Malhotra,
Y., Why Knowledge Management Systems Fail? Enablers and
Constraints of Knowledge Management in Human Enterprises . In Michael E.D.
Koenig & T. Kanti Srikantaiah (Eds.), Knowledge Management Lessons Learned:
What Works and What Doesn't, Information Today Inc. (American Society for
Information Science and Technology Monograph Series), 87-112, 2004.
Why Knowledge Management Systems Fail?
Enablers and Constraints of Knowledge Management in
Human Enterprises
Yogesh Malhotra, Ph.D.
Syracuse University, School of Management,
Syracuse, NY 13244, USA
yogesh(at)syr.edu, http://www.yogeshmalhotra.com/
Drawing upon lessons learned
from the biggest failure of knowledge management in recent world history and
the debacle of the 'new economy' enterprises, this chapter explains why
knowledge management systems (KMS) fail and how risk of such failures may be
minimized. The key thesis is that enablers
of KMS designed for the 'knowledge factory' engineering paradigm often unravel
and become constraints in
adapting and evolving such systems for business environments characterized by
high uncertainty and radical discontinuous change. Design of KMS should ensure that adaptation and innovation of
business performance outcomes occurs in alignment with changing dynamics of the
business environment. Simultaneously, conceiving multiple future trajectories
of the information technology and human inputs embedded in the KMS can diminish
the risk of rapid obsolescence of such systems. Envisioning business models not only in terms of knowledge
harvesting processes for seeking optimization and efficiencies, but in combination with ongoing knowledge
creation processes would ensure that organizations not only succeed in doing the thing right in the short term
but also in doing the right thing
in the long term. Embedding both
these aspects in enterprise business models as simultaneous and parallel sets
of knowledge processes instead of treating them in isolation would facilitate
ongoing innovation of business value propositions and customer value
propositions.
Keywords: Design of Successful Knowledge Management
Systems; Enablers and Constraints of Knowledge Management; Adaptive Systems for
Radical Discontinuous Change; Knowledge Harvesting and Knowledge Creation;
Information Processing and Sense Making; Strategic, Social, and, Behavioral
Aspects of Knowledge Management; Transformation of Knowledge Work and Knowledge
Organizations; Business Value Propositions and Customer Value Propositions;
Knowledge Management Failure; Business Model Innovation; New Business Models
1 Introduction
The advent of the era characterized by high uncertainty was announced by a recent Business Week (2001) cover story that determined September 11, 2001, as the day of the watershed event. On this day, the unprecedented combination of conventional means of terrorism inflicted their wrath on thousands of human lives in the World Trade Center twin towers despite policy-makers' preoccupation with unconventional means of terror. The basic premises guiding the knowledge processes of the intelligence machinery and policy-makers' decision models surmised that:
§ unconventional means pose greater risk compared with those posed by conventional means;
§ conventional means cannot reconfigure in unpredictable ways to pose greater risk than unconventional means;
§ impact of human and technology inputs can be determined with a safe margin of predictability;
§ hi-tech inputs always have greater impact than low-tech inputs;
§ human inputs play a lesser role compared with technology inputs and financial capital inputs in the input-outcome equation; and,
§ inputs rather than the execution strategy primarily determine the outcomes.
In retrospective it was found that all these assumptions were questionable. A review of the above assumptions guiding policy making decisions offers some interesting revelations listed below:
§ Pre-specified and pre-determined notions of unconventional and conventional, and, low-tech and high-tech inputs may not necessarily be always applicable;
§ Technology inputs and financial capital inputs may be less relevant factors in the input-outcome equation given unconventional strategy of execution that defines how creatively and innovatively inputs are deployed to produce unprecedented outcomes;
§ Human inputs may not necessarily play a lesser role than technology inputs or financial capital inputs in the input-outcome equation - given highly committed and motivated humans and their leaders, technology inputs and financial capital may assume a lesser role in the input-outcome equation.
Extending the same analysis to understand the recent debacle of 'new economy' enterprises also offers some interesting insights. Given the euphoria about the Internet technologies and the pitch of the venture capitalists and tech stock analysts and underwriters, Internet technology based businesses were summarily branded as unconventional in contrast to the conventional enterprises of the brick-and-mortar economy. It was assumed that conventional enterprises must get up on the Internet bandwagon if they had to survive in the future. It was assumed that given enough investment of venture capital, technology, and hype, any company could create and sustain successful business performance outcomes within a very short time. In summary, the following premises guided the euphoria about the Internet based companies which was compounded by the over-exuberance of media network reporters and the analysts:
§ unconventional means pose greater risk compared with those posed by conventional means;
§ conventional means cannot reconfigure in unpredictable ways to pose greater risk than unconventional means;
§ impact of human and technology inputs can be determined with a safe margin of predictability;
§ hi-tech inputs always have greater impact than low-tech inputs;
§ human inputs play a lesser role compared with technology inputs and financial capital inputs in the input-outcome equation; and,
§ inputs rather than the execution strategy primarily determine the outcomes.
Given the recent spate of Internet-based company failures, reversal of faith in the Net-based companies has been pervasive. This has happened despite the fact that widespread weaknesses are being observed in many sectors of the economy including many industries and companies that represent the tried-and-tested 'old economy'. It is time to reflect upon the lessons learned from the biggest failures of knowledge management in recent world history and the debacle of the 'new economy' enterprises. This is important to inform the prevailing myth about the intrinsic infallibility of 'old economy' enterprises in contrast with the 'new economy' enterprises despite dependence of both on the same fundamentals.
Based upon the earlier analysis, one can offer the following hypotheses that seem to offer a more robust basis for defining, implementing, and executing effective knowledge management systems.
§ The impact of human and technology inputs cannot be determined with safe margin of predictability as the business performance outcomes are separated from these inputs by intervening variables. Such variables include effective acceptance and utilization of technologies by humans; motivation and commitment for adoption of these technologies and for achieving the specified performance outcomes; and, contextual interpretation of information resulting in diverse subjective decisions and actions. Pre-specified outcomes may also become marginalized with the changing business environment when the inputs are consumed for doing the thing right even though it may not be the right thing any more.
§ Lo-tech and hi-tech inputs are constrained or enabled by knowledge workers who utilize these inputs as well as by the strategy of execution that may together produce different outcomes despite similar mix of the inputs. The contrast between lo-tech and hi-tech is based upon context-specific perspectives and as business contexts change, these contrasts may change or become immaterial with emergence of newer and unprecedented inputs.
§ The contrast between unconventional and conventional means of producing business performance outcomes is based upon context-specific perspectives. As business contexts change such contrast may become marginalized with emergence of newer and unprecedented means as well as unprecedented outcomes. Such contrast may also become marginalized if conventional means are configured in unprecedented ways to achieve unprecedented outcomes. In this discussion, it is observed that unprecedented business performance outcomes are realized as a result of new business value propositions and customer value propositions.
This article explains how both old and new economy enterprises having any mix of brick-and-click strategies are vulnerable to the above failures. Such failures result from the gaps between the input resources and the business performance outcomes, and, the gaps between the value these enterprises create and the value demanded by changing market conditions, consumer preferences, competitive offerings, and, changing business models, and, industry structures. KMS are often defined in terms of inputs such as data, information technology, best practices, etc., which by themselves may inadequately explain business performance outcomes. Often, moderating and intervening variables may play a significant role in skewing the simplistic relationships based upon correlation of the above inputs with business performance outcomes. Also, usefulness of such inputs and how they are strategically deployed are important issues often left unquestioned as 'expected' performance outcomes are achieved, but the value of such performance outcomes gets eroded by the dynamic shifts in the business and competitive environments. The remaining discussion will explain why KMS fail; how enablers of KMS designed for the 'knowledge factory' engineering paradigm become constraints in adapting and evolving such systems for business environments characterized by high uncertainty and radical discontinuous change; and, how risk of such failures may be minimized.
2 Knowledge Management for Routine and Structured Information
Processing
(Model 1 KMS)
Given the centrality of computerized information processing in most mainstream conceptualizations of knowledge management, most KMS primarily depend upon routines that are programmed in the logic of computational machinery and on data residing in data warehouses. [A detailed discussion about such definitions of knowledge management is available elsewhere (Malhotra 2000a, 2000b). A recent historical perspective of knowledge management is available in Prusak (2001).] Based on the pre-specification and pre-determination of the programmed logic connecting 'information inputs' and consequent 'information outcomes', such systems are based upon consensus, convergence, and, compliance to ensure adherence to organizational routines. The mechanistic model of information processing and control based upon compliance is not only limited to the computational machinery, but extends to specification of goals and tasks and the best practices and institutionalized procedures to achieve those pre-specified outcomes. Motivated by emphasis on optimization and efficiencies of scale, the above logic of knowledge management has evolved from 'scientific' Taylorism and the assembly line techniques applied by Henry Ford in the production of Model T.
Not surprisingly, the original versions of such KMS were reified in the interpretations of some Information Systems researchers who seemed to believe that technology inputs, rather than knowledge workers, would play a predominant role in the performance outcome equation discussed earlier. One example of such systems was offered in a popular Information Systems textbook published by professors at the Harvard Business School (Applegate et al., 1988, p. 44):
"Information systems will maintain the corporate history, experience and expertise that long-term employees now hold. The information systems themselves -- not the people -- can become the stable structure of the organization. People will be free to come and go, but the value of their experience will be incorporated in the systems that help them and their successors run the business."
Not surprisingly, many business and technology executives trained in similar reasoning have been trying to push for adoption of computer technologies for storing their employees' knowledge in computerized databases and programmed logic of the computing machinery with mixed results. Best practices, benchmarks, and rules tend to define the assumptions that are embedded not only in information databases, but also in the organization's strategy, reward systems, and resource allocation systems.
A recent interpretation of the same reasoning, illustrated in Figure 1, based upon pre-definition, pre-specification, and, pre-determination is offered in a definition popularized by the Gartner Group (cf: Oracle Magazine, 1998):
"Knowledge Management promotes an integrated approach to identifying, capturing, retrieving, sharing, and evaluating an enterprises information assets. These information assets may include databases, documents, policies, procedures, as well as the un-captured tacit expertise and experience stored in individual's heads."
Such inputs-oriented mechanistic and static representations of knowledge do not provide any hint as to how these inputs would affect business performance. Nor do they suggest how to deal with "associated emotions and specific contexts" (Nonaka & Takeuchi 1995, p. 63) that characterize tacit knowledge.
Recent thrust of some organizational knowledge management initiatives on archiving 'best practices' and 'what we know' to guide future decisions and actions is also based on a relatively predictable view of the business environment. Not surprisingly, this model of knowledge management guided by pre-specification and pre-determination of business logic with primary emphasis on optimizing the user of existing knowledge [reified in best practices, computational logic, data warehouses, etc.] has primarily focused on knowledge re-use over creation of new knowledge. This model is based upon managerial focus on seeking consensus and compliance to minimize variance so that pre-specified business performance outcomes are achieved. In this model of knowledge management, conformance to pre-specified and pre-determined business logic is expected to ensure pre-specified and pre-determined business performance outcomes are achieved.
Not surprisingly, many knowledge management practitioners and researchers who identify with Model 1 discussed above consider information and knowledge as synonymous constructs. In this perspective, both these constructs can be expressed in the computational rule based logic as well as in the form of data inputs and data outputs that trigger pre-defined and pre-determined actions in pre-programmed modes. There is limited, if any, scope for diverse interpretations of the same information by different agents, nor is there any need for multiplicity of meanings based upon the same information. Homogeneity of information-processing and control logic is desirable as it ensures that the model works as pre-specified. Inclusion of feedback and feedforward loops may often provide the mechanism for reactive fine-tuning of inputs for optimal conversion of the inputs into outcomes. [Information Systems researchers (cf: Churchman 1971; Mason and Mitroff, 1973; Malhotra, 1997) have discussed limitations of this model, particularly for environments characterized by uncertainty and radical change.].
Not surprisingly, the goal of KMS based on Model 1 is often characterized as "getting the right information to the right person at the right time." This model is based on the assumption that all relevant knowledge, including tacit knowledge can be stored in computerized databases, software programs, and, institutionalized rules and practices. The distinguishing features of this model are derived from the following assumptions:
(a) the same knowledge can be re-used by any human mind or computer to re-process the same logic to produce the same outcomes;
(b) the same outcomes will be needed and delivered again and again through optimal use of input resources;
(c) the system's primary objective is to achieve the most efficient means for transforming pre-specified inputs into pre-determined outcomes; and,
(d) there is no need for subjective interpretation of information - criticism and conflict must be minimized to achieve conformance and compliance.
Model 1 KMS are based on doing the thing right where the pre-specified inputs, processing logic, and, the outcomes are assumed to represent the right thing. The overriding belief is that designers of the systems and the knowledge managers have accurate and complete knowledge about the viability of the input-output transformation process as well as the viability of the performance outcomes that have been pre-defined.
The next section explains the contrasting Model 2 of KMS that is more suited for non-routine and unstructured sense making when deterministic controls encounter uncertain environments characterized by "wide range of potential surprise" that defy predictive logic (Landau and Stout, 1979). Interestingly, many of the limitations of Model 1 may be considered as strengths in Model 2 as the premises of pre-determination, pre-definition, and, pre-specification of meanings, actions, and, outcomes become less relevant.
3 Knowledge Management for Non-Routine and Unstructured Sense
Making
(Model 2 KMS)
In Model 2, the construct of knowledge may be better represented as intelligence in action as it is a composite construct resulting from interaction of data, information, rules, procedures, best practices and traits such as attention, motivation, commitment, creativity, and, innovation. This contrasting representation of knowledge as intelligence in action rather than static computerized representations of Model 1 is notable because of several reasons. The active, affective, and dynamic representation of knowledge makes better sense from a pragmatic perspective and is better aligned with theoretical representations of this construct beyond the domain of information technology management. It is active as knowledge is best understood in action - it is not the theory but the practice of theory that makes the difference. It is affective as it takes into consideration not only the cognitive and rational dimensions but also emotional dimensions of human decision-making. It is dynamic as it is based upon ongoing reinterpretation of data, information, and, assumptions while pro-actively sensing how decision-making process should adjust to future possibilities. From a pragmatic perspective, the dynamic representation of knowledge provides a more realistic construct where human and social interactions are present while situating this construct more proximal to performance outcomes as illustrated in Figure 2. [A detailed discussion about the active, affective, and dynamic representation of knowledge is available elsewhere (Malhotra 1999, Malhotra and Kirsch (1996), Malhotra (in press)).]
Model 2 provides a better representation of reality as it takes into consideration two key characteristics:
(a) what is done with data, information, and best practices depends upon subjective interpretation ("construction") of individuals and groups that transform these inputs into actions and performance; and,
(b) performance outcomes need to be continuously re-assessed to ensure that they indeed represent best business performance for the enterprise with respect to changing market conditions, consumer preferences, competitive offerings, and, changing business models, and, industry structures.
This view of knowledge management is consistent with some other
perspectives that have attempted to address the limitations of Model 1 that is
based upon "overdefinition of rules and overspecification of tasks"
(Landau and Stout, 1979). For instance, Churchman (1971) has emphasized that: "To conceive of
knowledge as a collection of information seems to rob the concept of all of its
life... Knowledge resides in the user and not in the collection."
Similarly, Nonaka
and Takeuchi (1995) had proposed the conceptualization of knowledge as justified belief in their argument
that, “knowledge, unlike information,
is about beliefs and commitment.” On a complementary note,
Davenport and Prusak (1998, p. 5) have defined knowledge as deriving from minds
at work: "Knowledge is a fluid mix of framed experience, values,
contextual information, and expert insight that provides a framework for
evaluating and incorporating new experiences and information. It originates in
the minds of knowers. In organizations, it often becomes embedded not only in
documents or repositories but also in organizational routines, processes,
practices, and norms."
As represented in Model 2, knowledge is a dynamic construct in contrast to static representations of Model 1 because diverse [individual and shared] meanings are possible based upon diverse interpretations of the same information inputs across different contexts and at different times. Processing of knowledge through the machinery of information technologies may still be represented by simplified, highly routine, and, structured forms that permit pre-definition, pre-programming, and, pre-determination of data inputs for achieving pre-specified performance outcomes. In contrast, human sense making processes represent a complete contrast, and, human decision-making is influenced by attention, motivation, commitment, creativity, and innovation of individuals and groups. [Related discussion on how "construction of meaning" differs from "the processing of information" is available in Bruner (1990), Kelly (1963), Malhotra (1999), Morris (1938), Strombach (1986). Detailed conceptualization and empirical validation of commitment and motivation constructs as applicable to effective technology use in knowledge work is available in Malhotra (1998), and, Malhotra and Galletta (1999). Contrast between information-processing capabilities of smart technologies and sense-making capabilities of humans is explained in Malhotra (2001a).]
4 Continuum of KM Systems Between Model 1 and Model 2
Model 1 works well in predictable and stable environments with primary focus on knowledge harvesting, and, knowledge re-use and replication. Under moderate levels of control, this model could be used for knowledge workers' goal and task specification to achieve pre-specified performance outcomes. This model may be susceptible to failure when creativity and innovation of knowledge workers overwhelms the controls inherent in the pre-specified logic of the input-output transformation process. It is also vulnerable to failure where attention and actions of knowledge workers are significantly influenced by their intrinsic motivation [rather than organizational or institutional rewards and punishments] and commitment to personal goals [rather than organizational or institutional goals]. The ideal scenario is to achieve perfect congruence between extrinsic motivation and intrinsic motivation, and, between organizational goals and individual goals, but this is a formidable challenge for designers of most organizational KMS. [A detailed discussion about the contrast between extrinsic motivation and intrinsic motivation, and, contrast between various levels of commitment is available elsewhere (Malhotra 1998).]
While Model 1 and Model 2 represent the extreme archetypes of KMS, most organizations need some combination of both depending upon their emphasis on knowledge harvesting and knowledge creation. Also, organizations and inter-enterprise value networks contain some business processes that primarily depend upon knowledge harvesting and others that primarily depend upon knowledge creation. This point can be appreciated by considering the two world's of business that often co-exist in many organizations - the world of bulk-processing industrial economy and the "world-of-reeverything" of the knowledge economy (Arthur 1996):
"The two worlds are not neatly split. Hewlett-Packard, for example, designs knowledge-based devices in Palo Alto, California, and manufactures them in bulk in places like Corvallis, Oregon or Greeley, Colorado. Most high-tech companies have both knowledge-based operations and bulk-processing operations. But because the rules of the game are different for each, companies often separate them—as Hewlett-Packard does. Conversely, manufacturing companies have operations such as logistics, branding, marketing, and distribution that belong largely to the knowledge world. And some products—like the IBM PC—start in the increasing returns world, but later in their life cycle become virtual commodities that belong to Marshall’s processing world."
Model 1 is relevant to the industrial world of bulk-economy production and Model 2 is relevant to the "world-of-re-everything". Optimization-based routinization of organizational goals and convergence is relevant for ‘freezing’ the meaning for achieving optimization-based efficiencies. However, ‘unfreezing’ of meaning embedded in information is critical for reassessing and renewing the routines embedded in business logic and business processes. Business enterprises will need to be facile in both modes despite the apparent contradiction in terms of the business logic and related assumptions. For instance, a key challenge for most organizations with institutionalized 'best practices' is to ensure that such practices remain open to critique, adaptation, and, replacement so that the enterprise is not caught in the death spiral (Nadler and Shaw 1995) of doing more of the same better and better with diminishing marginal returns (Drucker, 1994). Discontinuously changing environments impose upon the organization need for "creative synthesis" resulting from a "dialectical confrontation of opposing interpretations" (Mason and Mitroff 1973, p. 482). Although companies often separate the operations pertaining to the two worlds of business related to Model 1 and Model 2, both worlds need to be integrated in their business models. For example, given the diminishing margins in the PC markets due to increased competition, computer distributor Dell may need to shift its focus to distribution of servers or to hosting services. To do so effectively, however, it would need to start harvesting [using Model 1] knowledge that it created [using Model 2] earlier through experimentation, adaptation, and innovation related to servers or hosting and it is time to redefine the customer value propositions and the related business value propositions. [How the contrast between the two worlds applies to the use of information technology and knowledge management strategy is discussed in the context of business model innovation in Malhotra (2000d). How the world of 'old economy' and 'new economy' connect with their representation in terms of digitized information and knowledge has been discussed in business literature on virtual organizations, virtual products, and, virtual services, see for instance, Davidow and Malone (1993). More recent research relating knowledge management to virtual organizations and business model innovation is available in Malhotra (2000c) and Malhotra (2001b) respectively.]
As most business environments would include a combination of both stabilizing factors and destabilizing factors, real world KMS implementations should contain combinations of characteristics of both models. The processes of knowledge re-use and knowledge creation need to be balanced by integration of routine and structured information processing and non-routine and unstructured sense making in the same business model. Figure 3 depicts this representation of business model that includes simultaneous and parallel sets of knowledge harvesting and knowledge creation processes.
Prior arguments suggest the need for
skepticism about the myth of intrinsic infallibility of 'old economy'
enterprises in contrast with the 'new economy' enterprises despite dependence
of both on the same fundamentals. Both old and new economy enterprises having
any mix of conventional and unconventional brick-and-click strategies
are vulnerable to the failures resulting from the gaps between their inputs and
the business performance outcomes as well as from gaps between the value they
create and the value demanded by their customers. Also, despite similarity of
inputs and sought market shares, relative success of any business model will be
determined more by its execution than by the inputs especially in case of
rapidly changing environments. As illustrated in Figure 4, new business models
for the knowledge economy need to consider Internet and Web simply as elements
of the overall business strategy without getting caught in irrational exuberance or despair about
these means of producing business value. Just like other inputs and intervening
variables, inadequate understanding or application of Internet and Web should
not become the basis for outright rejection or degradation of these
technologies.
