KM Resources: Organizational Memory and the Completeness of Process

Organizational Memory and the Completeness of Process Modeling in ERP Systems:
Some Concerns, Methods and Directions for Future Research.

Please Note: To view the tables and figures which accompany this article you will need to download the Microsoft Word version of this article.

Eveline Van Stijn is a Ph.D. candidate at the Faculty of Technology & Management, University of Twente, Netherlands. She received her M.Sc. degree in Industrial Engineering from the University of Twente. She specialized in information management as well as in logistic management. The current research of Eveline Van Stijn focuses on the further development of the organizational memory mismatch approach applied to the implementation stage of ERP systems.

Anthony K.P. Wensley is Associate Professor of Accounting and Information Systems at the Rotman School of Management, University of Toronto, Canada. He teaches Information Systems, Financial Reporting and Decision Analysis. His research is in the areas of complex information systems, knowledge management, and e-commerce. He has been published in the International Journal of Project Management, the Journal of Information Technology, Expert Systems with Applications and Auditing and has authored chapters in books on Knowledge Management and Information Systems Strategy. Anthony Wensley is the Executive Editor of Knowledge & Process Management and on the Editorial Boards of the Journal of Information Systems, the Journal of Global Information Technology, and the Journal of Global Information Technology Management.

---

Organizational Memory and the Completeness of Process Modeling in ERP Systems:
Some Concerns, Methods and Directions for Future Research.

Eveline Van Stijn, Faculty of Technology and Management, University of Twente, Netherlands

Anthony Wensley, J.L. Rotman School of Management, University of Toronto, Canada

Abstract
Enterprise Resource Planning (ERP) systems not only have a broad functional scope promising to support many different business processes. They also embed many different aspects of the company's organizational memory. Disparities can exist between those memory contents in the ERP system and related contents in other memory media, such as individuals' memories, and the organizational structure and culture. It is our contention that, in general, these disparities or memory mismatches, as we will refer to them lead to under-performance of ERP systems.

In this paper we focus on potential memory mismatches that may arise with respect to the embedding of process knowledge within ERP packages. Packages such as SAP provide a varied and rich environment for process modeling. However, we suspect that there are still many instances where process knowledge is either lost or represented in different ways in different parts of the organization. As we will discuss, the results of such memory mismatches will often not become evident until the system is in-use.

The overall thrust of the paper is to identify a variety of concerns, intriguing questions and avenues for future research.

Key words and phrases: ERP systems, knowledge management, organizational memory mismatches, business process modeling, organizational change, organizational learning.

Introduction

In this paper we focus on problems that may arise after ERP systems have been implemented - the in-use phase as we will refer to it. Various problems can be identified regarding the ERP systems in-use. Because of the organizational unwillingness or inability to make technology upgrades (Markus and Tanis, 2000), the enterprise system may take on the appearance of a legacy system in disguise. Furthermore, the users may still be working around the system or maintaining old procedures, instead of learning the relevant ERP capabilities (Marcus and Tanis, 2000). These problems are not only technological and organizational in nature, but they also involve cognitive aspects, such as adjusting existing work methods, mental models, and data-models. Hence, to solve the problems and enhance the ERP system successfully, it is necessary to view the ERP system in a broad sense, including technological, organizational, and cognitive aspects.

Clearly there are many areas where the knowledge embedded in the ERP system may conflict with existing knowledge residing in organizational memory. We focus in this paper on the knowledge that relates to processes. Such knowledge may reside in many places within the organization. Some process knowledge is embedded in the way the activities that constitute processes are structured both temporally and spatially. Other knowledge may be recorded in process manuals that may record 'ideal' type processes as well as details of the functioning of processes on a regular basis. Yet other knowledge may reside in the heads of individuals who work directly with the processes themselves or in automated activities or sub-processes of the process concerned.

We provide a new approach to understanding why organizations' ERP systems may under-perform and provide an initial indication as to how organizations can enhance their ERP system in order to better realize the intended benefits. We adopt an organizational memory perspective for our investigation, because it integrates the technological, organizational, and cognitive aspects of the ERP development. In common with the structuration theory of IT (cf. Orlikowski and Robey, 1991), organizational memory theory places information systems in the context of human action, the organization, and organizational cognition.

ERP systems can be viewed as part of the organizational memory of an organization (cf. Stein and Zwass, 1995; Wijnhoven, 1999), with contents related to a diverse range of organizational memory contents located at other memory media, such as organizational processes, structure, and culture. This perspective allows us to conceptualize ways in which the knowledge embedded in the ERP system may be in conflict with other organizational knowledge - in particular, process knowledge. Based upon this organizational memory perspective, we develop what we call the organizational memory mismatch approach. Organizational memory mismatches are discrepancies between organizational memory contents located in the ERP system and related contents stored at other organizational memory media. Such memory mismatches cause under-performance of the ERP system which leads to a need for coping. Coping strategies are varied and may involve further enhancements to the ERP system or a variety of other strategies that we will only be able to address in a very cursory fashion in this paper but that are the focus of future research work.

Previous research on ERP systems and some initial concerns about process knowledge and its representation

Defining ERP systems
Before discussing the ERP implementation research, let us reflect upon the question of what an ERP system actually is. The functionality of ERP systems gives a first indication. Some functions of SAP R/3 are listed in Table 1. Currently, ERP systems are evolving into extended-ERP systems, embedding and providing support for inter-organizational processes such as e-commerce, and customer relationship management (CRM) (Kumar and Van Hillegersberg, 2000). Such added functionality is not considered part of the ERP system in this paper as we focus on internal organizational processes. However, we would note that the issues we raise about memory mismatches with respect to organizational processes and process knowledge become both more significant and more complex when ERP systems span organizations and process become inter-organizational processes.

Insert Table 1 about here

An important underlying assumption of ERP systems is that the tight/tighter integration of business processes should lead to higher organizational effectiveness. This integration between business processes, organization units, and information is visualized by SAP (ARIS, 2000) as shown in Figure 1. In R/3, organizational units control business processes and information flow using master and control data.

Insert Figure 1 about here

The integration of business processes is mainly realized by means of the R/3 applications as illustrated in Table 1 and Figure 1, as well as SAP Business Workflow. SAP Business Workflow is the workflow system of SAP AG, that is part of the SAP R/3 system. SAP Business Workflow is based on a Workflow Manager residing in the SAP R/3 BASIS system. The SAP Business Workflow Manager imposes the defined process logic, creates work items, determines the appropriate routing criteria, monitors deadlines, tracks the work item and maintains status information. SAP Business Workflow has components for workflow definition, workflow execution including deadline scheduling and escalation, workflow to application interfaces, and user interface (workflow inbox) (ASAP, 2000).

The question remains how many organizations actually achieve the proposed enterprise-wide integration. To what extent is the necessary process knowledge available within the organization in the first place? To what extent can it be 'discovered' through interaction with the individuals who participate in the process or manage the process? To what extent can it be represented using the process representation tools available in ERP systems? There are also likely to be significant problems when organizations implement limited functionality, thus not really realizing an enterprise-wide system. The process knowledge embedded in the ERP system may be inconsistent with the process knowledge retained in other parts of the organization leading to imperfect integration.

A second key premise underlying ERP systems is that they embody best practices in their reference models (Davenport, 1998; Kumar and Van Hillegersberg, 2000). "[…] Reference models supposedly reflect preferred business models including underlying data and process models as well as organizational structures (Kumar and Van Hillegersberg, 2000, p. 25)." Reference models are process models (often in broad sense, including function, data, and organization models (Scheer, 1998)) that are available from third parties, for example they may be supplied by SAP or ERP consultants. It is assumed that such reference models are based on theoretical and practical assumptions regarding best practices for a given process.

One can maintain that the reference models document the process know-how and in this way encapsulate such best practices (Scheer, 1998). Reference models may be adapted to company-specific requirements, thus evolving into company-specific models. The use of reference models may well speed up the implementation process (Scheer, 1998). Furthermore, "[…] reference models provided by software vendors as software documentation […] benefit the customer by utilizing business process know-how, providing the opportunity to compare business software solutions, or pinpointing positive or negative implementation issues (Scheer, 1998, p. 61)."

It may be noted that one of the potential problems with best-practice reference models is the problem of context. Just where do reference models apply? What are the relevant aspects of the business environment that need to be present for a reference model to actually constitute best practice? Processes exist within a rich context that includes aspects of the organizations products and services, its customers and suppliers, its employees and the organizational structure and culture. The danger is that much of this contextual information may be missing from the ERP system - it is often tacit and embedded in other organizational memory media. As a result the assumed context of the reference processes may conflict with the actual context in which the reference processes are actually implemented leading to under-performance of the ERP system. Furthermore, given the tacit nature of much of this contextual knowledge the specific reasons (in terms of knowledge conflicts or knowledge gaps) may be very difficult to diagnose or correct.

Another aspect of ERP systems is that they are primarily directed at the goal-attainment of the organization, being based on a rational and control oriented approach to modeling and managing organizations. An information system aiming at goal-attainment should provide support to manage the organization based on formulated goals and other performance criteria (Van Stijn, 1999). In this context, effectiveness may be defined in terms of such goal-attainment (cf. Bolwijn and Kumpe, 1990).

Other aspects of effectiveness that can be identified are adaptation and pattern maintenance (Quinn and Rohrbaugh, 1983; Stein and Zwass, 1995). Information systems focusing on adaptation should support the organization's attempts to adjust to the environment (Van Stijn, 1999). Finally, information systems directed at pattern maintenance should maintain the patterns and procedures that are anchored in the organization's culture (Van Stijn, 1999). Thus, ERP systems may be distinguished from other types of information systems, that may focus on adaptation, such as data mining systems, e-business, supply chain management systems, and systems that focus on the pattern maintenance, such as communication systems and skills index databases. It is worth noting that some process within the organization will be directed towards adaptation while others will focus on pattern maintenance. It is an interesting research question to determine to what extent ERP systems support these different types of processes. We certainly suggest later on in this paper that ERP systems may not adequately support the process that allow for the adaptation of processes as the organization and/or its environment changes.

As an interesting aside a significant issue with the implementation of ERP systems is the determination of the extent to which organizational processes need to be adjusted to those represented in the ERP system and the extent to which the ERP system's representations need to be adjusted. One way of approaching this dilemma is through a clearer understanding of the relationships between the organization's structure and processes and its culture. As we have noted above organizational processes embed considerable knowledge that can often only be 'decoded' through an extensive investigation of the context within which the process operate. The nature of this 'decoding' would seem to be a fruitful and creative area for future research.

In short, ERP systems are information systems aiming at organizational goal-attainment by integrating internal business processes, that may be shaped according to the vendor's best practices (implicitly) embedded in the reference models. An ERP system is thus more than information technology, because it also involves organizational and cognitive aspects (Van Stijn et al., 2000). That this broader understanding of ERP systems has an its impact on understanding the representation of process knowledge is further stressed and elaborated upon in the rest of this paper.

In the following section we review some of the prior research into the implementation stage of ERP systems. We argue that much of this research essentially misses the point for a number of reasons. In the first place it addresses issues relating to organizational memory and what we term 'memory mismatch' only tangentially. Secondly, we note that many of the problems associated with memory mismatch are more likely to arise in the in-use stage rather than the implementation stage.

Prior research on the implementation of ERP systems

Current ERP research has primarily focused on the ERP implementation stage, this stage being seen as an 'obstacle' to overcome first. However, it is self-evident that it is only after the ERP system has been implemented and is actually deployed or utilized that any success can be achieved (Fichman and Kemerer, 1999; Lassila and Brancheau, 1999). Some researchers discuss the implementation process itself. For instance, Kirchmer (Kirchmer, 1999) provides a normative model that describes how organizations should execute an ERP implementation (based upon the software supplied by SAP AG). Proposed descriptive models - that discuss how different organizations are actually implementing their ERP systems- are for example the structurational model (Volkoff, 1999) and the framework for organizational change (Boudreau and Robey, 1999).

Other researchers identify factors which contribute to successful and unsuccessful ERP implementations (e.g. Holland, Light and Gibson, 1999; Scott, 1999). Among those identified critical success factors are top management commitment, strategic vision, and training of users. None of these approaches explicitly recognize the need to identify the extent to which process knowledge is distributed across different organizational memory media. Further the implications of representing some of this knowledge in the ERP system and the subsequent interaction of this knowledge with process knowledge resident in other media is not considered.

Prior research on ERP performance measurement is scarce (cf. Rosemann and Wiese, 1999), and has mostly been conducted by practitioners, e.g. Deloitte Consulting (Deloitte, 2000). Implementation success is often measured in terms of cost and duration of the implementation (Bingi, Sharma and Godla, 1999). However, the overarching objectives associated with implementing ERP systems are to realize the promised benefits of enterprise systems. Typically, these benefits are in the form of reduced cycle times, reduced inventory costs, increased agility, or improvements in the availability of strategic decision information (Bingi, Sharma and Godla, 1999; Davenport, 2000). These benefits can clearly only be assessed during the in-use stage.

ERP packages also result in the formal representation of much of the knowledge of the organization as it relates to organizational strategy, structure, processes and so on. Thus ERP packages may be seen as contributors both to the capture and management of knowledge. There is a need to determine the success with which such knowledge is actually captured by the system. However, a word of caution is necessary here since knowledge may be refined, expanded and sometimes discarded during the implementation phase. Thus, there is a need to assess the extent to which actual pre-existing knowledge is appropriately represented in the ERP system.

As with many information technologies (Boudreau, M.C. and Robey, 1999), results of the ERP implementation efforts range anywhere from extreme failures to extreme successes. Some projects are abandoned before the actual go-live date, or are assumed to have played an important role in the bankruptcy of a company, such as FoxMeyer (cf. Scott, 1999). Other companies proudly report their successes, often through their software vendors and consultants, such as SAP AG (SAP, 2000). "Whereas the company used to require two weeks on average to ship to customers, 98 percent of products are now shipped within twenty-four hours. Financial closing times were cut in half, from twelve days to six. […] It has saved more on reduced inventory alone than its SAP system cost to install (Davenport, 2000, p. 7)." There is a danger here that improved performance in the short term may not be the result of improved process knowledge. There is a need for significantly more research into the location, nature and extent of process knowledge both before and after the implementation of ERP systems. Care must be taking to investigate how process knowledge stored on different knowledge media interact both before the implementation of the ERP system and after its implementation. It is also worth observing that organizations are likely to have both formal and information knowledge processes for maintaining and enhancing process knowledge. These processes must be identified and reconstituted in the post-ERP organization.

Research on ERP systems in-use concentrates on ERP performance evaluation and on identification of usage stage activities and problems. ERP performance may, for example, be evaluated based upon the Balanced Scorecard method (Rosemann and Wiese, 1999), measuring the degree to which the intended goals are actually obtained. An ERP performance evaluation helps in identifying problems and opportunities for further development of the enterprise system. It is important to note that such goals are dynamic and thus require that the performance measures evolve over time as well. Potential activities, problems and errors that may occur after the system has gone live have also been identified (Davenport, 2000; Marcus and Tanis, 2000), some of which are listed in Table 2.

Insert Table 2 about here.

From a process perspective we may identify a variety of ways in which organizational process may under-perform after the ERP system has gone-live. Processes may generate an unacceptable level of errors, they may be unstable and have performance that is difficult to predict, cycle times may be much longer than anticipated. Processes may also fail in unpredictable ways and may be difficult to trouble shoot and correct.

Although the need for ERP systems to respond to both major and minor changes in the organization and its environment after going live is undeniable there is a paucity of available research literature in this area. We would signal this as another fertile area for future research. Interesting research questions in this area are: What are the triggers for enhancement of ERP systems? In what ways can (should) the ERP system be enhanced to respond to these triggers? What knowledge is required of the organization and its members in order to both identify appropriate triggers and define and implement appropriate responses to them? As we note later in this paper, to the extent that the triggers relate to 'mismatches' between the contents of the organization's memory a variety of coping behaviors may be appropriate.

In the following section we discuss in detail the memory mismatch approach. Our memory mismatch approach seeks to provide a framework for classifying ways in which an implemented ERP system is broadly in conflict with some aspects of organizational memory as it exists at the time of implementation. We do not consider in the present paper a discussion of the processes that lead (or should lead) to modification and enhancement of organizational memory that may be missing or in conflict with processes that are in place to enhance ERP systems after they have gone live.

Although we introduce the organizational memory mismatch approach in a general way its applicability to specific contents of organizational memory as they relate to processes and process knowledge should be fairly clear.

The organizational memory mismatch approach

Organizational memory theory
Organizational memory may be defined as "[...] stored information from an organization's history that can be brought to bear on present decisions (Walsh and Ungson, 1991, p.61)." Next to information, other types of memory contents can be included, for instance knowledge (Stein, 1995) and paradigms (Wijnhoven, 1999). The memory contents may be stored at different locations or repositories (Walsh and Ungson, 1991; Wijnhoven, 1999). Organizational memory processes, such as search and retrieval (Stein, 1995), operate upon the memory base, thus enabling the actual use of the memory contents. These three aspects of organizational memory, contents, repositories and processes, are further discussed in the next subsections.

Organizational memory contents Organizational memory contents are the cognitive elements that form the memory base. Different authors label and classify the memory contents differently (Moorman and Miner, 1997; Robey, Wishart, and Rodriguez-Diaz, 1995; Stein, 1995; Walsh and Ungson, 1991). One may, however, distinguish four separate, more general types of memory contents, called information, knowledge, paradigms and skills here. Information is "[…] the flow of messages, while knowledge is created and organized by the very flow of information, anchored on the commitment and belief of its holder (Nonaka, 1994, p.15)." Thus, information may become knowledge when the receiver interprets the messages.

One may also put this distinction between information and knowledge (the second content type) differently: "[…] In the managerial, cognitive perspective, managers are assumed to be "information workers". That is, they spend their time absorbing, processing, and disseminating information about issues, opportunities, and problems. The most fundamental challenge faced by managers, however, is that their information worlds are extremely complex, ambiguous. […] Somehow they must see their way through what may be a bewildering flow of information to make decisions and solve problems. [They] meet this information challenge by employing knowledge structures to represent their information worlds and thus, facilitate information processing and decision making. […] A knowledge structure is a mental template that individuals impose on an information environment to give it form and meaning (Walsh, 1995, pp. 280-281)." Knowledge structures thus represent what are called 'interpretive schemes' in structuration theory, shared stocks of knowledge which help human actors to give the world meaning (Orlikowski and Robey, 1991). It is interesting to contemplate to what extent the knowledge structures that have been built up by individuals prior to the implementation of an ERP system are appropriate after the implementation of the system - do they allow individuals to behave appropriately? Can they work with the newly reconstituted processes, are they able to diagnose process failures or performance deviations appropriately?

The third content type, paradigms, consist of the organizational beliefs, governing values and norms (Wijnhoven, 1999). As structuration theory's 'norms', paradigms represent the beliefs and rules about 'what is good and what is bad', about what one should and should not do. The fourth content type, skills, are comparable to what some refer to as tacit (Nonaka, 1994) or soft knowledge (Anand, Manz, and Glick, 1998). Skills are capabilities of people, 'how they do things'. These capabilities thus have a personal quality, deeply rooted in action, commitment, and involvement in a specific context (Nonaka, 1994). Only if individual members of the organization are willing to and capable of sharing tacit knowledge or helping in the development of skills by others, is the organization able to have access to them.

The four identified content types may be independent of a specific application area or domain, but they may also be domain-dependent, depending on a specific business process, organizational unit, the organization in general, or the industry or the nation(s) in which the organization operates. Memory contents may be stored in one or more different retention media; these are discussed in the next subsection.

Organizational memory media Though some argue that such storage of memory contents may be interpreted metaphorically rather than literally, one can at least assume that the various repositories imply memory contents, such as knowledge and information. For instance business processes, or transformations, are based upon knowledge regarding what input is needed and what actions should be undertaken in order to produce a certain output. "[…] The logic that guides the transformation of an input into an output is embodied in these transformation (Walsh and Ungson, 1991, p. 65)." This logic may be called 'technological knowledge'. That is the knowledge about how to produce goods and services, understanding the effects of the input variables on the output (Bohn, 1994) The transformations occur throughout the organization and similarly, memory is preserved in a variety of procedures and formalized systems (Walsh and Ungson, 1991).

Next to transformations, Walsh and Ungson (Walsh and Ungson, 1991) describe the following storage media for organizational memory: individuals, culture, structure, ecology, external archives. The term ecology refers to the actual physical structure or workplace ecology of an organization (Walsh and Ungson, 1991). Such a physical setting often reflects the status hierarchy in the organization and helps to shape and reinforce behavior prescriptions in the organization. In a broader sense, other physical artifacts existing in the organization may be considered, including for instance the available machines, the products and product lines. Such physical artifacts "[…] embody, to varying degrees, the results of prior learning (Moorman and Miner, 1997, p. 93)." Additionally, information systems have been recognized as another important repository (Stein and Zwass, 1995; Wijnhoven, 1999). "[…] Information technology can also capture many routines stored in memory by embedding those routines within its programs and procedures. Through electronic storage, memory may become more accessible to organizational members (Robey, Wishart, and Rodriguez-Diaz, 1995, p.28)."

Strikingly, it should be noted that these identified storage media provide a more detailed overview of the organizational resources (individuals, ecology, information systems), realm of structure (culture, structure) and realm of action (transformations) identified in structuration theory (Orlikowski and Robey, 1991). Furthermore, it also places the organization in its environmental context by adding the external archives as another retention medium. An organization is associated with a number of stakeholders and other interested parties in its environment, for example former employees, competitors, and the government. Other parties involve companies that collect data on performance and sell this information to interested parties, news media and business historians (Walsh and Ungson, 1991).

Summarizing, memory contents may be stored at one or more different retention media, as illustrated in Table 3. These retention media have different opportunities and limitations for storing memory, and differ in speed, reliability, susceptibility to physical degeneration and availability (Wijnhoven, 1999). The next subsection deals with the organizational memory processes by means of which the organizational memory is actually realized and used.

Insert TABLE 3 about here

Organizational memory processes Organizational memory can be differentiated from general knowledge because it functions as a process and may be non-cognitive (Stein, 1995). In our opinion, this remark can be interpreted in the sense that the organizational memory base consists of the cognitive elements (memory contents). The media and the processes that operate on this memory base are non-cognitive. These defining processes of organizational memory are acquisition, retention, maintenance, and retrieval (Stein, 1995), as shown Figure 2.

Insert FIGURE 2 about here

Memory acquisition is the collection of new memory contents and memory maintenance is "[...] the process of adjusting existing memory to changed environments (application areas) in such a way that the basic part of the memory is still applicable despite these changes (Wijnhoven, 1999, pp. 172-173)." Memory maintenance, in other words, is about adapting and updating the memory. Other issues are forgetting obsolete memory, and integration of new memory with existing memory (Wijnhoven, 1999). Memory acquisition and memory maintenance together form the processes of organizational learning, for organizational learning is specifically concerned with the growth and change of organizational memory (Duncan and Weiss, 1979).

Memory retention is the storage of the memory contents in the memory media. Memory search and retrieval deals with finding and obtaining memory contents after storage.

Organizational memory mismatches
Based upon the previous discussion, an ERP system may be viewed as part of the organizational memory, being a retention medium (information system) that embeds memory contents. All four types of memory contents may be embedded in the ERP system. For example information regarding financial resources or technological knowledge regarding logistic planning are represented in the ERP system, e.g. logistic planning modules. Paradigms also underpin the ERP system, though they may be implicit for the user organization. For instance, paradigms concerning best practices (cf. Kumar and Van Hillegersberg, 2000) and effectiveness are included in e.g. inventory schedule modules. Skills could be included as well, either elicited in the form of routines or decision models, or in the form of a skill database in the human resource component of the ERP system, linking employees and skills.

It is our contention that organizational memory mismatches may exist between the memory contents of the ERP system and related memory contents in other memory media (Van Stijn and Wijnhoven, 2000). For instance, the sales planning component of the ERP system (the representation of the sales planning process) may be used to predict future sales based upon previous sales. However, the underlying assumption in the ERP system is that those sales in the past are representative for the future and that no specific 'events' have occurred that may alter the pattern. However, the sales manager may know that another company has started selling a similar product at a much lower price, which may be regarded as one of those events that disrupt the previous pattern. As a result a memory mismatch exists between the memory content of the ERP system and the memory content of the sales manager. Forecasts made with the ERP system may be systematically too high, which would have a negative consequence for the whole logistic and financial planning. Such a memory mismatch is very likely to lead to ERP under-performance, which means that the intended benefits of the system, and hence the organization as a whole, are not achieved.

Cognitive dissonance theory offers a starting point for the further definition of organizational memory mismatches. In his discussion of cognitive dissonance, the psychologist Festinger (Festinger, 1957) states that there are three possible relations between pairs of cognitive elements within an individual's mind, namely irrelevance, dissonance, and consonance. Irrelevance occurs when two elements have nothing to do with each other. When two elements are related to each other, they may either be consonant or dissonant. Two elements are dissonant if, they do not fit together, because they are inconsistent or contradictory (Festinger, 1957). For our discussion of memory mismatches, two extensions are made to Festinger's approach to cognitive dissonance. First, instead of comparing memory contents of one medium (the individual's mind), the memory contents of the ERP system are compared with those of other retention media. Related contents on the different media may be dissonant or consonant to each other. The second addition we make to Festinger's analysis is that we extend the concept of dissonance to include situations where memory contents are missing where they should be present and situations where memory contents are present on both media where only one instance of the memory content should be present (Van Stijn and Wijnhoven, 2000). We refer to the former situation as one of under-redundancy and the latter as over-redundancy. Thus, we distinguish three types of organizational memory mismatches, namely under-redundancy, inconsistency, and over-redundancy, as illustrated in Figure 3:

· "Type I. Under-redundancy
The memory content A in Figure 1 is missing in the other retention medium and memory content B is missing in the ERP system, but those memory contents should be present in both media.
· Type II. Over-redundancy
The memory content C in Figure 1 should not be duplicate (identical content present on both media), but should exist in either the ERP system or the other retention medium.
· Type III. Inconsistency
If for both media, memory content 'D' in Figure 1 should be the same, the memory content D in the ERP system is inconsistent with the memory content ~D (not D) in the other retention medium, and vice versa (Van Stijn and Wijnhoven, 2000)."

Insert FIGURE 3 about here

Such memory mismatches form the core of the organizational memory mismatch approach, where they are related to under-performance of the ERP system and coping behavior, i.e. further enhancement of the ERP system in broad sense. Organizational memory mismatches may be analyzed using the decision tree depicted in Figure 4. The memory contents of the ERP system are compared to the memory contents located at the other identified memory media. One can conclude that the organizational memory construct may be viewed as a further operationalization of the structuration theory, placing structuration theory from its social context to an organizational context.

Insert FIGURE 4 about here

Business process modeling and organizational memory revisited
The business process modeling sub-process is one of the sub-processes where it becomes particularly apparent that ERP systems and the related processes involve not only technological and organizational aspects, but also aspects that relate directly to organizational memory. In this paper, we mean by cognitive that it is related to the organizational memory. Since a complete discussion of all concerned organizational memory is beyond the scope of the paper, we will focus on the knowledge and paradigms with respect to the business process modeling. In his discussion of knowledge required for the ERP implementation, Chan (1999, p. 27) identifies three knowledge dimensions, namely:
1. Project management knowledge, referring to the knowledge required to manage the entire implementation process as a single project;
2. Business and management knowledge, referring to the knowledge about issues and knowledge to deal with those issues during/ after implementation. These issues are often people related and occur on a higher management level;
3. Technical knowledge, referring to the knowledge required to install and implement the ERP system.

It should be noted that the knowledge dimensions are defined rather vaguely, especially the second one. However, the discussion provides a more detailed division of the knowledge dimensions. Business process engineering (in the business and management knowledge dimension) and process engineering (in the technical knowledge dimension) (Chan, 1999) appear to be the two knowledge areas that are highly relevant to the business process modeling process. "[…] Business process engineering requires the organization to identify its business processes and its daily routines to carry out business functions. Organizations are becoming aware that understanding its' business processes has become increasingly important for competitive advantage. […] Organizations need to have the knowledge of how their business functions. This knowledge is critically important to the organization as it has to be aware of how it performs and how one process affects another in a domino effect (Chan, 1999, p. 30)."

Process engineering refers to "[…] knowledge required to map out the organization's business processes. […] Having this ability to map out the organization's tasks means that the organization can understand their business processes better and to have the capability to view the flow of information throughout the enterprise. Hence, organizations can determine what their critical information assets are and how to provide better leverage for them. The mapping of business processes may be assisted by the use of modeling tools from third-party vendors such as Visio, ARIS, or Intellicorp (Chan, 1999, p. 32)." Thus, the knowledge is related to the business modeling process as well as the outcome.

Process engineering knowledge may also be interpreted as being a skill of the modeler(s), to translate business process engineering knowledge into models that can be used to configure the ERP system and change the organization's processes. Business process engineering knowledge relates to the know-how, or the logic of the processes. Scheer (Scheer, 1998) argues that by modeling the business processes using ARIS, the process knowledge of the organization is stored and can be managed accordingly.

The key remark to be made here is that business process engineering knowledge, or process knowledge, may need to be interpreted in a broader sense, to include process paradigms, information and skills. All those cognitive elements are embedded in, or implied by, the ERP system as well, for instance regarding reference models, calculation methods and best practices. Process knowledge thus not only includes the know-how, or the process logic, identifying atomic tasks, when they are to be executed, and by which resources. It also includes know-why, referring to the explanations of why specifically those atomic tasks need to be executed, why in this order, and why by those resources.

It may also be necessary to answer questions regarding where tasks are executed. Furthermore, input/output relationships may be analyzed in detail. The level of detail may be dependent on the maturity, or stage, of the knowledge. Bohn (Bohn, 1994) identifies eight stages with respect to input/output knowledge, as are shown in Table 4. Next to eliciting the process knowledge, the process paradigms would be surfaced, giving insight in the beliefs, assumptions, values and norms underlying the processes.

Insert Table 4 about here

Process knowledge and paradigms in the organization are not necessarily the same as the process knowledge and paradigms underpinning the ERP system. When they do not match, one can speak of organizational memory mismatches (Van Stijn, 1999). If these mismatches are not reconciled, it is our contention that the organization will exhibit characteristic behaviors and is likely to under-perform in the usage stage of the ERP system(Van Stijn and Wijnhoven, 2000).

Discussion of the organizational memory mismatch approach

One question we would like to pose here is what is effective coping behavior? In other words, when does the further development of the ERP system in the broad sense, lead to performance improvements. This may depend on the type of memory mismatch, the nature of the memory contents, and the media involved, all of which are likely to influence the relative success of different types of coping. Another issue is related to the interpretive inflexibility of information systems (Käkölä and Koota, 1999), which may be related to the observation that ERP systems seem to become 'set in concrete' once they have been implemented (Bingi, Sharma, and Godla, 1999; Davenport, 2000). "This inflexibility results in part from agents with insufficient shared knowledge of: (1) the nature of social practices as a whole, (2) the articulation of these practices in time and space by the structural properties of organizations, (3) their own roles in the organization, and (4) the role of information systems as a structural property mediating work processes. [...] Dualistic information systems separate symbolic information from the material and social systems the symbols represent, hide the processing rules and retention structures in the software and database schemas, and blur the role of people as producers and consumers of information (Käkölä and Koota, 1999, p. 91)."

To overcome the interpretive inflexibility, individuals need to control their jobs again. "[…] Agents' ability to control their work depends on their domain and technology-related skills and knowledge as well as the shared stocks of knowledge that inform them about their own and other agents' roles and normatively regulated and sanctioned behaviors (Käkölä and Koota, 1999, p. 92)." The solution that Käkölä and Koota (Käkölä and Koota, 1999) sketch involves the development of a radically different structured information system. When looking at the further development of ERP systems, this approach may not be feasible since it may mean developing and implementing a completely different system than the ERP system in-use. However, it may be interesting to investigate the relation between the effectiveness of coping behavior and the presumed interpretive inflexibility of the ERP system.

Further research needs to answer questions regarding the occurrence and consequences of such organizational memory mismatches, as well as effective coping. Organizational memory mismatches can be solved in two ways (Van Stijn et al., 2000). One, by acquiring the ERP system's process knowledge and paradigms in the organization (changing the organization accordingly). Two, by changing the ERP system in such a way that it does incorporate the organization's process knowledge and paradigms, thus, customizing the package. This brings us to a question that is not only interesting from the organizational memory mismatch perspective, but also more in general. How to decide what aspects of the organization need to be adapted to the ERP package and what aspects of the ERP package need to be adapted to the organization, and under which conditions? This is an important question to be answered with respect to the business process modeling sub-process, because this decision is (implicitly) reflected in the business blueprint. However, AcceleratedSAP and ARIS do not address this question of how to decide this explicitly - it appears to be "just decided..."

Conclusion

In this paper we have provided a general discussion of issues relating to the representation of process knowledge during the implementation phase of ERP systems. We have suggested that ERP may well embed some, but by no means all, of the process knowledge that is resident in organizations. The concept of organizational memory provides an indication of the variety of memory media that are present in organizations and that potentially act as storage media for process knowledge.

It is our contention that considerable care should be exercised to identify exactly where different types of process knowledge reside in organizations. During implementation of ERP systems or other complex information systems decisions will have to be made as to what types of process knowledge can be represented in the ERP system and what types will continue to reside in other memory media. As we have noted, it is important to recognize that the process knowledge stored on different memory media has to interact in order that decisions can be taken and the knowledge refined and updated.

Given that many organizations are likely to have implemented ERP systems with only limited consideration of the above issues we have suggested that it will often be the case that what we have called memory mismatches are likely to arise during and after the implementation of ERP system. Further research is necessary to identify system behaviors that are likely to have arisen from such mismatches and the appropriate coping behaviors for the different types of mismatch.

We hope that this paper stimulates more research into the use of an organizational memory perspective in the development of ERP and other complex information systems that embed some types of organizational knowledge but by no means all of an organization's knowledge. In addition to ERP systems providing fertile ground for further research into organizational memory and knowledge management we believe that organizational memory and structuration theory provide a rich foundation for much of this research.

References

Anand, V., Manz, C.C. and Glick, W.H. An organizational memory approach to information management. Academy of Management Review (23: 4), 1998, pp. 796-809

ASAP AcceleratedSAP for 4.6B, software program (current: Dec. 1999)

Bingi, P., Sharma, M.K., and Godla, J.K. Critical issues affecting an ERP implementation. Information Systems Management (16:5), 1999, pp. 7-14

Bohn, R.E. Measuring and managing technological knowledge. Sloan Management Review (36:1), 1994, pp. 61-73

Bolwijn, P.T., and Kumpe, T. Manufacturing in the 1990s - productivity, flexibility and innovation, Long Range Planning (23:4), 1990, pp. 44-57

Boudreau, M.C. and Robey, D. Accounting for the contradictory organizational consequences of information technology: theoretical directions and methodological implications. Information Systems Research (10:2), 1999, pp. 167-185

Chan, R. Knowledge management for implementing ERP in SMES. SAPPHIRE'99, 1999, pp. 21-39

Davenport, T.H. Putting the enterprise into the enterprise system. Harvard Business Review, July - August, 1998, pp. 104 -112

Davenport, T.H. Mission critical: realizing the promise of enterprise systems. Boston (Ma.): Harvard Business School Press, 2000

Deloitte Consulting ERP's second wave. www.dc.com/services/secondwave/globalerp.pdf (current: Feb. 2000)

Duncan, R.B., and Weiss, A. Organizational learning: implications for organizational design. Research in Organizational Behavior (1), 1979, pp. 75 -123

Festinger, L. A theory of cognitive dissonance. Evanston, Ill. Row, Peterson, 1957

Fichman, R.G. and Kemerer, C.F. The illusory diffusion of innovation: an examination of assimilation gaps. Information Systems Research (10:3), 1999, pp. 255-275

Holland, C.P., Light, B. and Gibson, N. A critical success factors model for Enterprise Resource Planning Implementation. ECIS99, 1999, pp. 273-287

Käkölä, T.K. and Koota, K.I. Redesigning computer-supported work processes with dual information systems: the work process benchmarking service. Journal of Management Information Systems (16:1), 1999, pp. 87-119

Kirchmer, M. Business process oriented implementation of standard software: how to achieve competitive advantage efficiently and effectively. Berlin etc.: Springer, 1999

Kumar, K. and Van Hillegersberg, J. ERP experiences and evolution Communications of the ACM (43:4), 2000, pp. 23-26

Lassila, K.S. and Brancheau, J.C. Adoption and utilization of commercial software packages: exploring utilization equilibria, transitions, triggers, and tracks. Journal of MIS (16:2), 1999, pp. 63-90

Markus, M.L. and Tanis, C. The enterprise systems experience - from adoption to success In: Zmud, R.W., ed., Framing the domains of IT research: Glimpsing the future through the past. Cincinnati, OH: Pinnaflex Educational resources, Inc., 2000, pp. 173-207

Moorman, C. and Miner, A.S. The impact of organizational memory on new product performance and creativity. Journal of marketing research (34:1), 1997, pp. 91-106

Nonaka, I. A dynamic theory of organizational knowledge creation. Organization Science (5:1), 1994, pp. 14-37

Orlikowski, W. J. and Robey, D. Information technology and the structuring of organizations. CISR WP No. 220, 1991

Quinn, R.E. and Rohrbaugh, J. A spatial model of effectiveness criteria: towards a competing values approach to organizational analysis. Management Science (29:3), 1983, pp. 363-377

Robey, D., Wishart, N.A. and Rodriguez-Diaz, A.G. Merging the metaphors for organizational improvement: business process reengineering as a component of organizational learning. Accounting, Management & Information Technologies (5:1), 1995, pp. 23-40

Rosemann, M. and Wiese, J. Measuring the performance of ERP software - a balanced scorecard approach. Proceedings of the 10th Australasian Conference on Information Systems, 1999

SAP www.sap.com (current: Jan. 2000)

Scott, J. E. The FoxMeyer Drugs' bankruptcy: was it a failure of ERP? AMCIS99, 1999, pp. 223-225

Scheer, A.W. ARIS - Business process frameworks. Berlin etc.: Springer, 1998, 2nd edition.

Stein, E. Organizational memory: review of concepts and recommendations for management. International Journal of Information Management (15:1), 1995, pp. 17-32

Stein, E. and Zwass, V. Actualizing organizational memory with information systems. Information Systems Research (6:2), 1995, pp. 85-117

Van Stijn, E. Fixing memory misfits, Thesis University of Twente, 1999

Van Stijn, E. and Wijnhoven, F. Diagnosing organizational memory mismatches in the ERP usage stage. AMCIS2000, 2000

Volkoff, O. Using the structurational model of technology to analyze an ERP implementation. AMCIS99, 1999, pp. 235 - 237

Walsh, J.P. Managerial and organizational cognition: notes from a trip down memory lane. Organization Science (6:3), 1995, pp. 280-321

Walsh, J.P. and Ungson, G.R. Organizational memory. Academy of Management Review (16), 1991, pp. 57-91

Wijnhoven, F. Managing dynamic organizational memories: instruments for knowledge management. Pacific Grove etc.: Boxwood Press, 1999