Academy of Management Review2008, Vol. 33, No. 4, 825–845. MANAGEMENT INNOVATION JULIAN BIRKINSHAW GARY HAMEL London Business School MICHAEL J. MOL University of Reading We define management innovation as the invention and implementation of a man- agement practice, process, structure, or technique that is new to the state of the art and is intended to further organizational goals. Adopting an intraorganizational evolutionary perspective, we examine the roles of key change agents inside and outside the organization in driving and shaping four processes—motivation, inven- tion, implementation, and theorization and labeling—that collectively define a model of how management innovation comes about. Over the past half-century, scholars around research lab and GM’s invention of the M-formthe world have produced a vast body of aca- organization structure), the amount of detaileddemic research and writing on innovation. knowledge about how management innovationWhile most of this research has focused on var- is actually implemented is limited.ious aspects of technological innovation (e.g., In its broadest sense, management innova-Henderson & Clark, 1990; Utterback, 1994), the tion has, of course, received considerable re-trend over the last fifteen years has been toward search attention over the years. As we discussexploring other forms of innovation, such as pro- in the following section, there are four keycess innovation (e.g., Pisano, 1996), service inno- perspectives in the literature: (1) an institu-vation (e.g., Gallouj & Weinstein, 1997), and stra- tional perspective that focuses on the socio-tegic innovation (Hamel, 1998; Markides, 1997), economic conditions in which new manage-with a view to understanding how they are man- ment ideas and practices take shape (e.g.,aged and how they contribute to long-term firm Guillen, 1994); (2) a fashion perspective that ´success. focuses on the dynamic interplay between us- The focus in this article is on a relatively under- ers and providers of management ideas (e.g.,researched form of innovation—management Abrahamson, 1996); (3) a cultural perspectiveinnovation—and particularly the processes that focuses on how an organization reacts tothrough which it occurs. We apply a relatively the introduction of a new management prac-narrow definition of management innovation— tice (e.g., Zbaracki, 1998); and (4) a rationalspecifically, the invention and implementation perspective that focuses on how managementof a management practice, process, structure, or innovations—and the individuals who drivetechnique that is new to the state of the art and them— deliver improvements in organization-is intended to further organizational goals. al effectiveness (e.g., Chandler, 1962). There isWhile many of the landmarks of management also a related body of literature concernedinnovation are familiar to every business with the subsequent diffusion of managementscholar (e.g., GE’s development of the modern innovations across industries or countries (e.g., Guler, Guillen, & MacPherson, 2002). But ´ We thank Jos Benders, Rick Delbridge, Hakan Ener, Mar- useful as these bodies of literature are, theytine Haas, Michael Jacobides, Robert Kaplan, Olav Soren- have surprisingly little to say about the gen-son, Yiorgos Mylonadis, and seminar participants at London erative mechanisms by which new manage-Business School, INSEAD, HEC (Paris), Imperial College, and ment ideas are first created and put into prac-King’s College. Earlier versions of this research were pre-sented at the 2005 European Group for Organizational Stud- tice. To state the point slightly differently, ouries conference and the 2006 annual meeting of the Academy understanding of the processes of manage-of Management. ment innovation is currently very limited and 825Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyrightholder’s express written permission. Users may print, download, or email articles for individual use only.
826 Academy of Management Review Octoberis based largely on a few well-known exam- on insights from all four perspectives but relatesples, such as Chandler’s (1962) documentation most closely to the rational perspective.of the emergence of the M-form structure. Whatis required—and what we provide a first step Four Perspectives on Management Innovationtoward in this article—is a systematic andgrounded process theory of how management Proponents of the institutional perspectiveinnovation transpires. take a macrolevel and comparative approach to We focus on the specific actions individuals make sense of the institutional and socioeco-inside or outside the firm might undertake that nomic conditions in which particular manage-lead to the emergence of a management inno- ment innovations emerge. For example, Guillen ´vation—what we might call “management in- (1994) examined the impact of seven sets of in-novating,” as a way of capturing the poten- stitutional factors on the introduction of newtially critical role of human agency in the managerial ideologies and techniques acrossprocess. We address two specific questions. four countries; Cole (1985) focused on how theFirst, what is management innovation? How balance between labor market incentives thatcan we define management innovation in a are mostly set by the state, the relative strengthuseful and rigorous way that emphasizes its of industry associations, and the predispositiondistinctiveness? Second, and building on the of organized labor influenced the introduction offirst question, what are the processes through small-group activities in different countries; andwhich management innovation comes about? Kossek (1987) examined industry- and firm-levelWhat does the literature tell us about the typ- influences on the emergence of human resourceical sequence of actions followed by individ- management innovations. Normative beliefsuals inside and outside the organization that about what is progressive may drive manage-result in the creation of management innova- ment innovation, but those beliefs are also sub-tion? And to what extent can we induce a ject to long Kondratieff waves of economicgeneral set of arguments about the causal change in which new technologies occur andmechanisms through which management in- create performance gaps that then necessitatenovation takes place? The article concludes management innovation (Abrahamson, 1997;with some thoughts about the future research Barley & Kunda, 1992). The institutional perspec-agenda that might be pursued to further ad- tive measures innovation in terms of the dis-vance our understanding of management in- course around particular ideologies and also atnovation. the level of specific practices or techniques. It gives no direct consideration to the role of hu- man agency in shaping the process; instead, it WHAT IS MANAGEMENT INNOVATION? focuses on the preconditions in which an inno- Management innovation involves the intro- vation first emerges and then the factors thatduction of novelty in an established organiza- enable industries to adopt such innovations.tion, and as such it represents a particular form The fashion perspective focuses on how man-of organizational change. In its broadest sense, agement innovations emerge through the dy-then, management innovation can be defined as namic interplay between the managers who usea difference in the form, quality, or state over new management ideas and the “fashion set-time of the management activities in an organi- ters” who put forward those ideas (Abrahamson,zation, where the change is a novel or unprece- 1991, 1996). This perspective provides a wealth ofdented departure from the past (Hargrave & Van insight into how management fashions takede Ven, 2006; Van de Ven & Poole, 1995: 512). On shape, including a detailed understanding ofthe basis of this high-level definition, we iden- the typical attributes of managers who buy intotified four distinct perspectives on management these fashions (Gill & Whittle, 1993; Huczynski,innovation in the literature, as summarized in 1993; Jackson, 1986), as well as the ways in whichTable 1. These four should be seen as the dom- fashion setters shape incipient demand for theirinant perspectives around which research has ideas (Benders & van Veen, 2001; Clark, 2004;clustered in the past, rather than as theoreti- Kieser, 1997; Mazza & Alvarez, 2000). However, itcally comprehensive in terms of the domain that has little to say about the true origins of man-they cover. Our approach draws to some degree agement fashions, or why certain innovations
2008 Birkinshaw, Hamel, and Mol 827 TABLE 1 Key Features of Four Perspectives on Management Innovation InstitutionalFeatures Perspective Fashion Perspective Cultural Perspective Rational PerspectiveRepresentative Barley & Kunda Abrahamson (1991, Gill & Whittle (1992), Alange, Jacobsson, & ¨ papers (1992), Bendix 1996), Abrahamson Knights & McCabe Jarnehammar (1998), (1956), Cole (1985), & Fairchild (1999), (2000), Knights & Chandler (1962), Guillen (1994), ´ Clark (2004), Murray (1994), McCabe Damanpour (1987), Kossek (1987), Huczynski (1993), (2002), Stjernberg & Kaplan (1998), Strang & Kim Kieser (1997), Mazza Philips (1993), Zbaracki Kimberley & (2005), Weitz & & Alvarez (2000), (1998) Evanisko (1981), Shenhav (2000) Staw & Epstein Tichy & Sandstrom (2000) (1974), Yorks & Whitsett (1985)Core question What institutional How do aspects of the How do management What is the role of conditions give supply of and innovations shape, managers in rise to the demand for new and get shaped by, inventing and emergence and management ideas cultural conditions implementing new diffusion of affect their inside an management management propagation? organization? practices? innovations?Key factors Institutional Suppliers of new Culture of the Actions of key influencing conditions and ideas and the organization in which individuals driving the attitudes of major legitimacy of their the innovation is the process inside innovation groups of proposals introduced or outside the process influencers organizationRole of human Rarely discussed Rarely discussed Agents are important but Agents initiate and agency in constrained by power drive the process driving the relations and within an organiza- process traditions tional contextLevel of Firm plus industry/ Firm plus market for Firm plus individual Individual plus firm analysis country new ideasProcess of Progressive changes Cyclical process of Socially constructed Progressive changes change and in management hype then change process; in management outcome of ideology and/or disillusionment; no usually very little practice toward innovation practice, evidence that change in way of more effective ways sometimes toward innovation leads to working and of working; success more effective long-term benefits perpetuation of not guaranteed ways of working existing power relationsbecome fashions while others do not. The fash- level introduction of the innovation. One strandion perspective spans the macro and micro lev- of this literature takes a critical perspectiveels of analysis, with a concern both for the in- (Knights & McCabe, 2000; McCabe, 2002) whiledustry that supplies new management ideas the other adopts an intraorganizational processand for the behavioral reasons why individual perspective (Stjernberg & Philips, 1993; Zbaracki,managers choose to buy into those ideas. Man- 1998), but both share some common themes: aagement fashions can exist as abstract ideas or recognition that established organizations dorhetorics, or as specific practices or techniques. not change easily, that management innovation Proponents of the cultural perspective attempt has both rhetorical and technical components,to understand how management innovation and that the outcome of the introduction of ashapes, and gets shaped by, the culture of the management innovation is rarely what was in-organization in which it is being implemented. It tended by the senior executives who introducedoperates at the meso level of analysis by look- it. Unlike the two previous perspectives, the cul-ing at how individual attitudes toward manage- tural perspective provides some insight intoment innovation interact with the organization- how management innovations are imple-
828 Academy of Management Review Octobermented, though primarily from the point of view ideas are scientific management, total qualityof those who are being asked to participate in management (TQM), and the learning organi-the process, rather than those who are driving it. zation. While not identical, this concept of aThe outcome of management innovation accord- management idea is comparable to Guillen’s ´ing to this perspective is typically a reinforce- (1994) notion of an organizational ideology,ment of the status quo (McCabe, 2002). This per- along with Barley and Kunda’s (1992), Abraham-spective does not deny that changes can occur son’s (1996), and Suddaby and Greenwood’sas a result of management innovation, but the (2005) notions of management rhetoric.1 At theforces at work in large organizations typically more operational level we can identify man-dampen its impact. agement practices, management processes, The rational perspective builds on the management techniques, and organizationalpremise that management innovations are in- structures2 (Alange et al., 1998; Guillen, 1994) ¨ ´troduced by individuals with the goal of making as different facets of the rules and routines bytheir organizations work more effectively. Ac- which work gets done inside organizations (forcording to this perspective, an individual puts the sake of readability, we use the term man-forward an innovative solution to address a spe- agement practices throughout the remaindercific problem that the organization is facing, and of the article to cover this full range of activi-he or she then champions its implementation ties). In definitional terms this article focusesand adoption (Burgelman, 1983; Howell & Hig- on management innovation at the operationalgins, 1990). Some studies from this perspective level—that is, in terms of the generation andhave favored a case study methodology (e.g., implementation of new practices, processes,Chandler, 1962; Tichy & Sandstrom, 1974), structures, or techniques— because this is thewhereas others have used large-sample quanti- level at which observable changes take placetative approaches (Damanpour, 1987; Kimberly in the way work is done and the management& Evanisko, 1981), but all studies span the micro- innovation process can be witnessed. But, asmacro levels of analysis by focusing on the ac- will become clear, there is an important inter-tions of key individuals within an organizational action between the development of new man-and environmental context. There is also a sub- agement practices and new managementtheme within this perspective concerned with ideas, so our theoretical arguments will givethe links between management and technolog- due consideration to both levels of analysis.ical innovation, which suggests that they may Second, how new does an innovation have tocoevolve (Damanpour & Evan, 1984; Ettlie, 1988; be? There are two equally valid points of viewGeorgantzas & Shapiro, 1993). in the literature. Abrahamson (1996) and Kim- berly (1981) define an innovation as “new to the state of the art,” which essentially meansAn Operational Definition of ManagementInnovation 1 This review of the literature highlights the Guillen states that organizational ideologies “can serve ´ as cognitive tools that managers use to sort out the complex-very different approaches researchers have ities of reality, frame the relevant issues, and choose amongused to make sense of the phenomenon of man- alternative paths of action” (1994: 4). Abrahamson definesagement innovation, and it helps us to focus on rhetoric as “spoken or written discourse that justifies the usethree key questions that arise as we seek to of a set of techniques for managing organizations or theirdevelop an operational definition. employees” (1996: 259). 2 First, what exactly is being innovated? By bracketing these elements together, we end up with a broad phenomenon. But the distinctions among practice,There is little consistency in the terminology process, structure, and techniques are not clean, either con-within or across the four perspectives, but we ceptually or empirically, so it would be difficult to definebelieve it is useful to separate out two levels management innovation in a way that excluded one or otherof analysis. At the more abstract level are of them. Moreover, our analysis suggests that there are im-management ideas, defined by Kramer as portant similarities across the different forms of manage- ment innovation, especially with regard to how they are“fairly stable bodies of knowledge about what generated. We therefore find it useful to bracket togethermanagers ought to do. . . . a system of assump- management innovations across these areas, “as if theytions, accepted principles and rules of proce- constitute a homogeneous entity” (Alange, Jacobsson, & ¨dure” (1975: 47). Examples of management Jarnehammar, 1998: 7).
2008 Birkinshaw, Hamel, and Mol 829without known precedent. But many other re- that ultimately change the organization’s goalssearchers implicitly see innovation as “new to (Selznick, 1957).the organization” so that, for example, the ini- In sum, we define management innovation3 astial introduction of a TQM program to an or- the generation and implementation of a man-ganization might be categorized as a manage- agement practice, process, structure, or tech-ment innovation (e.g., McCabe, 2002; Zbaracki, nique that is new to the state of the art and is1998). Our interest in this article is in new to intended to further organizational goals (see Ta-the state of the art innovations, for the primary ble 2 for a list of examples). And while it is not areason that this is the area where existing necessary component of our definition, it isknowledge is the most limited. But the bound- worth reinforcing that our perspective on man-ary between the two definitions is blurred: if agement innovation gives conscious attention toone considers a spectrum of approaches to the the individuals who drive the process. Indeed,implementation of management practices, on one of the themes of this article is the need tothe left side an organization might buy an “off increase the emphasis on human agency inthe shelf” practice from a consultancy, and management innovation while not losing sighton the right side it might come up with a of the contextual dynamics that are the focus ofcompletely novel innovation of its own. Our the institutional and fashion perspectives. Asinterest is on those innovations toward the McCabe puts it, “What is required is an under-right side of the spectrum, where the level of standing of innovation as part of a far moreadaptation to the specific context of the inno- complex social process: interrelated to the wayvating organization is high and where there is in which individuals interpret, act, and ascribe meaning to the world” (2002: 509).a considerable level of uncertainty regardingits outcome. Third, what is the purpose of management Management Innovation versus Relatedinnovation? Proponents of both the fashion and Constructscultural perspectives see management innova- Having established an operational definitiontion as having little lasting impact on the orga- of management innovation, we still need tonization, whereas those of the institutional and make a prima facie case that a theoretical dis-rational perspectives view management inno- cussion of the process of management innova-vation as generating positive outcomes for the tion—in its own right—is necessary.innovating firm and/or for society as a whole. As We propose three key factors that make man-suggested earlier, our focus is aligned most di- agement innovation distinctive. First, there arerectly with the rational perspective in that we important differences in the nature of the outputsview management innovation as intending to of management innovation and technological in-further the organization’s goals, which may in- novation that affect how the respective processesclude both traditional aspects of performance unfold. Management innovations are typically(e.g., financial goals) and softer aspects (e.g., tacit in nature and “difficult if not impossible toemployee satisfaction). This is appropriate be- protect by patent” (Teece, 1980: 464); they are alsocause it helps to explain why firms are prepared relatively “difficult to observe, define and to iden-to engage in the costly and somewhat risky pro-cess of management innovation in the firstplace. This approach serves to underline the 3 It should be noted that there are several related terms inimportant point that not all management inno- use: managerial innovation (Kimberly & Evanisko, 1981), ad-vations are ultimately successful. For example, ministrative innovation (Damanpour & Evan, 1984), organi- zational innovation (Alange et al., 1998; Damanpour & Evan, ¨Volvo experimented for many years with cellu- 1984; Kimberly & Evanisko, 1981), and management innova-lar manufacturing, with the intention of deliver- tion (Abrahamson, 1991; Kossek, 1987; Stata, 1989). Organiza-ing significant benefits, but the innovation was tional innovation is sometimes used to imply any type ofultimately discontinued (Berggren, 1992). More- innovation generated by organizations, including new prod-over, it should also be noted that goals are ucts. Administrative innovation typically refers to a narrow range of innovations around organizational structure andrarely entirely exogenous to the organization; human resource policies and does not include innovationsindeed, the process of innovating can result in in, for instance, marketing or operations management. Wethe introduction of new practices or programs therefore prefer the term management innovation.
830 Academy of Management Review October TABLE 2 Examples of Management InnovationExample How It Fits the Definition of Management InnovationModern research lab (e.g., Hargadon, 2003) A new structure to manage the technological innovation process; intended to improve technological and product innovationsDivisional (M-)form (e.g., Chandler, 1962) A new organizational structure for dealing with complex, multiple-product, and multiple-market firmsToyota production system (e.g., Ohno, 1988) A new set of practices and processes aimed at improving production efficiency and reducing wasteTotal quality management (e.g., Zbaracki, 1998) A new set of practices and processes aimed at reducing quality defects and improving customer satisfactionDiscounted cash flow (e.g., Pezet, 1997) A new technique intended to improve investment and budgeting decisions by adding a temporal dimensionSpaghetti organization (e.g., Foss, 2003) A new organizational structure with the objective of increasing employee initiatives and overcoming problems of hierarchyCellular manufacturing (e.g., Berggren, 1992) A new process for managing tasks inside a production unit aimed at improving employee satisfaction and production outputNASA new organization (e.g., Carroll, Gormley, Bilardo, A new structure and practice for teams to perform complex Burton, & Woodman, 2006) modeling and analysis without colocationActivity-based costing (e.g., Kaplan, 1998) A new practice and technique for assigning costs aimed at providing more realistic cost assessmentsModern assembly line (e.g., Hounshell, 1984) A new set of practices and processes with the goal of improving production efficiency and lowering costsBalanced scorecard (e.g., Kaplan, 1998) A new technique and practice for integrating various types of information with the aim of making more informed decisionsQuality of work life (e.g., Yorks & Whitsett, 1985) A new set of practices and processes around the job design of employees with the goal of improving their happiness at worktify system borders for” (Alange et al., 1998: 8). ¨ and uncertainty arises because of a fear that theTaken together, these attributes allow a higher innovation will have negative consequences forlevel of subjective interpretation on the part of the the individual and/or the organization. If an or-potential user than is common with technological ganizational change is proposed that has al-innovations, which, in turn, increases the impor- ready been successfully implemented else-tance of the social and political processes fol- where (e.g., the installation of a new IT system),lowed by the proponents of the innovation. its proponents can allay the concerns of individ- Second, very few organizations have well- uals by referring back to those prior successes,established and specialized expertise in the but if the change is new to the state of the art, thenarea of management innovation. A typical large the task of reducing ambiguity and uncertainty isorganization might employ tens or hundreds of much harder. Of course, all types of innovationscientists with technological innovation skills generate uncertainty and ambiguity, but their im-but few, if any, with proven management inno- pact in the case of management innovation isvation skills (the closest are organization devel- likely to be more far-reaching because of the restopment consultants, who seek systemic ways of of the attributes identified above.improving the overall effectiveness and health Taken together, these attributes suggest thatof the organization). This lack of expertise both the management innovation process can poten-heightens the uncertainty of management inno- tially require fundamental changes in the rou-vation for people across the organization and tines or DNA of the organization4 (Argyris &increases the need for external support. Schon, 1978) that make it very difficult to under- ¨ Third, the introduction of something new to take in an effective manner, and significantlythe state of the art creates ambiguity and uncer-tainty for the individuals in an organization.Ambiguity arises because of a lack of under- 4 We are indebted to an anonymous reviewer for suggest-standing of the intended value of the innovation, ing this point.
2008 Birkinshaw, Hamel, and Mol 831harder than the generic process of organizational main in use in a relatively small number ofchange (where the change is just new to the firms. But it is also possible for managementorganization rather than the state of the art) or fashions that are expressed in highly abstractthe process of technological innovation (where terms to spur management innovations. For ex-the innovation is relatively more tangible and ample, the knowledge management fashion ofless system dependent). These factors, in turn, the early 1990s led individuals and organiza-highlight the need for management innovators tions to put in place specific practices, such asto seek out distinctive approaches to building communities of practice, that were managementthe legitimacy of the new practice to make it innovations in their own right. We return to theacceptable to the various constituencies in the relationship between management innovationorganization (Ashforth & Gibbs, 1990; Green- and management fashion in the discussion.wood, Hinings, & Suddaby, 2002; Suchman, 1995). One such approach is likely to be a greater THE PROCESSES OF MANAGEMENTdegree of emphasis on independent validation INNOVATIONfrom external sources to establish the legitimacyof the new practice than would be the case for a The second part of this article addresses thegeneric organizational change activity (where question “What are the processes through whichprevious successful changes can be referred to) management innovations come about?” Build-or a technological innovation (which is more ing on our conception of what makes manage-likely to have objective benefits and/or a tech- ment innovation unique, we develop a frame-nical standard to which it subscribes). Such ex- work that highlights the four interlinked phasesternal sources can be useful providers of both of the process and the roles played by two keymoral and cognitive legitimacy in the absence sets of stakeholders. This framework is thenof hard evidence that management innovation fleshed out using theoretical arguments and ex-will be valuable and can allow the innovators to amples from the management literature.“manipulate the environmental structure by cre- The framework, illustrated in Figure 1, hasating new audiences and new legitimating be- two dimensions. The horizontal dimension con-liefs” (Suchman, 1995: 587). sists of four phases of the innovation process: (1) A second approach is likely to be for the inno- motivation is concerned with the facilitatingvators to focus their efforts on organizations (or factors and precipitating circumstances thatspecific units within organizations) with prior lead individuals to consider developing theirexperience in management innovation, on the own management innovation; (2) invention is anbasis that these organizations/units understand initial act of experimentation out of which a newthe challenge faced by the management inno- hypothetical management practice emerges; (3)vators and are therefore likely to be more toler- implementation is the technical process of es-ant of the uncertainty and ambiguity it brings tablishing the value of the new management(Kossek, 1989). In legitimacy-seeking terms, this innovation in vivo (i.e., in a real setting); and (4)can be seen as a strategy to “select among mul- theorization and labeling is a social processtiple environments in pursuit of an audience whereby individuals inside and outside the or-that will support current practices” (Suchman, ganization make sense of and validate the man-1995: 587). agement innovation to build its legitimacy. It is also useful to briefly consider the differ- This four-phase process builds on the in-ence between a management innovation and a trafirm evolutionary perspective advanced bymanagement fashion—a “relatively transitory Burgelman (1991) and Zbaracki (1998), wherebycollective belief that a management technique changes perceived in the environment (motiva-[or idea] leads to rational management tion) lead to variations in management prac-progress” (Abrahamson, 1996: 257). For the most tices (invention), some of which are then subjectpart, management innovations can be thought to internal selection (implementation) and reten-of as potential management fashions: some, tion (theorization and labeling). We expect thesuch as Six Sigma and the balanced scorecard, process to be shaped in large part by the con-become management fashions when they get scious and deliberate actions of key individuals,taken up by a significant number of manage- but we also recognize there is a role for unin-ment fashion users; others either die out or re- tended actions by individuals and random
832 Academy of Management Review October FIGURE 1 Management Innovation Process Frameworkchanges inside the organization in affecting the action researchers alongside the internal teamprocess of management innovation. during the implementation phase, and they can As per the vertical dimension in Figure 1, we play a role in theorizing about and labeling theexpect two groups of individuals to shape the innovation (Chandler, 1962; Kaplan, 1998; Pezet,process: (1) internal change agents, who are the 1997; Stjernberg & Philips, 1993; Yorks & Whit-employees of the innovating company proactive sett, 1985).in creating interest in, experimenting with, and A key feature of this framework is that itvalidating the management innovation in ques- does not assume a simple left-to-right se-tion (DiMaggio, 1988; Howell & Higgins, 1990), quence of activities. As Zbaracki observes, theand (2) external change agents, who, similar to processes of innovation typically will be com-Guillen’s (1994) management intellectuals and ´ plex, recursive, and occurring “in nested andAbrahamson and Fairchild’s (2001) idea entre- repeated cycles of variation, selection and re-preneurs, are independent consultants, academ- tention” (1998: 612). We address this point byics, and gurus proactive in creating interest in, focusing our attention on how individuals it-influencing the development of, and legitimiz- erate between the adjacent cells in the frame-ing the effectiveness and retention of new man- work, identifying ten core activities.5 For ex-agement practices (DiMaggio, 1991). As sug- ample, the activity “problem-driven search”gested earlier, we expect external change involves internal change agents’ iteratingagents to play a major role in management in-novation because they provide legitimacy and 5expertise in many different phases of the pro- It is quite possible to propose activities that span non- adjacent boxes—for example, between boxes 1 and 6. Ourcess. They can give credibility to the original analysis, however, suggests that such activities are proba-idea that sparks off the experiment inside the bly rare, so in the interests of keeping the framework rela-company, they can act as sounding boards or tively simple, we focus on the ten core activities.
2008 Birkinshaw, Hamel, and Mol 833back and forth between motivation and inven- agement practices is driven by the identificationtion, whereas the activity “agenda setting” of a novel problem—a perceived shortfall be-involves interaction between internal and tween the organization’s current and potentialexternal change agents (cf. Burgelman, 1983, performance6 (Barley & Kunda, 1992; Cyert &1991). March, 1963; Guillen, 1994). A perceived shortfall ´ Figure 1 identifies the ten core activities (indi- can be caused by a problem that underminescated by the double arrows and text spanning the current performance but also by opportunitiesboxes) and the nature of the innovation or its con- that may exist and the anticipation of environ-stituent parts (indicated by the numbered text mental changes (Cyert & March, 1963; Ocasio,within each box). Figure 1 also indicates the im- 1997). In some instances individual managers inportant role of context in shaping management an organization may attribute this shortfall sim-innovation. Organizational context is the admin- ply to a failure to execute under existing ar-istrative and social mechanisms that man- rangements, but in others they may identify aagement can manipulate to shape the behav- specific problem or opportunity vis-a-vis their `iors of actors in the organization (Bower, 1970; existing management practices. They engage inBurgelman, 1983) and will have a direct impact a problem-driven search process that begins(positive or negative) on the ability of internal with existing and proximate contacts, and oncechange agents to pursue the core activities asso- they find a satisfactory solution, they terminateciated with management innovation. Environmen- the search and implement the solution.tal context is the broad set of stimuli— exogenous In cases where the individuals choose to lookto the focal organization—that shapes the man- outside their own organization for a solution,agement discourse (Guille 1994) and thereby in- ´n, they are confronted with a management fash-fluences the priorities and efforts of external ion-setting community, which shapes the beliefchange agents as they engage with organizations. systems of users as to what is rational and ped-While these two aspects of context potentially in- dles its particular solutions to users’ problemsfluence all activities associated with manage- or perceived opportunities. Constrained both byment innovation, we discuss them in detail only in the pressure to conform to the norms of ration-those places where their role is critical. ality of the organization’s institutional field and by the costs of evaluating multiple competing offers, managers will often choose to adopt theMotivation Phase solution that appears to be the most progressive The motivation phase refers to the precondi- and legitimate (Abrahamson, 1996). This, oftions and facilitating factors that lead individu- course, is the process through which manage-als in a company to be motivated to experiment ment fashions spread.with a new management innovation. It ad- Sometimes, however, managers will choose todresses the question “Under what conditions, or experiment with developing their own solutionsin what circumstances, do executives deem ex- to the problem or performance shortfall they areisting management practices to be inadequate addressing. In the language of institutional the-for their needs?” The answer to this question is ory, such an act can potentially “appear irratio-far from straightforward because it is necessary nal and retrogressive” (Abrahamson, 1996: 263),not only to identify the conditions under which but there are several conditions under which itexecutives search for new management innova- may transpire—when the pressure to try some-tions but also to specify the circumstances in thing new overcomes the pressure to conform towhich they choose not to adopt one of the exist- externally arbitrated management norms. Weing solutions that can be obtained in prefabri- suggest two such conditions. The first is wherecated form from the so-called management fash-ion-setting community (Abrahamson, 1996). Formanagement innovation to occur, in other 6 We acknowledge that the notion of a perceived shortfallwords, the market for management fashions has implies some cognitive process through which environmen- tal changes are converted into action. In other words, someto fail. individuals will be able to interpret changes, whereas oth- Internal change agents. Consider first those ers may not. These cognitive processes, however, are not ouron the “demand” side of the market. Established central concern here, and, hence, we simply assume theytheory suggests that the demand for new man- take place.
834 Academy of Management Review Octoberinternal change agents are able, through an ronment that require management attention.agenda-setting activity, to frame the problem or But, as above, this is only part of the story be-opportunity such that internal stakeholders cause many external change agents see theirview it as genuinely new or as something that role as stimulating managers (through the agen-cannot be resolved by buying an existing solu- da-setting process) to adopt an existing or fash-tion from the fashion-setting community. Con- ionable practice, rather than to create a newsider, for example, the Danish hearing-aid man- one. We suggest that the nature of the manage-ufacturer Oticon. Its CEO at the time, Lars ment knowledge that external change agentsKolind, was able to convince his employees and share with their internal counterparts is an im-board of directors that Oticon faced a significant portant factor in motivating management inno-threat to its viability from large competitors, vation. One can identify a spectrum of manage-such as Philips and Siemens, and this was suf- ment knowledge, with new ideologies and ideasficient for him to push through a radical innova- at the more abstract end and new practices andtion that he labeled the spaghetti organization techniques at the more practical end. Those ex-(Foss, 2003; Lovas & Ghoshal, 2000). ternal agents who focus on the practical end of The second broad condition is when the or- the spectrum, with standardized or “off theganizational context is supportive of new think- shelf” solutions to the problems facing manag-ing and thereby enhances the degree of freedom ers, will encourage the adoption of managementfor internal change agents to pursue novel fashions. In contrast, those external changeideas. The notion of a supportive organizational agents who focus on the more abstract end ofcontext has been conceived in two broad ways the spectrum will more likely provide a fertilein the literature, both of which are potentially environment for management innovation be-relevant here. One set of arguments focuses on cause of the “interpretive viability” of theirthe role of management in creating an informal ideas—that is, the extent to which these ideascontext that encourages individuals to take ini- can be adapted to multiple agendas (Benders &tiative (Ghoshal & Bartlett, 1994). For example, van Veen, 2001; Clark, 2004).the more exposure managers have to different External change agents can interact both di-industries and organizations, the more receptive rectly and indirectly with internal changethey are likely to be to ideas for new practices agents in agenda setting.7 They generate their(e.g., Oldham & Cummings, 1996). The other set influential points of view by linking their inter-of arguments is more concerned with the formal pretation of changes in the environmental con-processes of the organization and the extent to text with agenda-setting conversations aboutwhich they institutionalize the pursuit of new or the practical issues executives face. They arebetter ways of working. For example, we might also influenced by prior cases of managementexpect the rigor of the decision-making process innovation they have been involved with; in Fig-in the organization to have a positive influence ure 1 this feedback loop is indicated by the threeon internal change agent motivation because, horizontal processes (idea contextualizing, ideaby clarifying the pros and cons, the uncertainty refining, and reflective theorizing) that we dis-and ambiguity associated with an idea are re- cuss in greater detail below.duced. In sum, internal change agents evaluate a Invention Phaseproblem or opportunity through an agenda-setting dialogue with external change agents Invention refers to either random or plannedthat helps to establish its novelty, and with ref- variations in management practices, some oference back to the supportiveness of the currentorganizational context. To the extent that the 7 An example of direct interaction is the U.K.-based “Be-problem or opportunity can be framed as novel yond Budgeting Round Table” (Hope & Fraser, 2003), whoseand the context is supportive, the preconditions founders worked actively with many organizations to helpfor management innovation exist. them make sense of the limitations of their traditional bud- geting systems. An example of indirect influence is Tom External change agents. The role of external Peter’s 1987 book, Thriving on Chaos, which was the inspi-change agents in motivating management inno- ration for an innovative organization structure at Welling-vation begins with their ability to identify new ton, a Canadian insurance company (Birkinshaw & Mol,threats and opportunities in the business envi- 2006).
2008 Birkinshaw, Hamel, and Mol 835which subsequently are selected and retained the birth of activity-based costing per se cameby the organization (Burgelman, 1991; Campbell, when Kaplan and the corporate executives be-1965). It is the phase in which a hypothetical new gan to interact.practice is first tried out in an experimental way. Finally, trial and error occurs when the feed- Internal change agents. Figure 1 suggests back about a new idea comes from trying it outthree ways in which internal change agents in practice, rather than from how well it solvesmight come up with a hypothetical new practice: an existing problem or how well it fits with theproblem-driven search, trial and error, and idea ideas of an external change agent. We can ex-linking with external change agents. While pect trial and error to be an important part ofeach of these subprocesses has value in its own any effective management innovation (when un-right, our expectation is that invention is more dertaken in combination with other activities),likely when they are applied in combination, but it is also possible for trial and error to be anjust as new technologies typically arise through unintended or ad hoc starting point for thenovel combinations of existing ideas and prac- whole process. For example, furniture retailertices (Hargadon, 2003: 65; Kogut & Zander, 1992; IKEA allowed its customers to pick up their ownSchumpeter, 1947). flat-pack products from the warehouse because Problem-driven search is a conscious and of- of staffing shortages at a busy time, and thisten planned activity in which individuals seek practice proved so effective (though in ways thatto create a new practice in response to a specific were not foreseen when it was first tried out)problem or opportunity (Cyert & March, 1963). that it was rapidly implemented in other storesChandler’s (1962) description of Alfred Sloan’s (Bartlett & Nanda, 1990). New practices canintroduction of the M-form structure suggests a emerge through serendipitous events of thisprocess of this type: Sloan’s proposed changes type, and they can also occur when an existingin 1920 were a direct response to the complexity practice is adapted to fit different circumstancesthat had been created by bringing together five (Czarniawska & Sevon, 2005; Mamman, 2002; ´independent businesses. As Sloan himself Sturdy, 2004).noted, “I wrote the ‘Organizational Study’ for External change agents. The role of externalGeneral Motors as a possible solution for the change agents in the invention phase mirrorsspecific problems created by the expansion of that of internal change agents. In other words,the corporation after World War I” (Sloan, 1963: their ability to come up with a new idea for32). management practices is a function of three of- Idea linking is when individuals in the orga- ten-linked activities: idea contextualizing, ideanization make connections between the new refining, and idea linking. Idea contextualizingideas proposed by external change agents and involves speculating on new ways of workingthe experimental efforts underway inside the that potentially address threats or opportunitiesorganization. Such connections can be viewed in the business environment. This is a commonas a form of brokering between relatively dis- activity among management thinkers, involvingparate networks (Granovetter, 1973; Hargadon, a back-and-forth interaction between the myr-2003), and they can be nurtured by encouraging iad of issues faced by managers on the one sideindividuals to read widely and to attend confer- and the set of possible solutions on the other.ences and other networking events. For exam- For example, Davenport and Prusak (2003: 179)ple, the concept of activity-based costing was describe how their initial ideas about knowl-developed by Robert Kaplan, a business school edge management emerged through the re-professor, through conversations at a conference search agenda of the Ernst & Young Centre forwith executives at Scovill Corporation and John Business Innovation.Deere Component Works about new cost mea- Idea refining can be viewed as a form of dis-surement approaches they were experimenting ciplined imagination (Weick, 1989), in which thewith (Kaplan, 1998: 98). Kaplan had been devel- external change agent works through the impli-oping his own ideas about ways of overcoming cations of a particular idea in terms of how itthe failure of existing cost measurement sys- might work in practice or in other contextualtems (idea contextualizing), and the Scovill and settings. Campbell (1974) viewed this activity asDeere executives had been experimenting in- “ideational trial and error”; it is directly analo-side their own organizations (trial and error). But gous to the process of trial and error that inter-
836 Academy of Management Review Octobernal change agents go through, but it occurs in proven methodology for tapping into externalthe conceptual domain. sources of technology (Birkinshaw, Crainer, & Idea linking, as discussed earlier, involves Mol, 2007). The other activity is reflective exper-reconciling the external change agent’s knowl- imenting, in which internal change agents eval-edge base (which is typically deep in terms of uate progress against their broader body of ex-academic discipline or functional expertise) perience. For example, Stjernberg and Philipswith the context-specific ideas of internal made the following observation about how suchchange agents. For example, activity-based individuals can be most effective:costing emerged through a combination of dis- As the [innovation] attempt proceeds, he [the in-satisfaction with existing accounting methods ternal change agent] needs to be able to learnand Kaplan’s perspectives on the changing from the consequences of his own actions and topressures on manufacturing companies, but it alter these actions accordingly. He will be morethen required an explicit link to Scovill and capable of seeing and learning how to manage the change and the learning dilemmas if he has aDeere for the concept to be put into practice well-developed capacity for reflection (1993: 1199).(Kaplan, 1998). Taken together, these three activities can be Organizational context also plays an impor-viewed as alternative but complementary ap- tant role in facilitating or inhibiting the imple-proaches to theory development: idea contextu- mentation of new ideas. Zbaracki (1998) ob-alizing is about developing new solutions to ex- served that the reaction of employees toisting problems, idea refining is about working implementing new management practices isthrough the consequences of an idea through a generally negative: they are likely to be intimi-series of “thought trials” (Weick, 1989), and idea dated by innovations, particularly if the innova-linking is an inductive-deductive loop through tions have a significant technical componentwhich concepts are reconciled with empirical and the employees are mostly ignorant of theirevidence. potential benefits. But the cultural perspective on management innovation suggests employ- ees’ reactions will also vary according to theirImplementation Phase personal circumstances and the immediate The implementation phase consists of all the work environment in which they are placedactivity on the “technical” side of the innovation (Knights & McCabe, 2000). The implementationafter the initial experiment up to the point where process is therefore likely to involve careful ma-the new management innovation is first fully neuvering by internal change agents as theyoperational. Like Zbaracki (1998), we distinguish focus their efforts on those parts of the organi-between the technical elements of the work and zation that are more amenable to change. As thethe rhetorical elements that are concerned with literature on technological innovation de-theorizing and labeling the innovation (dis- scribes, such tactics include pursuing corridorscussed in the next section). Our description of of indifference through the organization, build-this phase involves making sense of the actions ing coalitions of senior executives to supportof internal and external change agents in imple- their ideas, framing innovation as an opportu-menting an in vivo new practice, as well as nity and not a threat, accessing resources be-understanding the ways existing employees re- yond the individual’s control, and maintaining aact to it and influence its implementation generally tenacious and persistent attitude(Lewin, 1951). (Howell & Shea, 2001; Rothwell et al., 1974; Schon ¨ Internal change agents. Figure 1 indicates two 1963; Wrapp, 1967).primary activities that internal change agents Taken as a whole, the literature suggests thatengage in as they attempt to implement an in implementation transpires through a dialecticalvivo new practice. One is trial and error, in process (Van de Ven & Poole, 1995). Internalwhich progress is achieved by monitoring and change agents try out the proposed new prac-making adjustments against the original con- tice, and they evaluate its progress against thecept. For example, to develop Procter & Gam- original idea (trial and error), its conceptual va-ble’s “Connect and Develop” innovation pro- lidity (reflective experimenting), and the reac-cess, its originator, Larry Huston, observed that tions of other employees (i.e., the organizationalhe ran six years of experiments before he had a context). Some aspects of the new practice may
2008 Birkinshaw, Hamel, and Mol 837prove to be unworkable, and the reactions of vention experiments” (Argyris & Schon, 1991: 86). ¨employees may in some cases be directly op- In terms of our framing, the external changeposed to what is being pursued. But after sev- agent therefore plays a dual role, oscillatingeral iterations, an outcome will often emerge back and forth between his or her thought ex-that is a synthesis of the opposing forces periment about what might make sense in the(Knights & McCabe, 2000; Zbaracki, 1998). In world of management ideas and the in vivo im-other cases the internal resistance generated by plementation of what actually works in thevarious aspects of the organizational context world of practice. This dual role potentially of-may be sufficiently strong that the experimental fers great insights to both worlds. Unfortunately,new practice does not get taken forward at all. though, the evidence suggests interventions of External change agents. The role played by this type are on the decline. While action re-external change agents in the implementation search has an illustrious past (Emery & Trist,phase is less clear-cut than in other phases. 1960; Lewin, 1946), it has lost ground in recentExternal change agents lack deep contextual decades to more passive forms of research, aknowledge of the focal organization, as well as point we return to in the discussion.the accountability for results that most internalchange agents face, so they rarely play an ac- Theorization and Labeling Phasetive role in actually implementing new ideas invivo. However, we suggest they potentially play The fourth phase results in a theorized newa critical indirect role in making management practice— one that is retained and institutional-innovation happen. ized within the organization. While an effective The essence of the external change agent’s implementation, as described above, is clearly arole is to create a thought experiment (analo- necessary part of the process, the intangible andgous to an in vitro experiment performed by a system-dependent nature of management inno-biologist before a new molecule is tried out in vation means that the results of the implemen-vivo in a live body). External change agents tation are likely to be highly unclear for severaldraw from their prior experience (reflective the- years (Teece, 1980). We therefore expect thatorizing) and their deep knowledge of a particu- there will be an important rhetorical componentlar conceptual domain (e.g., an academic disci- associated with a successful management inno-pline or a functional competency) to sharpen vation. Key change agents will seek to make thetheir new idea (idea refining), and on the basis case with constituencies inside and outside theof the insights gained, they attempt to influence organization that the new practice is legitimate,and direct the implementation efforts of the in- even though this new practice represents (byternal change agents (idea testing). There is definition) a departure from the tried-and-testedsome evidence of what this set of activities looks offerings of the fashion-setting communitylike in practice. For example, Stjernberg and (Abrahamson, 1996; Suchman, 1995).Philips (1993) highlight the roles external We view this phase as consisting of two inter-change agents play as facilitators and sounding linked elements: theorization and labeling. The-boards, and Kaplan provides a thoughtful ac- orization is increasing “the zone of acceptancecount of his own experiences in this area: by creating perceptions of similarity among adopters and by providing rationales for the During this process of intimate engagement with implementation, the action researcher [i.e., the practices to be adopted” (Greenwood et al., 2002; external change agent] not only advances the Strang & Meyer, 1993; Suddaby & Greenwood, theory underlying the concept, but also becomes 2005; Tolbert & Zucker, 1996). In the context of a skilled practitioner. . . . Such skill also enables this article, theorization is therefore first about the action researcher to distinguish between the- building a logical rationale for the link between ory limitations vs. poor implementations when companies experience difficulties applying the an organization’s opportunities and the innova- innovation (1998: 106). tive solution that is being put in place, and sec- ond about expressing that logic in terms that These activities can, as Kaplan suggests, be resonate with key constituencies inside or out-thought of as a form of action research—where side the organization. Labeling refers to the se-the aim is to “build theories within the practice lection of a name for the management innova-context itself, and test them there through inter- tion in question that reflects its theorization.
838 Academy of Management Review OctoberLabels have been shown to have a significant reflective experimenting (whereby the neweffect on the acceptability of management prac- practice is interpreted in light of the internaltices to various constituencies (Eccles & Nohria, change agents’ broader body of experience) and1992; Kieser, 1997). theory linking with external change agents (by Internal change agents. The primary role of talking to them directly, by reading their books,internal change agents in this phase is to build or by listening to them speak).legitimacy for the innovation among employees External change agents. The role of externalof the organization.8 As a means of defusing the change agents in the theorization and labelingwidespread skepticism toward management in- phase is twofold. First, they have an importantnovation that employees often exhibit (Knights role to play in building cognitive legitimacy in-& McCabe, 2000), internal change agents will side the organization, because their status asoften theorize about the value of the new prac- independent experts means they are broughttice and label it in such a way that employees in—for example, as speakers at companysee its potential value, and also see it as con- events—to verify both the significance of thesistent with the prevailing norms of the organi- challenge the organization is facing and the va-zation. The outcome, in other words, is a new lidity of the proposed innovation as a responsepractice that has been theorized vis-a-vis the ` to that challenge. This form of input is referredimmediate organizational context (whereas ex- to in Figure 1 as theory linking.ternal change agents focus on theorizing be- External change agents also play a major roleyond this immediate context). in building legitimacy for the innovation beyond It is useful to apply Suchman’s (1995) three the boundaries of the organization. This is oftenbasic forms of legitimacy to help clarify the ap- deemed by the organization to be a worthwhileproaches used here. Pragmatic legitimacy (ap- activity, because most employees have somepealing to employees’ self-interested calcula- level of awareness of how their organization istions) is pursued by showing early evidence of viewed by external constituencies (through cus-the innovation’s value and by allaying employ- tomers and outside partners, friends, or the me-ees’ concerns, but such evidence is likely to be dia), so their opinion of the innovation is shapedhard to come by in the early stages of implemen- to some degree by what external constituenciestation. Moral legitimacy (a positive normative say about it. Influential media such as newspa-evaluation through consistency with the organi- pers and magazines have been shown in otherzation’s value system) is pursued by playing up contexts to play an important role in legitimat-how the innovation builds on previous changes ing the actions of individual executives and or-the company has been through and/or that the ganizations (Deephouse, 2000; Mazza & Alvarez,organization has a tradition of trying out new 2000; McQuail, 1985; Pollock & Rindova, 2003),ideas. Finally, cognitive legitimacy (the devel- and we would expect them to play an importantopment of plausible explanations for the inno- role here as an indirect communication channelvation that mesh with larger belief systems and through which employees’ attitudes toward thethe experienced reality of the audience’s daily innovation are shaped.life) is pursued by showing that management The externally focused theorization and la-innovation is a necessary solution to a specific beling process involves a set of challengesand novel challenge the organization is facing different from that of the internally focused(Tolbert & Zucker, 1986). process. The external constituency is typically This form of internally focused theorization management intellectuals, such as seniorand labeling is best performed by internal leaders in other organizations, journalists,change agents because of their existing credi- consultants, and academics (Guillen, 1994). ´bility with employees (Stjernberg & Philips, These individuals operate at a more abstract1993). It is achieved through a combination of level than employees (i.e., they are likely to focus on management ideas rather than prac- 8 tices), they are more positively disposed to- There is also potentially a role for internal change ward management innovation than employeesagents in building legitimacy for external stakeholders, al-though external change agents typically do this rather more (because they run no personal risk of failure),effectively. We discuss this possibility in the following sec- and they have a less detailed understandingtion. of the innovation than employees. As a result
2008 Birkinshaw, Hamel, and Mol 839of these differences, the approaches used to and personal networks of external changebuild legitimacy for the innovation to external agents, they have greater credibility throughaudiences are likely to be somewhat different. their personal championing of the process,Pragmatic legitimacy can only be achieved which may help to establish the moral legiti-vis-a-vis external constituencies by demon- ` macy of the innovation.strating that the innovation is yielding valu-able outputs. Procter & Gamble claimed, forexample, that the company increased the per- DISCUSSION AND AVENUES FORcentage of new products arising from external FUTURE RESEARCHideas to 40 percent as a result of its Connectand Develop innovation process (Huston & Here we have argued that management in-Sakkab, 2006). However, such evidence is rel- novation is an important phenomenon in theatively hard to put together in the early stages field of management and that the generativeof a management innovation, for the reasons mechanisms through which it occurs (i.e.,we have discussed. management innovation processes) are theo- Moral legitimacy is pursued by seeking a retically interesting in their own right, andpositive normative evaluation of the innova- also relatively poorly understood. We have de-tion among managerial intellectuals, which veloped a framework highlighting the impor-may involve showing how the innovation is tant roles of internal and external changeprocedurally consistent with existing manage- agents in the process and the ways these twoment practice (e.g., Six Sigma was positioned sets of actors interact with one another. Ouras a successor to total quality management; framework suggests a number of importantHarry & Schroeder, 2000), or it may involve insights, and it opens up some interesting an-demonstrating the credentials of the organiza- gles for further research.tion as a proven high performer with a trackrecord of innovation. Cognitive legitimacy, in Sequencing of Management Innovationcontrast, is typically pursued by framing the Activitiesinnovation as a logical solution to one of thegeneric challenges or problems that all large We first observed that the process of manage-organizations face. In Kieser’s words, “The im- ment innovation does not always proceed as aplementation of the new principles is pre- linear sequence of activities from motivationsented as unavoidable [by management gu- through to theorization and labeling. For exam-rus], because the old principles are bound to ple, an organization that suffers from too muchfail in the face of the menacing dangers” (1997: “smart talk” (Pfeffer & Sutton, 2000) may have57). This approach is similar to the pursuit of several initiatives that are well progressed incognitive legitimacy with employees, except terms of motivation and theorization and label-here the arguments will be expressed in more ing, but with no commensurate investment inabstract or generic terms. invention and implementation. In such a case External change agents typically develop the appropriate managerial intervention mighttheir knowledge of a particular innovation be to focus attention on implementation as athrough prolonged interactions with internal means of establishing which initiatives arechange agents (through idea-linking, idea- worth pursuing, whereas other organizationaltesting, and theory-linking activities) and settings might require different interventions.through their own reflective theorizing. External However, at the moment we know little aboutchange agents have the skills for contextualiz- the relative effectiveness of different sequencesing the innovation in terms of contemporary of activities, which makes it difficult to offer anybusiness challenges, as well as the necessary coherent advice to managers about how to im-contacts with media organizations. It should be prove the quality of their interventions. Ourobserved that internal change agents can also framework focuses on the “activities,” such ashelp to build legitimacy for management inno- trial and error, that take place between the ad-vation with external constituencies by writing jacent cells in Table 1 as a means of highlight-articles or books and speaking at conferences. ing that innovation is an iterative process. ButAlthough they may lack the theorization skills an important next step in making sense of the
840 Academy of Management Review Octoberoverall process of management innovation oscillate between the two roles over the coursewould be to examine the actual sequencing and of their careers (Davenport & Prusak, 2003).phasing of activities over time. While some in- One avenue for further research, then, is tonovations may follow a linear sequence of ac- take a closer look at the key change agentstivities from left to right, others do not; for exam- involved in management innovation and the ex-ple, Procter & Gamble’s Connect and Develop tent to which they are able to take on hybridprogram was developed as a concept by its internal/external roles. A second line of inquirychampion, Larry Huston, long before the compa- might be to consider the extent to which internalny’s CEO articulated the need for it (Birkinshaw and external change agents are acting in har-et al., 2007). Likewise, while we would expect mony. Here we have assumed that both partiesmost management innovations to be initiated have a more or less common objective—namely,primarily by internal change agents, it is possi- to implement a successful management innova-ble to identify others, such as T-Groups (Benne, tion. However, future research might want to1964; Blake, 1995), where many of the core activ- relax this assumption and consider the extent toities were driven by external change agents. which the two parties are truly aligned. For ex-Future research should attempt to map out and ample, in attempting to build legitimacy for amake sense of the sequences that actually occur management innovation among internal andin practice. The historical record is not particu- external constituencies, internal change agentslarly helpful in this regard, because writers will may downplay the scale of change required totypically impose their own structure on a pro- their colleagues (perhaps by emphasizing con-cess in order to make sense of it. Research will sistency with prior norms or a low level of risk)therefore need to be done on contemporary to make the change more palatable, while ex-cases, and, where possible, these cases should ternal change agents may exaggerate the scalebe followed in real time to avoid problems of of the proposed innovation (perhaps by position-retrospective sensemaking bias. ing it as the antidote to dramatic changes in the industry) as a way of generating interest amongThe Role of Internal and External Change external audiences. These differences in posi-Agents tioning can potentially have deleterious conse- quences for the individuals involved, as well as Another core element of our framework is the for the long-term success of the innovation.distinction between internal and externalchange agents. As a matter of definition, inter- A third avenue for future research that alsonal change agents are employees of the focal builds on the internal versus external distinc-organization whereas external change agents tion is to examine the locus of management in-are not, which, in turn, implies that internal novation. Our framework assumes that it ischange agents will typically have superior possible and meaningful to identify the organi-knowledge and networks inside the organiza- zation in which a new management practice istion and greater accountability for delivering first implemented. While this approach is validresults than their external counterparts. How- vis-a-vis the existing cases we mentioned, there `ever, it is important for future research to con- may be cases where it is less valid in the future.sider this distinction more carefully, since it Increasingly, economic activity transpiresmay not always be clear-cut in practice. Con- through nonfirm networks, such as open-sourcesultants, for example, are sometimes seconded software communities, so we can expect innova-to their client companies during a change pro- tive ways of organizing to emerge that enablecess, and ethnographic researchers will often nonfirm coordination of this type. It is also pos-become employees in the organizations they are sible, although less likely, that more manage-studying for significant periods of time. In both ment innovations will emerge in vitro in thecases these external actors actually become in- future, perhaps through the efforts of academicsternal actors on a temporary basis. Moreover, rather than the trial and error of practicing man-there is evidence that some individuals are able agers, in which case the locus of innovation,to switch back and forth between internal and again, would not be the organization. Futureexternal change agent roles during a single studies should therefore give careful attentionproject (i.e., as action researchers), while others to the unit of analysis at which management
2008 Birkinshaw, Hamel, and Mol 841innovation is studied, since there are several to which management innovation helps organiza-possible models that could be followed. tions to fulfill their goals are equally important. It seems likely, for example, that certain manage- ment innovations will offer more potential forManagement Innovation and Management competitive advantage than others, depending onFashion the extent to which they are valuable, rare, and We have argued that the management inno- hard to imitate (Barney, 1991), but this argumentvation process is triggered when the market for remains open to empirical testing.fashion fails—that is, when an organization pur- The consequences of management innovationsues its own novel practice rather than one sug- are complex, because so many different stake-gested by the fashion-setting community. How- holders are potentially affected. It is necessary toever, this argument obscures the important separate out at least three different sets of conse-point that, in many ways, the management fash- quences: (1) the impact of management innovationion process has important similarities to the on various performance metrics inside the inno-management innovation process: both involve vating firm; (2) the impact on the performance andsignificant roles for internal change agents legitimacy of subsequent adopters of the innova-(Abrahamson  calls them “users” of man- tion; and (3) the benefits of management innova-agement fashions) and external change agents tion to society as a whole, in terms of improve-(“suppliers” of management fashions), as well ments of such things as productivity or quality ofas complex interactions between the two. And work life. As noted earlier, there has been someboth can be framed in evolutionary lan- research on the second of these (e.g., Staw & Ep-guage—in terms of the introduction of some- stein, 2000), but the first and third remain largelything new to the organization that subsequently unexplored. Future research might therefore ex-gets selected and retained, or not. amine why certain types of management innova- A useful direction for future research, then, will tion take longer to yield dividends than others,be to look more closely at how the processes of whether some management innovations spurmanagement innovation and management fash- waves of related innovation, and how often andion interact. It may be possible, for example, to under what circumstances management innova-identify ways in which external change agents, tion creates firm-specific competitive advantage.such as consultants, influence the emergence ofmanagement innovation, either by suppressing The Role of Academia in Managementthe level of novelty in the focal organization’s cho- Innovationsen solution (e.g., by pushing their own off-the-shelf solutions, regardless of the user’s agenda) or Finally, this article offers some initial thoughtsby enhancing it (e.g., by encouraging users to de- on the role academics can play in the process ofvelop their own agendas and by putting forward management innovation. Like Abrahamson andideas with interpretive viability). Another ap- Fairchild (2001), we are concerned that academicsproach might be to examine the conditions under may be losing out to other members of the fashion-which a management innovation gets picked up setting community, such as consultants and gu-by the fashion-setting community and turned into rus, in terms of their ability to influence practice.a management fashion. At an abstract level such Our framework suggests some possible ways for-practices are likely to have highly “progressive” ward. One is for academics to become more cre-and contemporary labels (Abrahamson, 1996) and ative in the development of new ideas andare likely to exhibit high levels of external change thought experiments that organizations might putagent involvement, but there is room for a much into practice. Another is to become more engagedgreater level of clarity on what these conditions in the activity we call “idea testing,” whereby thelook like in detail. academic engages closely with the focal organi- zation and brings his or her insight to bear on the particular problem the organization is grapplingManagement Innovation and Firm Performance with. This concept of engaged scholarship (Van de While our focus in this article was primarily on Ven & Johnson, 2006) has a long history, but itsprocess issues, questions about why individuals legitimacy as a valid form of scholarship has fal-engage in management innovation and the extent tered in recent decades. Another alternative is a