriptions after every process of configuration. While the consultant called these inscriptions manuals, the project coordinator refused to call them manuals, as if he just had spoken with Latour himself: “I prefer to call it ‘job instruction’. […] A manual tells you how the system works; a job instruction tells you how to work, with the system.” This quote implies an interaction between these instructions, the ERP system and the key-users, since job instructions were said to show users and the system how they should interact. The job instructions did, however, depict the ERP system as a passive tool instead of an actor.

“[When you have] a new patient and an enquiry for a new device, you need those two job instructions. You need this one for adding the patient [to the system] and this one for the enquiry [for a new device]. They have to follow [the job instructions] step by step.”

The interaction between the key-users and the ERP system, often through multiple spokespeople such as the project coordinator or the consultants, made the system more ‘real’. Little by little, the system became less obscure, and thus more visible. Also inscribing the translations made the ERP system more visible, since these inscriptions escorted the ERP system in its travels (Callon, 1991) and could show other actors how to interact with the system.

The interessement and enrolment of the steering committee

Apart from the key-user network, another alliance network was formed. This alliance network was called the ‘steering committee’ and consisted of the CEO of HEALTHCO, several executives from within the parent company and the project coordinator. Two other actors were named as part of this committee: FANDOR’s project manager and a management director of FANDOR. However, since these were not (yet) enrolled in this network, they were no part of it. According to Latour (2005, p. 34), “social aggregates are not the object of an ostensive definition—like mugs and cats and chairs that can be pointed at by the index finger—but a performative definition”.

The steering committee had to define the strategic framework in which the key-users were allowed to make configuration choices. Also, they had to resolve problems the key-user group could not resolve. In the course of this study, one steering committee meeting had taken place in April, 2013. Four other meetings were planned. Every meeting can be seen as a separate translation where the actors needed to be re-enrolled in order to ‘keep [their] support’. These meetings are held with the use of WebEx, a videoconferencing application.

“[I]n such a steering committee they operate at a higher level: do I achieve my budget? What are the pitfalls? Risks, what are we up against? That sort of stuff is discussed within the steering committee. […] Most of the time they talk about money, and about the status of the project. Sometimes, when things do not run smoothly, [FANDOR’s] project leader reports this to the steering committee. They talk about things like: OK, we bought this product and assumed it could do this and this, that’s what we’re told. Now we see it’s not there, how are we going to solve this? That’s real ‘steering committee-talk’.” (Lead consultant FANDOR)

(25)

25

“Our structure is very complex, but we use several services from [the parent company], so their support is needed. […]It’s important to keep [their] support and to ensure a positive attitude of them towards the project. We do this by involving them in the decision-making process and by informing them.” (CEO HEALTHCO)

So, how did this make the ERP system more visible? This had to do with the informational needs of the parent company in order to be enrolled, since these were abundant. “Unfortunately, that’s part of the game. I say unfortunately, but it does mean it’s thoroughly examined and well-organised.” HEALTHCO’s CEO showed several documents, coming from a blue binder out of his office cabinet. These documents contained inscripted communication with the supplier, but mostly with the parent company. Many of these documents described characteristics of the ERP system, either realised or envisioned. The collection and inscription of this information made the black box more visible (cf. Chua, 1995). The documents provided to the parent company became interessement devices, both rendering parts of the system visible and tying together the system’s interests with the parent company’s interests.

Towards a more visible ERP system

In short, the formation of these two alliances has rendered the ERP system more visible, yet complete visibility was never realised. I believe this never will be realised, since the parts of the system that were opened, were also shut again when the network was satisfied with its alignment. Nevertheless, the formation of these alliances did achieve a situation where parts of the system could be (re-)assessed and configured. It also made the ERP system more influential, since power is contingent upon the successful enrolment of actors (Latour, 1986). Moreover, since it now has more and stronger ties within other actors, it will be easier to eventually establish connections with even more organisational actors.

(26)

26 Punctualisation of actor-networks

Let us consider the subject of punctualisation by drawing on the formation of the actor-network of the key-users. Punctualisation offers a way to draw on ‘networks of the social’ without having to deal with endless complexity (Law, 1992). By addressing the ‘key-user group’, a consultant does not need to address each key-user individually. Heterogeneous actors are thus translated into a seemingly homogeneous actor. By assigning the same interests and roles to the different key-users, these heterogeneous actors became punctualised. When an actor-network is punctualised, a black box is created. This renders the content invisible and sets the focus on the inputs and outputs of the actor-network. Therefore, using punctualisation the consultants can draw on the key-user group without dealing with all individuals separately, much like why I, in turn, punctualise these consultants for this research.

Punctualisation, however, is precarious, due to the fact that actors within the actor-network can desert the network, actors from outside the network can join the network, or alliances within the

(27)

27

network can change. Moreover, when differences in interests are not acknowledged, one might miss important information. The project coordinator addresses this heterogeneity among the key-users: “Well, [The key-user group] is somewhat diffuse, because if you look at the key-users that are there, they come in all sorts of levels in terms of knowledge and the ability to think along, how they act in such a process. So, […] a key-user group is defined, but you can’t say that everyone participates at an equal level. […]You can’t say: we measure everyone by the same standards. For example, there are now three key-users who are there to learn how it all works, and indeed have to pass on this knowledge and, when we go live, answer questions of others, but these are not the people you can expect to have a ‘helicopter view’ and can create a work-around13_{for certain matters that do not totally fit […].”}

Initially this actor-network of key-users failed to act as promised, and thus the network collapsed. The only way to know why the actor-network failed was to look inside, which was now possible, since the actor-network had broken down:

“The key-users are the experts in the daily routines. [However,] some have, unfortunately, only a limited insight in the interaction between departments. […] [W]e noticed this only after we had started. […] People were too much focused on their own department. Look, something can be easy to work with within a department, but can sometimes obstruct cooperation between departments.” (HEALTHCO’s CEO)

Only after looking inside the black box, it became obvious that some of the key-users were unable to take on a ‘helicopter view’ and thus construct an interface between different departments. This ability is essential for the implementation of an ERP system, since such systems aim to integrate business functions. They eventually fixed this by adding an actor to the network of key-users. “Now there is a key-user who was not really appointed as one, but is present in almost every session. [She] knows the processes very well and is really needed.” (The project coordinator)

“Fairly quickly after the start [of the implementation] she joined the key-users. The commitment of [this] manager customer support really is inevitable.” (HEALTHCO’s CEO)

This shows that when actors punctualise an entire network into a node, this can lead to a failure to examine the composition of a group or team. Punctualising an actor-network into a black box may degenerate it into a failing network without being able to assess why it fails. In order to understand why a black box does not do what we want it to do, we have to open it and review its contents. The reopening of the black box of key-users was vital to the translation of the ERP system. Not being able to establish a ‘helicopter view’ to achieve cooperation between departments would be disastrous for the translation of the system, for both the promise of the ERP system and the aim of the CEO were to achieve integration.

Stalling management control decisions

ERP systems have the ability to affect many organisational practices, such as management control (Dechow & Mouritsen, 2005). Surprisingly, despite the importance attributed to management control by the CEO, in the beginning phase of the implementation management control issues did not seem a priority. Management control was not (yet) enrolled in the network of the ERP system.

13_{A plan or method to circumvent a problem (as in computer software) without eliminating it.}

Opening the black box of ERP: The implementation of an ERP system and the effect on management control