EXPLOITATION, EXPLORATION, AND PROCESS … · MARY I. BENNER University of Pennsylvania MICHAEL L. TUSHMAN ... ature with contingent theory and testable prop-ositions about process

^ Academy ol Management Review2003, Vol. 28, No, 2, 238-256,

EXPLOITATION, EXPLORATION, ANDPROCESS MANAGEMENT: THE PRODUCTIVITY

DILEMMA REVISITED

MARY I. BENNERUniversity of Pennsylvania

MICHAEL L. TUSHMANHarvard Business School

We develop a contingency view of process management's influence on both techno-logical innovation and organizational adaptation. We argue that while process man-agement activities are beneficial ior organizations in stable contexts, they are fun-damentally inconsistent with all but incremental innovation and change. But dynamiccapabilities are rooted in both exploitative and exploratory activities. We argue thatprocess management activities must be buffered from exploratory activities and thatambidextrous organizational forms provide the complex contexts for these inconsis-tent activities to coexist.

Twenty-five years ago, Abernathy (1978) sug-gested that a firm's focus on productivity gainsinhibited its flexibility and ability to innovate.Abernathy observed that, in the automobile in-dustry, a firm's economic decline was directlyrelated to its efficiency and productivity efforts.He suggested that a firm's ability to competeover time was rooted not only in its ability toincrease efficiency but also in its ability to beefficient and innovative simultaneously (Aber-nathy, 1978: 173; Hayes & Abernathy, 1980). Strat-egy and organization theorists have similarlyobserved that dynamic capabilities are an-chored in a firm's ability to both exploit andexplore (Ghemawat & Costa, 1993; March, 1991;Weick, 1969). A firm's ability to compete overtime may lie in its ability both to integrate andbuild upon its current competencies while si-multaneously developing fundamentally newcapabilities (Teece, Pisano, & Shuen, 1997).

Twenty-five years after Abernathy's observa-tions, the pressures for organizations to meetmultiple, often inconsistent, contextual de-mands have escalated (e.g., Christensen, 1998;Tushman & O'Reilly, 1997). The notion of bal-

We thank Diane Burton, Don Hambrick, Andy Henderson,Monica Higgins, Lee Fleming, David Garvin, GiovanniGavetti, Peter Kolesar, Nitin Nohria, Nelson Repenning, San-dra Sucher, Ruth Wageman, former associate editor Devlennings, and three anonymous reviewers for comments onearlier drafts oi this article.

ance between exploitation and exploration, orbetween incremental and radical organization-al change, has been a consistent theme acrossseveral approaches to research in organization-al adaptation {e.g.. Brown & Eisenhardt, 1998;Burgelman, 1994; Gavetti & Levinthal, 2000;Levinthal & March, 1993; March, 1991; Tushman& Romanelli, 1985). Yet this need for dual organ-izational capabilities arises in the context of awave of managerial activity and institutionalpressures focusing on process management andcontrol {e.g., Adler, 1993; Cole, 1998; Hackman &Wageman, 1995; Hammer & Stanton, 1999;Winter, 1994).

Process management, based on a view of anorganization as a system of interlinked pro-cesses, involves concerted efforts to map, im-prove, and adhere to organizational processes.Initially based on the seminal work of Ishikawa(1985), Deming (1986), and Juran (1989), processmanagement practices became popular as acentral element of total quality management(TQM) programs in the 1980s (Hackman & Wage-man, 1995). Since then, these practices have con-tinued to spread as a core element of a continu-ing progression of quality-related initiatives,including the Malcolm Baldrige National Qual-ity Award, the International Organization forStandardization's Series 9000 program (ISO9000), business process reengineering, and,more recently. Six Sigma programs. By 1992

238

2003 Bennei and Tushman 239

every Fortune 100 firm had adopted TQM prac-tices (Nohria, 1996).

While some suggest that interest in TQM hadwaned by the 1990s (Powell, 1995), thousands oforganizations were subsequently certified in theISO 9000 process management program (QualityDigest, 1999), and many companies, includinghighly visible ones like GE, Honeywell, 3M,Amazon.com, Toshiba, and Ford have recentlyembraced Six Sigma (e.g., Feyder, 2001; Gabor,2001; Honeywell, 1998). Process managementpractices have been institutionally mandatedas government agencies or powerful buying or-ganizations require adoption by their suppliers{e.g., Harrington & Mathers, 1997; Westphal,Gulati, & Shorten, 1997).

Process management's contribution to im-proving manufacturing efficiency has led to itsmigration beyond operations to other parts oforganizations—for instance, to adjacent pro-cesses for selecting and developing technologi-cal innovations (Brown & Duguid, 2000; Sitkin &Stickel, 1996). As the variation-decreasing andefficiency-oriented focus of process manage-ment spreads to centers of innovation, or varia-tion creation activity in organizations, it increas-ingly affects an organization's dynamiccapabilities. Yet there has been a lack of re-search about how these institutionally man-dated and pervasive practices affect technolog-ical innovation or adaptation.

Much existing literature is prescriptive andaimed at educating managers on implementingprocess management practices. Process man-agement's proponents have promoted processimprovement practices as universally beneficialfor organizations, spurring continuous innova-tion that results in efficiency improvements, costreductions, improved customer satisfaction,and, ultimately, higher profits (Hammer & Stan-ton, 1999; Harry & Schroeder, 2000; ISO, 1999).Reflecting these assumptions, empirical re-search on process management's effects hasbeen limited to assessing the financial perfor-mance implications from process managementadoption (e.g., Ittner & Larcker, 1997; Powell,1995; Samson & Terziovski, 1999).̂ The results of

' In other empirical research scholars explore implica-tions of more broadly defined TOM programs. While processimprovement practices may be included in an organization'sadoption of TQM, this is not necessarily the case (Westphalet al,, 1997, Zbaracki, 1998). Further, since it is not clear from

these studies have been equivocal. Researchaimed at resolving the debate suggests that dif-ferential outcomes from TQM adoption arise be-cause organizations implement different prac-tices under the TQM umbrella (Westphal et al.,1997; Zbaracki, 1998). Similarly, Sitkin, Sutcliffe,and Schroeder (1994) argue for an alternative setof TQM practices when task environments arehighly uncertain.

Although these studies provide insight intopossible contingencies, their focus is specifi-cally on TQM programs. Similarly, much of theorganizational literature on the topic of processmanagement has been focused on TQM (e.g.,Easton & Jarrell, 1998; Hackman & Wageman,1995; see also the July 1994 Total Quality specialissue of AMR and Cole & Scott, 2000). Research-ers have made no attempt to build theory thatlinks this TQM literature to broader concepts ofprocess management or with more recent re-search on dynamic capabilities (e.g., Eisenhardt& Martin, 2000; Teece et al., 1997).

We extend the process management literatureby developing a model and testable proposi-tions about how process management activitiesaffect both technological innovation and organ-izational adaptation. We explore how both tech-nological and organizational contexts moderatethe relations between process-focused activitiesand organizational adaptation (e.g., Sitkin et al.,1994), arguing that process management tech-niques stabilize and rationalize organizationalroutines while establishing a focus on easilyavailable efficiency and customer satisfactionmeasures. Although increased efficiency resultsfrom these dynamics in the short run, they alsotrigger internal biases for certainty and predict-able results and ensure process management'sprogression to more activities throughout thefirm. This diffusion of process managementtechniques favors exploitative innovation at theexpense of exploratory innovation. We arguethat while exploitation and inertia may be func-tional for organizations within a given techno-

most research on TQM whether and to what extent the or-ganizations involved undertook process-focused practices,interpreting or comparing existing research is hampered byheterogeneity in the bundle of TQM practices adopted. Thethree studies cited break out process-focused activities fromother practices within TQM and, thus, are particularly relevantto our topic. We also review the larger body ol existing re-search on process management and TQM later in this paper.

240 Academy of Management Review April

logical trajectory or for existing customers, thesevariance-reducing dynamics stunt exploratoryinnovation and responsiveness to new customersegments (Henderson et a l , 1998; Sterman, Re-penning, & Kofman, 1997). Finally, we explorehow ambidextrous organizational forms providebuffered contexts such that exploitation and ex-ploration can coexist (Bradach, 1997; Tushman &O'Reilly, 1997).

We proceed in three sections. In the first wedefine process management techniques and ex-plore empirical evidence of their impact on organ-izational outcomes. In the second we explore theeffects of process management techniques on ex-ploratory as well as exploitative innovation. Wealso consider the structural features and moderat-ing effects of ambidextrous organizational de-signs that isolate process control activities fromexploratory activities. Finally, in the third sectionwe explore how process management practices,through their influence on both technological in-novation and organizational inertia, affect adap-tation. We also explore the organizational andenvironmental contingencies that moderate pro-cess management's effects.

This paper contributes to research in severalways. We extend the process management liter-ature with contingent theory and testable prop-ositions about process management's effects onorganizations. We also contribute to organiza-tion and strategy research considering how sta-ble processes influence outcomes like techno-logical innovation, organizational adaptation,and failure (Christensen & Bower, 1996; Hannan& Freeman, 1984; Henderson & Clark, 1990;Levinthal & March, 1993; Nelson & Winter, 1982).Our model helps resolve inconsistent empiricalfindings in the process management literature,while providing a base for further research intothe relationship between process managementactivities and dynamic capabilities. Although inprior research scholars have examined how rou-tines influence organizational outcomes, theyhave not fully explored the effects on innovationor adaptation of institutionally mandated qual-ity programs aimed at stabilizing organization-al routines and processes.

BACKGROUND: THE PROMISE AND REALITYOF PROCESS MANAGEMENT

The process revolution has been marked by ashift from the view of organizations as a collec-

tion of departments with separate functions andoutputs to a view of them as systems of inter-linked processes that cross functions and linkorganizational activities (Dean & Bowen, 1994).Processes are collections of activities that, takentogether, produce outputs for customers (Garvin,1998; Ittner Sc Larcker, 1997). Customers includenot only external consumers of the organiza-tion's products or services but also a series ofinternal recipients at linkage points betweenprocesses, as outputs from upstream processesbecome the inputs for subsequent processes. Al-though programs and awards like TQM, ISO9000, Six Sigma, the Malcolm Baldrige Award,and business process reengineering differ inscope and approach, they share a core focus onmeasuring, improving, and rationalizing organ-izational processes (cf. Hammer & Champy,1993; Harrington & Mathers, 1997; Harry &Schroeder, 2000; Ittner & Larcker, 1997; Powell,1995; Repenning, 1999; Winter, 1994).

Process management entails three main prac-tices: mapping processes, improving processes,and adhering to systems of improved processes.Once underlying processes have been recordedthrough process mapping, process improvementinvolves using measures of process effective-ness and statistical methods to continuallyeliminate variation in processes and outputs(Hackman & Wageman, 1995; Harry & Schroeder,2000). Process improvement involves not onlyrationalizing individual work processes but alsostreamlining the handoffs between processes(Garvin, 1995; Harry & Schroeder, 2000). As or-ganizational participants are trained in effec-tive ways to facilitate cross-functional teammeetings and learn standard approaches foridentifying and solving problems, they tightlyintegrate and coordinate a broad set of activi-ties throughout the organization (Hackman &Wageman, 1995; Repenning, 1999).

Process management programs also stressongoing adherence to the resulting mapped andimproved processes. Repeatable processes al-low organizations both to reap the benefits ofimprovement efforts and to continue incremen-tal improvements (Hackman & Wageman, 1995;Harrington & Mathers, 1997; Mukherjee, Lapre, &Van Wassenhove, 1998). The ISO 9000 program,in particular, has a strong, explicit focus on ad-hering to documented processes. ISO 9000 certi-fication requires third-party audits to periodi-cally ensure that an organization follows its

2003 Benner and Tushman 241

documented practices (Cole, 1998; Harrington &Mathers, 1997). Similarly, the final stage of SixSigma implementation requires coordinatingstandardized best practices throughout an or-ganization (Harry & Schroeder, 2000).

Process management's proponents cite nu-merous organizational benefits from process im-provement activities. These include increasedyields and less rework and waste as stream-lined processes eliminate non-value-added ac-tivities. Tighter intraorganizational linkagesalso increase efficiency by streamlining thehandoffs between activities, thus speedingproduct development and delivery times (Dean& Snell, 1996; Garvin, 1995). Further, productsthat result from improved processes are likely tobetter satisfy customers, leading to increasedrevenues and, ultimately, increased profits.

Yet empirical research on the effects of pro-cess management practices fails to yield con-clusive evidence of these promised benefits. Thefew empirical studies that consider the implica-tions of process-focused techniques show no ev-idence that these techniques are consistentlyhelpful Powell (1995) and Samson and Terzi-ovski (1999) found no relationship between pro-cess management utilization measures and or-ganizational performance. Ittner and Larcker(1997) found process management techniquesassociated with increased performance in theautomobile industry, but decreased perfor-mance in the computer industry. Others havenoted that poor financial performance has fol-lowed the process-focused efforts necessary forwinning the Baldrige Award (Garvin, 1991; Hill,1993). Winners like Motorola, Cadillac, and Fed-eral Express have suffered subsequent financialsetbacks, leading some Wall Street analysts torecommended shorting the stocks of BaldrigeAward winners (Garvin, 1991).

In related research scholars also have ex-plored the organizational effects of TQM pro-gram adoption. Because TQM programs includeactual implementation of a different bundle ofassociated practices within each organization(Hackman & Wageman, 1995; Westphal et al.,1997), it is difficult to determine whether organi-zations implement process-focused techniques.Even so, this research has also produced con-flicting or paradoxical results. Kearns and Nad-ler (1992) and Wruck and Jensen (1994) report onTQM efforts that led to substantial organization-al improvements. Similarly, event studies have

shown performance benefits from adoption of aTQM program (Easton & Jarrell, 1998) and fromwinning a quality award (Hendricks & Singhal,1996), and Dean and Snell (1996) found that TQMas one component of integrated manufacturingwas associated with performance benefits.

In contrast. Analog Devices' adoption of aTQM program resulted in short-term improve-ments in efficiency, such as higher yields andless waste, but this was soon followed by finan-cial performance far worse than the industryaverage (Sterman et al., 1997). Similarly, proac-tive quality efforts at Alcoa initially led to com-petitive gains (Kolesar, 1993), yet soon after thisquality effort was initiated, Alcoa's CEO PaulO'Neill lamented the slow pace of change andinitiated a "quantum-leap improvement" (Kole-sar, 1993: 161). Henderson and colleagues (1998)found that despite significant returns to processimprovements at Hewlett Packard, these activi-ties were underfunded out of concerns that pro-cess-focused efforts would impede product inno-vation. Sitkin and Stickel (1998) found that TQMefforts in an R&D setting led to organizationalconflicts, distrust, conformity, and, in turn, todecreased innovation. And Staw and Epstein(2000) found that while TQM adoption was asso-ciated with higher CEO compensation, therewas no association between adoption and fi-nancial performance.

In other empirical research scholars haveaimed at resolving these inconsistent findings.This work suggests that outcomes fall short ofanticipated benefits, because while firms adoptTQM, they fail to fully implement the associatedefficiency-generating practices (Easton & Jarrell,1998; Westphal et a l , 1997; Zbaracki, 1998) or failto give them sufficient time or the right culturein which to work (e.g., Cameron & Barnett, 2000;Sterman et a l , 1997). Further, it may be that, inimplementing TQM and related process im-provement activities, firms drive in fear, which,in turn, triggers resistance to these administra-tive practices (Repenning, 2000).

It may also be that the effects of process man-agement activities are contingent on the techno-logical and organizational contexts in whichthey are applied. The initial successes of pro-cess management activities have spurred theirspread beyond manufacturing into upstream ac-tivities, such as processes for selecting and de-veloping technological innovations, or down-stream, into distribution, sales, and service


activities. Indeed, the 9001 version of the ISO9000 program involves processes for product de-sign, development, and service, while Design forSix Sigma similarly promotes extending processcontrol techniques into R&D, including productdesign and development activities (Harry &Schroeder, 2000; ISO, 1999). Process manage-ment practices have also been applied to seniorteam processes (e.g., Garvin, 1995; Ghoshal &Bartlett, 1995; Hammer & Stanton, 1999).

Process management practices may beadopted because of bandwagon effects, createdwhen other high-status organizations adoptthem (Abrahamson, 1996; Abrahamson & Rosen-kopf, 1993; Meyer & Rowan, 1977; Staw & Epstein,2000). As these techniques are embraced byhighly visible organizations like IBM, GE, 3M,and Amazon.com, among others (Hammer &Stanton, 1999), they become taken for grantedand viewed as sensible approaches for operat-ing organizations (cf. Meyer & Rowan, 1977).These pressures may lead to process manage-ment utilization, even in organizational or envi-ronmental contexts where it is ineffective oreven harmful.

Further, once adopted, the spread of processmanagement techniques beyond manufacturinggives rise to unintended effects on innovationand adaptation. Activities focused on measur-able efficiency and variance reduction drive outvariance-increasing activities and, thus, affectan organization's ability to innovate and adaptoutside of existing trajectories (e.g., Levinthal &March, 1993; Sutcliffe, Sitkin, & Browning, 2000;Weick, 1995). Core capabilities may become corerigidities or competency traps in rapidly chang-ing environments (Leonard-Barton, 1992;Levinthal, 1991; Levitt & March, 1988).

Thus, although process management activi-ties are employed to help organizations adapt,to the extent that these practices increase iner-tia and limit variation creation, they may in-stead impede adaptation in all but routine con-texts (e.g., Levinthal, 1997a). Yet there has beenno effort to build testable theory to explore thesebroader implications of process managementactivities for innovation and adaptation.

PROCESS MANAGEMENT ANDTECHNOLOGICAL INNOVATION

Technological innovation is a central engineof organizational adaptation (Brown & Eisen-

hardt, 1997; Levinthal, 1991; Nelson & Winter,1982; Tushman & Nelson, 1990). Thus, to under-stand how process management techniques af-fect organizational adaptation, we first addresshow they affect both exploratory and exploit-ative innovation. A firm's ability to innovate pro-vides its senior team with options either to rein-force or destabilize a technological regime(Burgelman, 1994; McGrath, 1999; Tushman &Murmann, 1998). Innovations generated withinor absorbed by firms provide variation to proac-tively shape or reactively respond to technolog-ical transitions (Brown & Eisenhardt, 1997;Cohen & Levinthal, 1990; Tushman & O'Reilly,1997). An organization's dynamic capabilitiesdepend on simultaneously exploiting currenttechnologies and resources to secure efficiencybenefits and creating variation through explor-atory innovation (Ghemawat & Costa, 1993;March, 1991: Teece et a l , 1997). As process man-agement techniques focus on continuous im-provement in routines and variation reduction(Anderson, Rungtusanatham, & Schroeder, 1994;Hackman & Wageman, 1995; Harry & Schroeder,2000), their increased utilization in an organiza-tion affects the balance between exploratoryand exploitative innovation (Benner & Tushman,2002).

Process management influences technologi-cal innovation in several ways. Process-focusedactivities stabilize the resource allocation anddecision processes that determine which tech-nological projects will be supported (cf. Chris-tensen & Bower, 1996). Process managementtechniques also tighten Internal communicationlinkages and affect the types of technologicalchanges that are recognized and addressed(Henderson & Clark, 1990). Programs like Designfor Six Sigma or ISO 9001 also ensure that pro-cess management activities extend to innova-tion design and development processes (e.g.,Harry & Schroeder, 2000). We argue that processmanagement activities will facilitate some in-novation types but dampen others. We first needto clarify innovation types.

Types of Innovations

We classify innovations along two dimen-sions: (1) their proximity to the current techno-logical trajectory and (2) their proximity to theexisting customer/market segment (Abernathy &Clark, 1985). On the technological dimension.


different innovation types have contrasting de-terminants and organizational effects (Morone,1993; Tushman & Smith, 2002). Incremental inno-vation, characterized by small changes in atechnological trajectory, builds on the firm's cur-rent technical capabilities, while radical inno-vation fundamentally changes the technologi-cal trajectory and associated organizationalcompetencies (Dosi, 1982; Green, Gavin, &Smith, 1995). Innovation can be further catego-rized by how it affects existing subsystemsand/or linking technologies (Baldwin & Clark,2000; Henderson & Clark, 1990; Tushman & Mur-mann, 1998). Modular innovations affect sub-system or component technology, leaving link-ing mechanisms intact, while architecturalinnovations involve changes in how subsystemsare linked together (Henderson & Clark, 1990;Iansiti & Clark, 1994). For example, the addition ofa stepper motor was a modular innovation in thephotolithography industry, while the move from14- to 8-inch disk drives was an architectural in-novation in the disk drive industry (Christensen &Bower, 1996; Henderson & Clark, 1990).

Innovations have also been classified bywhether they address the needs of existing cus-tomers or are designed for new or emergentmarkets (Christensen & Bower, 1996). Productsdesigned for new customer sets are often organ-izationally disruptive and require significantdepartures from existing activities. For exam-ple, the advent of digital photography not onlyrepresented a technological change from chem-ical-based film but also involved new distribu-tion channels for electronic cameras. For somephotography firms, the change in marketing anddistribution presented a challenge equal orgreater to the technological change (Tripsas &Gavetti, 2000). Similarly, progressively smallerdisk drives were innovations for emergent cus-tomer sets in the disk drive industry. These tech-nologically simple innovations created organi-zational challenges for incumbents, and theyfound it difficult to respond (Christensen, 1998).

Incremental technological innovations andinnovations designed to meet the needs of ex-isting customers are exploitative and buildupon existing organizational knowledge. Incontrast, radical innovations or those for emer-gent customers or markets are exploratory,since they require new knowledge or depar-tures from existing skills (Levinthal & March,1993; March, 1991).

Process Management's Effects on Innovation

In this section we develop propositions aboutthe relationship between process managementand innovation. Our propositions and the linksbetween them are outlined in Figure 1.

Because the implementation of process man-agement techniques initially entails an explicitfocus on innovation and change in organization-al activities, it is likely to spur innovation (Win-ter, 1994). Tacit organizational routines are re-vealed and made explicit (Brown & Duguid,1991), leading to richer cognitive models of or-ganizational activities (Repenning, 1999); firmscan learn before doing (Pisano, 1996). Team ap-proaches to problem solving, decentralized de-cision making, and colocated information facil-itate the search for solutions to help drive initialimprovements in efficiency in organizational ac-tivities (see also Hackman & Wageman, 1995,and Wruck & Jensen, 1994). In underperformingfirms the initial organizational changes associ-ated with process management may well beradical (e.g., Kearns & Nadler, 1992), or, in thecase of lethargic competitors, process manage-ment efforts can lead quickly to substantial per-formance gains (e.g., Kolesar, 1993).

While this innovative phase of process man-agement entails substantive change, suchchange is focused on improving efficiencywithin an existing technological trajectory. Thefounders of process management focused on in-cremental and exploitative innovation, ratherthan radical, architectural, or exploratory inno-vation. The associated process-oriented toolsand techniques, thus, are aimed at making anorganization more efficient through incrementalimprovements in processes and outputs (Ander-son et a l , 1994; Harry & Schroeder, 2000). Indeed,the prescriptive literature encourages organiza-tions to view improvements as controlled exper-iments that involve repetition of practices andmeasurement prior to making small, testablechanges (e.g., Harry & Schroeder, 2000). More-over, process improvement techniques apply tothe organization's current activities; innovationand change in those activities are accomplishedby paring from existing routines. Once changed,these improved routines become standardizedbest practices.

After this initial phase of innovation, repeti-tion of organizational routines through adher-ence to best practices leads to deepened com-

244 Academy oi Management Review April

"a

0)uu

0"os

1 T

ech

cal

.ogi

oau.-'oG01

*

du

u

iogi

0a(J.2Sc01

t-i

uc'cQz

DlCD

.do


petence (e.g., Pisano, 1996). As an organizationlearns and increases its efficiency through rep-etition of a set of activities, its subsequent inno-vation is increasingly incremental (Levinthal &March, 1993; Levitt & March, 1988). Thus, whileprocess management activities drive innova-tion, the focus on variation reduction, search forincremental improvements in routines, and in-creased proficiency through repetition of organ-izational activities ensures that this innovationwill be in the neighborhood of existing capabil-ities. As process management activities extendin an organization over time, more and tighterlinkages between routines further constrain in-novation to incremental changes in existing ac-tivities and outputs.

Proposition 1: Increases in processmanagement practices promote incre-mental innovation.

Process management activities also aim atimprovements in processes and outputs for theorganization's existing customers. Measures ofcustomer satisfaction are collected and used asthe basis of improvement and coordination ef-forts (Cole, 1998; Garvin, 1988; Hackman &Wageman, 1995; Powell, 1995), and such effortsare reinforced by supportive resource allocationsystems. As the locus of decision making ispushed down in the firm, multiple levels of thehierarchy are focused on meeting customerneeds (Winter, 1994). Von Hippel (1988) andChristensen and Bower (1996) show how persis-tent resource allocation processes among equip-ment suppliers and disk drive manufacturerschanneled innovation away from product devel-opment for emergent customer sets. Thus, asprocess management creates an organizationalfocus on better understanding and satisfyingexisting customers, it also channels innovationinto areas that benefit existing customers.

Proposition 2; Increases in processmanagement practices promote inno-vation for existing customer sets.

Henderson and Clark (1990) found that, in thephotolithography equipment industry, increas-ingly stable routines and communication link-ages within firms constrained their ability toinitiate and respond to changes in subsystemand linking technologies. Incumbent firmstreated seemingly minor architectural innova-tions as incremental innovations, with disas-

trous results. Process management techniquesstabilize organizational routines and tighten thelinkages between them, yet they make cross-boundary, cross-community linkages more diffi-cult (Sitkin & Stickel, 1996). Organizations fo-cused on incremental enhancements of currenttechnology treat architectural innovation asmerely incremental, fail to forge linkages acrossorganizational boundaries, and, in turn, under-perform (Henderson, 1993; Lawless & Anderson,1996; Tripsas, 1997). Although incremental inno-vation may, in some circumstances, actually ac-commodate architectural or modular innova-tions, adherence to standardized best practicesensures repetition of practices through thesestable linkages within local domains, and anorganization's ability to actually take advan-tage of subsystem and linking technologies isstunted.

Proposition 3; increases in processmanagement practices decrease ar-chitectural innovation.

Process management techniques, by designand intent, exploit existing capabilities. Whereshort-term performance pressures, the demandsof existing customers, and ease of measurementdominate, exploitation overwhelms exploration(Levinthal & March, 1993; March, 1991; Sitkin eta l , 1994). Process management affects variationcreation through the use of statistical tech-niques designed to reduce variation, and alsothrough repetition of linked sets of streamlinedorganizational processes. Rapid exploitation, asemployees carry out activities in coordinatedprocesses, produces measurable, short-termbenefits. As an organization achieves faster newproduct introductions, cost reductions, or im-provements in customer satisfaction, these out-comes drive the organization to persist withgreater zeal in extending process activities andlinking disparate parts of the firm (Levinthal &March, 1993). Managers with documented suc-cesses from process management may be pro-moted and charged with extending such activi-ties to more domains. Those opposed to theencroachment of process management in cre-ative or innovative domains may opt out of theorganization, further increasing the influence ofprocess-intensive activities (Brown & Duguid,2000).

As systems of improved routines are repeatedthroughout an organization, the organization be-


comes more competent and efficient, but var-iation in the outcomes of those activitiesdecreases (March, 1991). The objective of stabi-lizing organizational routines and "attackingvariation" (Harry & Schroeder, 2000) contrastswith the variation-increasing roles played byinternal corporate venturing units (Burgelman,1983), as well as with redundant and overlap-ping information designed to increase organiza-tional innovation and variety (Nonaka & Takeu-chi, 1995). This explicit focus on incrementalinnovation and increased proficiency with localsearch makes it unlikely that process manage-ment activities will produce innovations thatsignificantly depart from the neighborhood ofthe organization's existing technological or mar-ket competencies.

Further, over time, process management influ-ences variation creation by affecting the selec-tion of innovations. Short-term, easy-to-measureefficiency improvements make vague, uncer-tain, difficult-to-quantify exploratory activitiesless attractive (Levinthal & March, 1993; March,1991). As process management practices spreadin an organization, the predominant measuresof effectiveness are increasingly focused onspeed, efficiency, and reductions in costs orwaste. These dynamics lead to selecting inno-vations that leverage efficient, streamlinedmanufacturing or distribution processes or thatutilize materials that are cost-effectively ob-tained from streamlined purchasing processes.Such innovations build on existing firm capabil-ities and tend to be closer to existing products.

Because innovation development and selec-tion processes are linked with other improvedprocesses, the focus shifts away from creatingvariation. A focus on refining project selectionprocesses to yield continuous improvement inthe speed or success rates of new products tipsproject selection toward those with greater pre-dictability, lower variation, and increased cer-tainty. Thus, as the reach of process manage-ment activities extends further into research,R&D project selection activities, or product de-velopment, radical innovation projects increas-ingly give way to more certain, incremental ac-tivities (e.g., Henderson et a l , 1998). As processintensity increases, even structures designed toproduce radical innovations (e.g., heavyweightteams or independent units) increasingly willproduce innovations close to past innovations

(e.g.. Brown & Duguid, 2000; Sitkin & Stickel,1996; Tripsas & Gavetti, 2000).

Incremental innovation associated with pro-cess management reduces significant explor-atory activity and learning outside the existingtechnological trajectory (Levinthal & March,1993; March, 1991). The path-dependent nature ofinnovation suggests an even longer-lasting ef-fect of process management practices. Past in-novative activities play a role in future innova-tion by providing a firm with a knowledge basethat allows it to absorb technological compe-tence from external sources (Cohen & Levinthal,1990; Levitt & March, 1988). An organization thatlacks exploration in one period may be excludedfrom areas of future exploratory activity be-cause it has no relevant knowledge base (e.g.,Cohen & Levinthal, 1990; Teece et a l , 1997).Since process management techniques reduce afirm's exploratory activities, its absorptive ca-pacity will be stunted. Therefore, the firm is lesslikely to produce subsequent innovations thatincorporate new technologies.

Proposition 4: Increases in processmanagement practices decrease radi-cal innovation.

Measures of customer satisfaction drive im-provement efforts through an increasingly coor-dinated system of processes. As improved man-ufacturing or distribution processes createmeasurable improvements for existing custom-ers, process management will drive innovationsthat continuously improve products for existingcustomers. Emergent customer sets or marketsegments often do not lend themselves to mea-surement or are highly uncertain (Sitkin et a l ,1994; von Hippel, 1988). Thus, exploratory inno-vation into new markets becomes increasinglyunattractive compared to the short-term, tangi-ble, measurable successes from further improv-ing existing capabilities (Christensen, 1998;Leonard-Barton, 1992).

Through process management practices, anorganization becomes increasingly skilled atproducing outputs that leverage existing knowl-edge about inputs, technologies, manufacturingtechniques, or distribution channels. New inno-vations that further utilize these capabilitieswill benefit from these efficiencies and lendthemselves to even more measurable successes.As these codified and tightly linked processesbegin to further influence the development and

2003 Bennei and Tusbman 247

delivery of technological innovations, innova-tions for emergent or new markets will besqueezed out. This is likely to further exacerbatethe tendency to innovate for existing customersthat Christensen and Bower (1996) observed inthe disk drive industry.

Proposition 5: Increases in processmanagement practices decrease inno-vation for new customer sets.

Exploration, Exploitation, and AmbidextrousOrganizations

Abernathy (1978) highlights the inconsisten-cies between activities focused on productivityimprovements and cost reductions and those fo-cused on innovation and flexibility. He ques-tions whether it is possible for organizations topursue both types of activities simultaneously.Although both types of activities are importantfor organizational survival, exploration and ex-ploitation are contradictory organizational pro-cesses (Adler, Goldoftas, & Levine, 1999; March,1991; Teece et al., 1997). How can organizationsbalance these potentially conflicting activities?The organizational literature has stressed theimportance of a balance between efficiency andexploration, but there are multiple points ofview on how organizations actually strike thisbalance.

Hedberg, Nystrom, and Starbuck (1976) sug-gest that organizations engage in multipleforms of learning by alternating between differ-ent organizational designs and by being bothconsistent and inconsistent in their actions. Sim-ilarly, Brown and Eisenhardt (1998) suggest thatdynamic capabilities are rooted in an organiza-tion's ability to rhythmically switch betweenmore organic and more mechanistic structures.In contrast to alternating between organizingmodes, others argue for creating loosely coupledorganizations, where the experimenting unitsare highly buffered from the exploiting units(Levinthal, 1997b; Weick, 1976). For example,Burgelman (1991) describes buffered-inducedand autonomous processes at Intel, whereasLeonard-Barton (1995) describes experimentingunits distinct from units focused on more incre-mental innovation. At the extreme, Christensen(1998) suggests that because of the disruptivenature of the technology, experimenting unitsmust be completely separated from exploiting

units. In these loosely coupled organizationalforms it is unclear where the integration requiredto drive streams of innovation is accomplished.

Ambidextrous or dual organizational formsare organizational architectures that build inboth tight and loose coupling simultaneously(Bradach, 1997; Sutcliffe et a l , 2000; Tushman &O'Reilly, 1997). These organizational forms arenot loosely coupled, nor do they switch betweencontrasting structures. Rather, ambidextrous or-ganizations are composed of multiple tightlycoupled subunits that are themselves looselycoupled with each other. Within subunits thetasks, culture, individuals, and organizationalarrangements are consistent, but across sub-units tasks and cultures are inconsistent andloosely coupled. Strategic integration—the abil-ity to drive innovation streams and take advan-tage of contrasting organizational capabili-ties—-occurs at the senior team level of analysis.

In contrast to Lawrence and Lorsch's (1967)design, ambidextrous organizational designsare composed of highly differentiated butweakly integrated subunits. While the explor-atory units are small and decentralized, withloose cultures and processes, the exploitationunits are larger and more centralized, with tightcultures and processes. Exploratory units suc-ceed by experimenting—by frequently creatingsmall wins and losses (Sitkin, 1992). Becauseprocess management tends to drive out experi-mentation, it must be prevented from migratinginto exploratory units and processes. In contrast,exploitation units that succeed by reducing vari-ability and maximizing efficiency and controlare an ideal location for the tight coordinationassociated with process management efforts.

These contrasting, inconsistent units must bephysically and culturally separated from oneanother, have different measurement and incen-tives, and have distinct managerial teams (Bra-dach, 1997; Nonaka, 1988, 1991; Sutcliffe et a l ,2000; Tushman & O'Reilly, 1997). For example, inHewlett Packard's (HP's) Scanner Division, themore routine flatbed scanners had a completelydifferent organizational architecture from theemerging consumer/knitting technology scan-ners. These distinct units were physically sepa-rated from one another and had their own man-agement teams.

Because dynamic capabilities are rooted indriving streams of innovation in a product class(Morone, 1993; Teece et al., 1997), these highly


differentiated but loosely coupled subsystemsmust be strategically integrated by the seniorteam. Such strategic linkage is anchored bycommon aspiration levels and a senior teamthat both provides slack to the experimentalsubunits and holds the differentiated units tofundamentally different selection and searchconstraints (Levinthal & March, 1993; Rotemberg& Saloner, 2000). To be effective, ambidextroussenior teams must develop processes for estab-lishing new, forward-looking cognitive modelsfor exploration units, while allowing backward-looking experiential learning to rapidly unfoldfor exploitation units (Gavetti & Lsvinthal, 2000;Louis & Sutton, 1989).

To create dual organizational structures, se-nior teams must develop techniques that permitthem to be consistently inconsistent as theysteer a balance between the need to be smalland large, centralized and decentralized, andfocused both on the short term and long term,simultaneously (Gavetti & Levinthal, 2000; Hed-berg et al., 1976; Weick, 1995). If the locus ofstrategic integration is low in the firm, experi-mentation will be stunted, as the short-term suc-cesses of rapid exploitation will drive out explo-ration. If this integration is at too high a level ina multidivisional firm, or done across indepen-dent firms (e.g., Christensen, 1998), the underly-ing understanding of an innovation stream's dy-namic will be dampened, and the ability to drivedisruptive or radical change restricted.

For example, at HP, when the Scanner Divi-sion's general manager and his senior team pro-vided the integration between the unit's twohighly differentiated subunits, the division wasable to effectively both explore and exploit.However, when this general manager was pro-moted, the consumer unit was spun out to asector executive. Because this corporate execu-tive did not deeply understand the competitiveissues in the scanning industry, he cancelledthe more exploratory consumer unit when it didnot show quick profits.

While complex and politically difficult, ambi-dextrous organizational forms permit a firmwith highly differentiated units to drive processmanagement, with its associated variation re-duction and control, as well as exploration andoption creation. Experimental units providevariation from which the senior team can learnabout and, in turn, bet on the future, even as theexploitation units build capabilities for short-

term effectiveness (McGrath, 1999). These inter-nally inconsistent operating modes must bestrategically linked by the senior team throughtheir aspirations and actions and through a lim-ited set of core values (Hambrick, Nadler, &Tushman, 1998).

Thus, the impact of process management ac-tivities on innovation outcomes will be contin-gent on organizational context. In ambidextrousorganizational forms, process management ac-tivities drive exploitative innovation, even asthey are buffered from intruding on exploratoryinnovation.

Proposition 6: In the context of an ambi-dextrous organizational form, increasesin process management practices in-crease exploitative innovation but donot dampen exploratory innovation.

PROCESS MANAGEMENT ANDORGANIZATIONAL ADAPTATION

As technological capabilities affect organiza-tional fate, so, too, does process management'sinfluence on innovation have implications fororganizational adaptation. Process manage-ment activities, through their impact on organi-zational processes, accentuate organizationalinertia and slow responsiveness to technologi-cal transitions. We now explore process man-agement's effects on organizational adaptationin contrasting technological contexts.

Technology Cycles

Organizational environments are character-ized by cycles of technological variation, alter-nating between periods of incremental changeand periods of rapid innovation (Abernathy &Utterback, 1978; Sanderson & Uzumeri, 1995;Tushman & Anderson, 1986). Organizations out-side an industry often introduce discontinuoustechnological advances, and this triggers peri-ods of rapid innovation and change (Foster,1986; SuU, Tedlow, & Rosenbloom, 1997). Thisperiod of rapid innovation, or era of ferment,ends when a dominant technology designemerges (Abernathy, 1978; Anderson & Tush-man, 1990; Tushman & Rosenkopf, 1992). Withthe emergence of a dominant technological re-gime, the nature of technical change shifts fromproduct innovation to a relatively long period of


process innovation and incremental refine-ments of the selected technology (Abernathy &Utterback, 1978; Ettlie & Reza, 1992; Utterback,1994). This period of incremental technicalchange, in turn, is punctuated by the subse-quent technological discontinuity.

Thus, in the videocassette recorder (VCR) mar-ket, after a period of variation between alterna-tive formats, VHS became the industry standard.Once this standard emerged, subsequent inno-vation focused on incremental improvementsand cost reductions in VHS technology (Cu-sumano, Mylonadis, & Rosenbloom, 1992). Simi-lar innovation dynamics have been found in arange of industries, including typewriters, tele-visions, computers, disk drives, operating sys-tems, browsers, and microprocessors (see Tush-man & Murmann, 1998, and Van de Ven, Angle, &Poole, 1989).

Adaptation in Stable Environments

Because of the shifting nature of innovationrequirements embedded in technology cycles,organizations must develop capabilities tomove with, if not shape, these cycles (Hender-son, 1993; Tushman & O'Reilly, 1997). Duringeras of incremental change, organizations thatsustain incremental innovation will be more ef-fective than those that initiate variance-increas-ing innovation. As process management activi-ties stimulate incremental innovation, theseactivities will benefit organizations, when tech-nological environments are characterized byincremental refinements of an existing techno-logical design. Greater consistency and effi-ciency will be beneficial in periods of incremen-tal change. Rapid improvements in customersatisfaction, combined with reduced waste andcosts, will likely further enhance organizationaleffectiveness.

Organizations with routines and proceduresstabilized through process management activi-ties, thus, are likely to do well in stable or pre-dictable contexts (Donaldson, 1995; Hannan &Freeman, 1984). Indeed, Ittner and Larcker (1997)found that process management was positivelyassociated with performance in the stable autoindustry, but not in the more dynamic computerindustry. These environmental contingenciesare also supported by Sitkin et a l (1994) andSterman et al. (1997).

Proposition 7: Increases in processmanagement practices will improveperformance in eras of incrementaltechnological change.

Tighter coordination and repetition of activi-ties embedded in standardized best practicesincrease an organization's speed and efficiency.For example, streamlined linkages between pro-cesses for innovation and manufacturing allowfor rapidly screening and developing innova-tions that best leverage downstream manufac-turing or distribution capabilities. Further, asprocesses for identifying and addressing prob-lems and opportunities in the environment arefurther refined and routinized by process man-agement's influence, decision making and prob-lem solving also become more efficient (Miller,1993). Increasing organizational proficiency inrecognizing and addressing recurring chal-lenges leads to stable and increasingly efficientcommunication channels and information filters(Henderson & Clark, 1990; Tyre & Orlikowski,1994). These stable patterns of communicationand interaction lead to the development ofnorms, rules, and roles, which further channelindividual and group behavior into streamlinedactivities that carry out an organization's mis-sion more efficiently (Hackman, 1992; Nadler &Tushman, 1998; Repenning, 1999). Over time, sta-ble procedures and norms also drive increaseddemographic homogeneity within the organiza-tion, further speeding decision making andproblem solving (Keck & Tushman, 1993; Wil-liams & O'Reilly, 1998). These factors drive anincreasingly tightly integrated organization fo-cused on fast response to existing customerrequirements.

Proposition 8: Increases in processmanagement practices will speed or-ganizational responsiveness duringeras of incremental technologicalchange.

Adaptation in Turbulent Environments

Conversely, the ability to develop new tech-nological capabilities rapidly is especially crit-ical in environments characterized by rapid in-novation and change (Brown & Eisenhardt, 1997;Teece et a l , 1997; Tushman & Anderson, 1986).Periods of technological ferment are character-ized by substantial product variation and mar-


ket uncertainty. Responding to environmentaluncertainty and variation requires similar vari-ety within the firm (Daft & Weick, 1984). A reduc-tion in variance-increasing activity within thefirm prevents it from registering and/or respond-ing to environmental uncertainty (Burgelman,1994). Process management's variation-reducingfocus restricts the development of alternativesfor responding to environmental changes. Forexample, Sitkin and Stickel (1996) found thatTQM efforts in R&D settings not only bred dis-trust and dissension across the firms but alsodrove out variability in R&D.

Organizational outcomes are affected by de-layed or inadequate responses to environmentalturbulence (Henderson, 1993; Meyer, Brooks, &Goes, 1990). For example, IBM's relatively slowresponse to personal computers resulted in thesuccessful entry of other less inertial competi-tors that fundamentally changed the nature ofthe computer industry (Mitchell, 1989). Slow orincompetent responses to environmental shiftsprevented incumbents from retaining their lead-ership positions in the disk drive (Christensen &Bower, 1996), photolithographic equipment (Hen-derson & Clark, 1990), and watch (Glasmeier,1991) industries, among others (Tushman &O'Reilly, 1997). Similarly, TQM practices havebeen associated with lower performance in thedynamic computer industry environment (Ittner& Larcker, 1997; Sitkin et a l , 1994).

Proposition 9: Increases in processmanagement practices will decreaseperformance in eras of technologicalferment.

The same practices that help an organizationlearn and achieve efficiency more quickly canalso impede an organization's adaptation to ma-jor technological transitions (Levinthal, 1991,1997a). Over time, as process management ac-tivities permeate an organization, the increas-ingly stable, tightly linked, and efficient rou-tines that span an organization make anythingmore than incremental organizational changedifficult (Hannan & Freeman, 1984). Organiza-tions that have honed their skills in making in-cremental changes in processes and productsdevelop momentum that works to impede majorchange (Miller & Friesen, 1980) or that trans-forms core competencies into rigidities (Leo-nard-Barton, 1992).

Case studies of process management effortsand simulation studies of problem-solving dy-namics provide evidence of this tendency forincreased resistance to change resulting fromprocess management activities. Following suc-cessful implementation of a process improve-ment program. Analog Devices' financial perfor-mance fell far below the industry average. Thecompany subsequently underwent a major reor-ganization, including a reorientation of its prod-uct development efforts away from its core busi-ness and into emerging markets (Sterman et a l ,1997: 505). While process improvement effortswere underway, internally generated innova-tions necessary for successful performance in achanged environment were not executed.

Extending the Analog Devices study, Repen-ning (1999) argues that, over time, process im-provement activities trigger cycles of increasedmanagerial control and bureaucratic proce-dures, which undermine initial gains. Similarly,Alcoa CEO's frustration with the slow pace ofcontinuous improvement might also be ex-plained by the increased inertia associated withpractices focused on incremental change andimprovement (Kolesar, 1993). Finally, Levinthal's(1997b) modeling of adaptation dynamics showsthat tightly coupled organizations are subjectedto heightened inertia and high failure rates inchanging environments. Because organizationalinertia is so strong, successful firms movethrough environmental change by initiating re-orientations (Tushman & Romanelli, 1985).

Thus, the influence of process managementtechniques on integrating and coordinating pro-cesses can drive rapid exploitation and effi-ciency, but also longer-term momentum andresistance to change. Although tightly coordi-nated and streamlined processes in product de-velopment and manufacturing may allow forrapid response with incremental extensions orenhancements of current capabilities, the asso-ciated inertia is likely to make such an or-ganization slower to respond to subsystem, link-ing, or radical technological change. Indeed,empirical results suggest that, in the digitalcamera industry, more extensive process man-agement activities dampened a firm's ability tokeep up with rapid technological changethrough new product introductions (Benner,2002). These results were particularly strong forphotography industry incumbents entering thenew digital camera industry, suggesting that


the incremental improvements in routines thatenhanced their efficiency in film technologymay have been inappropriate for responding tothe rapid change in digital technology.

Proposition 10: Increases in processmanagement piactices will slow or-ganizational responsiveness in eras oftechnological ferment.

Moderating Effects of AmbidextrousOrganizational Forms

To maximize short-term performance and sur-vive in periods of incremental technologychange, firms need to accentuate incrementalchange, momentum, and inertia associated withprocess activities. Multiple functions and activ-ities must be linked seamlessly throughout theorganization to efficiently deliver to and satisfyexisting customers. Without this concerted re-finement of capabilities, firms may not survivelong enough to face or initiate technologicalchange.

But process intensity also stunts incumbents'ability to take advantage of internally gener-ated opportunities for discontinuous change(Cooper & Smith, 1992; Foster, 1986). For exam-ple, even though Swiss producers invented thequartz movement, it was American and Japa-nese firms that actually introduced this techno-logical discontinuity (Glasmeier, 1991). Thus,process management promotes only one dimen-sion of a firm's dynamic capabilities.

As firms achieve short-term measurable suc-cess with process management, they are likelyto increase their commitment to process inten-sity and broaden its influence to ever more pro-cesses. Yet firms must be capable of forgettingtheir past, breaking rules and traditions, andincreasing variation in the service of architec-tural and/or radical innovation or in meeting theneeds of new customer segments (Hedberg etal., 1976; Weick, 1995). Because competencies arehard to develop and the rates of environmentalchange are substantial, we argue that dynamiccapabilities are not rooted in sequential atten-tion or rhythmic pacing (e.g.. Brown & Eisen-hardt, 1998) but, rather, in exploiting and explor-ing simultaneously (Sutcliffe et al,, 2000;Tushman & O'Reilly, 1997),

Over time, process management crowds outexploratory innovation. A decrease in explor-

atory behaviors stunts an organization's abilityto adapt in environments characterized by tech-nological ferment and uncertainty. Ambidex-trous organizational forms isolate process man-agement activities in subunits where reducingvariation and increasing control are strategi-cally vital. An ambidextrous organizational de-sign allows for uncoupling the variance-decreasing units and activities from those unitswhere variation is critical.

For example, at CibaVision, a single seniorteam created highly differentiated subunits,where one subunit pursued low-cost lenses evenas another subunit developed a substitute (dis-posable) lens. Similarly, at Ciba's Crop Protec-tion Division, a single general manager builtmultiple internally inconsistent subunits in or-der to drive an innovation stream to keep plantshealthy. One unit based in Switzerland focusedon the efficient development of chemical-basedfungicides. Anchored in molecular biology, an-other unit, based in the United States, focusedon developing a seed that would grow plantsnot needing fungicides. If this latter unit weresuccessful, it would cannibalize the chemical-based unit in Switzerland (Tushman 8E O'Reilly,1997).

These experimental, variance-creating unitsprovide the options from which a firm's seniormanagers can select in order to shape an evolv-ing innovation stream, and they must be buff-ered from the enticing spread of efficiency im-provement in the existing technology. Theseideas further suggest that organizational formmoderates the impact of process management.

Proposition 11: In the context of an am-bidextious organizational form, in-creases in process management prac-tices will enhance responsiveness andperformance during eras of incremen-tal technological change hut willhave no effect on responsiveness orperformance during eras of technolog-ical ferment.

DISCUSSION

In the twenty-five years since Abernathy's ad-monition about the productivity dilemma, therehas been an ever-increasing influence of theIdeology of process management on organiza-tions. Over the past two and a half decades.


firms have had substantial institutional pres-sures, irrespective of their organizational or en-vironmental contexts, to adopt TQM, Six Sigma,or BPR; to get ISO 9000 certified; or to compete forthe Baldrige or Deming awards. While the sym-bolic importance of process management maybe important, its substantive benefits are muchless clear.

Our review suggests that inconsistent out-comes of process management practices can bereconciled with attention to the context in whichthese practices are employed. Process manage-ment activities are positively associated withorganizational effectiveness in a limited set ofconditions: during periods of stability or incre-mental change and for incremental innovationor existing customers. In contrast, in a muchwider set of conditions—during eras of ferment;in turbulent environments; for new customersegments; and for architectural, modular, andradical innovation—process management activ-ities are less conducive to organizational effec-tiveness. Under these frequently occurring con-ditions, process management activities buildresistance to change and momentum and, inturn, inhibit organizational variability. These in-ertial outcomes of process management activi-ties stunt a firm's ability to adapt.

But organizations must outcompete rivalsboth in the short and long run. Organizationsthat must meet current customer requirementsand new customer demands do not have theluxury of choice—they must deal simulta-neously with the inconsistent demands of ex-ploitation and exploration. Process manage-ment capabilit ies speed exploitation andefficiency, and while they may allow organiza-tions to survive in the short run, they simulta-neously dampen the exploration required forlonger-term adaptation. Ambidextrous organiza-tional forms reconcile these paradoxical de-mands by building internally inconsistent archi-tectures within a single organization—contrasting architectures that retain the benefitsof experimentation and variability, along withthe benefits of exploitation and process control.These tightly coupled, internally inconsistentarchitectures must be tactically uncoupled.However, to drive streams of innovation, theseinconsistent units must be strategically inte-grated by the senior team. It may be that heter-ogeneous senior team capabilities coupled with

complex organizational architectures are at theroot of dynamic organizational capabilities.

Implications for Future Research

Our model provides a starting point for iutureresearch on how process management practicesaffect firm dynamic capabilities. We advancetestable propositions about the effects of theseinstitutionally mandated practices on both tech-nological innovation and adaptation. Moreover,our propositions provide a basis for understand-ing how the effects of process managementpractices unfold over time in an organization.Future empirical work to test and modify thesepropositions must incorporate a longitudinalperspective and assess how increases in suchactivity over time affect an organization, ratherthan compare differences associated with nom-inal adoption of quality programs across firms.

While our ideas may be relevant for all typesand sizes of firms, they apply most readily tofirms whose strategies include both exploitativeand exploratory innovation. Our discussion con-cerns the effects of increased process manage-ment intensity for those firms already chal-lenged with balancing such activities. Ourpropositions are therefore less relevant for firmswhose strategies focus solely either on exploi-tation or exploratory innovation. In addition,these ideas may not apply to small start-ups intheir initial phases—not yet challenged withbalancing exploitation and improvements of aninitial project with the exploratory developmentof subsequent products. Future research shouldfurther test the boundaries of our propositions.

Our review also suggests that, to understandthe relations between process activities and or-ganizational outcomes more fully, we need tomore explicitly couple research and theory inprocess management, organizations, and strat-egy. There is much to be gained by integratingthese perspectives on the phenomena and bytaking a contingency approach to process activ-ities and organizational outcomes. Technologi-cal, environmental, and structural conditionsclearly moderate the relations between processactivities and organizational outcomes. If dy-namic capabilities require exploitation as wellas exploration, more complex organizationalforms are required, which, in turn, demand morecomplex senior team capabilities. This review,then, indicates the potential in exploring more


deeply the relationships among organizationalarchitectures, senior team behaviors, innovationstreams, and process management.

Implications for Managerial Practice

Finally, our review has implications for prac-tice. The powerful institutional pressures toadopt process management practices have cutacross industries and firms, regardless of age orsize. Despite the coercive pressure and promiseof legitimacy, managers need to exercise greatcare in when and where to adopt these prac-tices. Whereas in stable, technologically certainsettings these practices may be productive, inuncertain or technologically complex contextsthese practices may be quite counterproductive.

Indeed, the usefulness of process manage-ment practices may be much more constrainedthan the popular literature suggests. Processmanagement and its associated technologiesand philosophies are conservative and resistantto anything but incremental or competence-enhancing innovation. This variance-hostile fo-cus on incremental change and existing custom-ers, from the senior team to lower levels of thefirm, severely stunts a firm's dynamic capabili-ties. It is the promise of ambidextrous organiza-tional forms and heterogeneous senior teamsthat provides the possibility of building organi-zations capable of both celebrating process ac-tivities as well as limiting their damage.

REFERENCES

Abernathy, W. J. 1978. The producfivify dilemma. Baltimore:Johns Hopkins University Press.

Abernathy, W. J., & Clark, K. 1985. Mapping the winds ofcreative destruction. Research Policy, 14: 3-22.

Abernathy. W. J., & Utterback. J. M. 1978. Patterns of indus-trial innovatian. Technology Review, 80(7): 40-47.

Abrahamson, E. 1996. Management fashion. Academy ofManagement Review. 21: 254-285.

Abrahamson, E., & Rosenkopi, L. 1993. Institutional and com-petitive bandwagons: Using mathematical modeling asa tool to explore innovation diffusion. Academy of Man-agement Review, 18: 487-517.

Academy of Management fleview. 1994, 19(3).

Adler, P. S. 1993. Time-and-motion regained. Harvard Busi-ness Review, 71(1): 97-108.

Adler, P. S., Goldoitas. B,, & Levine, D. 1999. Flexibility versusefliciency? A case study of model changeovers in the

Toyota production system. Orgonizafion Science, 10:43-68.

Anderson. J. C, Rungtusanatham, M., & Schroeder, R. G. 1994.A theory of quality management underlying the Demingmanagement method. Academy of Management Re-view, 19: 473-509.

Anderson. P., & Tushman. M. L. 1990. Technological discon-tinuities and dominant designs: A cyclical model oitechnological change. Administrative Science Quar-terly, 35: 604-633.

Baldwin, C. & Clark, K. 2000. Design ruJes: The power ofmodularity. Working paper. Harvard Business School.Boston.

Benner, M. 2002. Dynamic or static capabilities? Process man-agement and adaptation io technological change. Work-ing paper. The Wharton School, University of Pennsyl-vania, Philadelphia.

Benner, M., 8E Tushman, M. 2002. Process management andtechnological innovation: A longitudinal study of thephotography and paint industries. Working paper. TheWharton School, University of Permsylvania, Philadelphia.

Bradach, J. 1997. Using the plural form in the management ofrestaurant chains. Administrative Science Quarterly, 42:276-303.

, J. S.. & Duguid, P. 1991. Organizational learning andcommunities of practice. Organization Science, 2: 40-47.

Brown, J. S.. & Duguid, P. 2000. The sociaJ life of information.Boston: Harvard Business School Press.

Brown, S, L., & Eisenhardt, K. M. 1997. The art of continuouschange: Linking complexity theory and time-paced evo-lution in relentlessly shifting organizations. Administra-tive Science Ouarteriy, 42: 1-34.

Brown, S. L.. & Eisenhardt, K. M. 1998. Competing on fhe edge:Strategy as structured chaos. Boston: Harvard BusinessSchool Press.

Burgelman, R. 1983. A process model of internal corporateventuring in a diversified major firm. AdministrativeScience Quarfer/y, 28: 223-244.

Burgelman, R. 1991. Intra-organizational ecology of strategymaking and organizational adaptation. OrganizationScience, 2: 239-282.

Burgelman. R. 1994, Fading memories: A process theory ofstrategic business exit. Administrative Science Quar-terly, 39: 24-56.

Cameron, K., & Barnett, C. 2000, Organizational quality as acultural variable: An empirical investigation of qualityculture, processes, and outcomes. In R. E. Cole & W, R.Scott (Eds), The quality movement & organization the-ory; 271-294. Thousand Oaks, CA; Sage.

Christensen. C. M. 1998. The innovator's dilemma: When newtechnologies cause great firms to fail. Boston: HarvardBusiness School Press.

Christensen, C. M., & Bower. J. L. 1996. Customer power,strategic investment, and the failure of leading firms.Strategic Management Journal, 17: 197-218.

Cohen. W. M., & Levinthal, D. A. 1990. Absorptive capacity: A


new perspective on learning and innovation. Adminis-trative Science Quarterly, 35: 128-152.

Cole, R. E. 1998. Learning from Ihe quality movement: Whatdid and didn't happen and why? CalHornia Manage-ment Review. 41(1): 43-73.

Cole, R. E., & Scott, W. R. (Eds.). 2000. The quality movement& organization theory. Thousand Oaks, CA: Sage.

Cooper, A., & Smith, C. 1992. How established firms respondto threatening technologies. Academy oi ManagementExecutive, 6(2): 55-70.

Cusumano, M., Mylonadis, Y., & Rosenbloom, R. 1992. Stra-tegic maneuvering and mass market dynamics: The tri-umph oi VHS over Beta. Business History Review, 66:51-95.

Daft, R. L., & Weick, K. E. 1984. Toward a model of organiza-tions as interpretation systems. Academy oi Manage-ment Review, 9: 284-295.

Dean, J. W., Jr., & Bowen, D. E. 1994. Management theory andtotal quality: Improving research and practice throughtheory development. Academy o/ Management Review.19: 392-418.

Dean, J. W., Jr., & Snell, S. !996. The strategic use of inte-grated manufacturing: An empirical example. S(ra(egicManagement Journai. 17: 459-480.

Deming, E. W. 1986. Out oi crisis. Cambridge, MA: MIT Press.

Donaldson, L. 1995. American anti-management theories oimanagement. Cambridge: Cambridge University Press.

Dosi, G. 1982. Technological paradigms and technologicaltrajectories. Research Policy, 11: 147-162.

Easton, G. S., & Jarreli, S. L. 1998. The effects of total qualitymanagement on corporate performance: An empiricalinvestigation. Journal oi Business, 71: 253-307.

Eisenhardt, K.. & Martin, J. 2000. Dynamic capabilities: Whatare they? Sfrafegic Management/ournal, 21: 1105-1121.

Ettlie, J. E., & Reza, E. M. 1992. Organizational integration andprocess innovation. Academy oi Management Journal.35: 795-827.

Feyder, S. 2001. 3M to use favorite GE quality program; SixSigma adoption to be companywride. MinneapoJis StarTribune, January 25: ID.

Foster, R. 1986. Innovation: The attacker's advantage. NewYork: Summit Books.

Gabor, A. 2001. Quality revival, Part 2: Ford embraces SixSigma. New York Times, June 13: C5.

Garvin, D. A. 1988. Managing quality. New York: Free Press.

Garvin, D. A. 1991. How the Baldrige Award really works.Harvard Business fleview, 69(6): 80-93.

Garvin, D. A. 1995. Leveraging processes for strategic advan-tage. Harvard Business Review, 73(5): 77-90.

Garvin, D. A. 1998. The processes of organization and man-agement. Sloan Management Review. 39(4): 33-50.

Gavetti, G., & Levinthal, D. 2000. Looking forward and look-ing backward: Cognitive and experiential search. Ad-ministrative Science Quarterly, 45: 113-137.

Ghemawat, P., & Costa, J. 1993. The organizational tensionbetween static and dynamic efficiency. Strategic Man-agement/ourna/, 14: 59-73.

Ghoshal, S., & Bartlett, C. A. 1995. Changing the role of topmanagement: Beyond structure to processes. HarvardBusiness Review, 73(1): 86-96.

Glasmeier, A. 1991. Technological discontinuities and flexi-ble production networks: The case of Switzerland andthe world watch industry. Research Policy, 20: 469-485.

Green, S., Gavin, M., & Smith, L. 1995. Assessing a multidi-mensional measure of radical innovation. IEEE Transac-tions on Engineering Management, 42(3): 203-214.

Hackman, J. R. 1992. Group influences on individuals. InM. D. Dunnette (Ed), Handbook o/ industrial and organ-izational psychology: 1455-1525. Palo Alto, CA: Consult-ing Psychologists Press.

Hackman, J. R., & Wageman, R. 1995. Total quality manage-ment: Empirical, conceptual, and practical issues. Ad-ministrative Science Quarterly, 40: 309-342.

Hambrick, D., Nadler, D., & Tushman, M. 1998. Navigatingchange: How CEOs, top teams and boards steer transfor-mation. Boston: Harvard Business School Press.

Hammer, M., 8f Champy, J. 1993. Reengineering the corpora-tion: A manifesto /or business revolution. New York:Harper Business.

Hammer, M., & Stanton, S. 1999. How process enterprisesreally work. Harvard Business Review, 77(6): 108-118.

Hannan, M. T., & Freeman, J. 1984. Structural inertia andorganizational change. American Sociological Review,43: 149-164.

Harrington, H. ]., & Mathers, D. D. 1997. ISO 9000 and beyond:From compliance to periormance improvement. NewYork: McGraw-Hill.

Harry, M. J., 8f Schroeder, R. 2000. Six Sigma: The break-through management strategy revolutionizing theworld's top corporations. New York: Currency.

Hayes, R. H., & Abernathy, W. J. 1980. Managing our way toeconomic decline. In M. L. Tushman & W. L. Moore (Eds.),Readings in the management oi innovation: 11-25.Marshfieid, MA: Pitman.

Hedberg, B., Nystrom, P., & Starbuck, W. 1976. Camping onseasaws: Prescriptions for self-designing organizations.Administrative Science Quarter//, 21: 41-65.

Henderson, R. 1993. Underinvestment and incompetence asresponses to radical innovation. Hand Journal oi Eco-nomics. 24: 248-269.

Henderson, R. M., & Clark, K. B. 1990. Architectural innova-tion: The reconfiguration of existing product technolo-gies and the failure of established firms. AdministrativeScience Quarterly, 35: 9-30.

Henderson, R., De! Alamo, J., Becker, T., Lawton, J., Moran, P.,& Shapiro, S. 1998. The perils of excellence: Barriers toeffective process improvement in product-driven firms.Production and Operations Quarterly. 7(1): 2-18.

Hendricks, K. B., & Singhal, V. R. 1996. Quality awards and


the market value oi the firm: An empirical investigation.Managemenf Science, 42: 415-436.

Hill, R. 1993. When the going gets tough: A Baidrige Awardwinner on the line. The Executive, 7(3): 75-79.

Honeywell. 1998. Annual report. Minneapolis: Honeywell Inc.

Iansiti, M., & Clark, K. 1994. Integration and dynamic capa-bility. Industrial and Corporate Change, 3: 557-606.

International Organization ior Standardization (ISO). 1999.ISO 9000 "one two three." http://www.isoxh/iso/en/iso9000-14000/tour/123.html, accessed December 12, 2002.

Ishikawa, K. 1985. What is total quality control? The Japaneseway. Englewood Cliifs, NJ: Prentice-Hall.

Ittner, C. D., & Larcker, D. 1997. The performance effects ofprocess management techniques. Management Science,43: 522-534.

Juran, J. 1989. Juran on leadership /or quality. New York: FreePress.

Kearns, D., & Nadler, D. 1992. Prophets in (he darJc. New York:Harper.

Keck, S., & Tushman, M. 1993. Environmental and organiza-tional context and executive team structure. Academy ofManagement Journal, 36: 1314-1344.

Kolesar, P. I. 1993. Vision values milestones: Paul O'Neillstarts total quality at Alcoa. CalHomia ManagementReview. 35(3): 133-165.

Lawless, M. W.. & Anderson, P. C. 1996. Generational tech-nological change: Effects of innovation and local rivalryon performance. Academy oi Management Journal. 39:1185-1217.

Lawrence, P., & Lorsch, J, 1967, Organizations and environ-ments. Boston: Harvard Business School Press.

Leonard-Barton, D. 1992. Core capabilities and core rigidi-ties: A paradox in managing new product development.Strategic Managemenf/ournal, 13: 111-125.

Leonard-Barton, D, 1995. Wei/springs o/Jcnowiedge. Boston:Harvard Business School Press.

Levinthal, D. 1991, Organizational adaptation and environ-mental selection—interrelated processes of change. Or-ganization Science: 140-145,

Levinthal, D. 1997a, Three faces of organizational learning:Wisdom, inertia and discovery. In R, Garud, P, Nayyar, &Z. Shapira (Eds,), recfino/ogicai innovation: Oversightsand foresights: 167-180, Cambridge: Cambridge Univer-sity Press.

Levinthal, D. 1997b. Adaptation on rugged landscapes. Man-agement Science, 43: 934-950.

Levinthal, D,, & March, I. G. 1993. The myopia of learning.Strategic Management Journal, 14: 95-112.

Levitt, B,, & March, ]. G. 1988, Organizational learning. An-nual Review of Sociology, 14: 319-340,

Louis, M., & Sutton, R, 1989, Switching cognitive gears: Fromhabits of mind to active thinking. Human Heiations, 44:55-75.

March, J, 1991. Exploration and exploitation in organization-al learning. Organization Science, 2: 71-87.

McGrath, R. 1999. A real options logic for initiating technol-ogy positioning investments. Academy of ManagementReview, 22: 974-996.

Meyer, A., Brooks, G,, & Goes, ], 1990. Environmental jolts andindustry revolutions. Strategic Management/ournai, 11:93-110,

Meyer, J, W,, & Rowan, B, 1977, Institutionalized organiza-tions: Formal structure as myth and ceremony, Ameri-can Journal ol Sociology, 83: 364-385.

Miller, D, 1993, The architecture of simplicity. Academy ofManagement Review. 18: 116-138.

Miller, D., & Friesen, P. H, 1980. Momentum and revolution inorganizational adaptation. Academy of ManagementJournal, 23: 591-614,

Mitchell, W, 1989, Whether and when? Probability and tim-ing of incumbents' entry into emerging industrial sub-fields. Administrative Science Quarterly. 34: 208-234,

Morone, J. 1993. Winning in high tech markets. Boston: Har-vard Business School Press,

Mukherjee, A,, Lapre, M,, 8f Van Wassenhove, L, 1998. Knowl-edge driven quality improvement. Management Sci-ence, 44{Supplement}: S35-S49.

Nadler, D. A., & Tushman, M. L. 1998. Competing by design.New York: Oxford University Press,

Nelson, R,, & Winter, S, 1982. An evoiutionaiy theory of eco-nomic change. Cambridge, MA: Harvard UniversityPress,

Nohria, N. 1996, From the M form to the N form. WorkingPaper, Harvard Business School, Boston,

Nonaka, I., 1988, Toward middle-up-down management.SJoan Management fleview, 29(3): 9-18,

Nonaka, I, 1991, The knowledge-creating company. HarvardBusiness Review, 69(6): 96-104,

Nonaka, 1,, & Takeuchi, H, 1995. The knowledge-creatingcompany. New York: Oxford University Press.

Pisano, G, 1996. Learning before doing in the development ofnew process technology. Research Policy, 25: 1097-1119,

Powell, T, C. 1995, Total quality management as competitiveadvantage: A review and empirical study. StrategicManagement Journal, 16: 15-37.

Quality Digest. 1999. Database of ISO 9000-certified compa-nies. http://www.qualitydigest,com, accessed December12, 2002.

Repenning, N. 1999, Vou measure what you get: Toward atheory of process improvement and change. Workingpaper, MIT, Cambridge, MA.

Repenning, N, 2000. Drive out fear (unless you can drive it in):The role of agency and job security in process improve-ment. Management Science. 46: 1385-1396.

Rotemberg, J,, & Saloner, G. 2000. Visionaries, managers andstrategic direction, fland Journal of Economics, 31: 693-716.

Samson, D,, 8f Terziovski, M. 1999, The relationship betweentotal quality management practices and operational


performance, /ournal of Operations Management. 17:393-409.

Sanderson, S., & Uzumeri, M, 1995. Product platforms anddominant designs: The case of Sony's Walkman. Re-search Policy, 24: 583-607.

Sitkin, S, 1992, Learning through failure: The strategy ofsmall losses. Research in Organizational Behavior, 14:232-266,

Sitkin. S,, & StickGl. D, 1996, The road to hell: The dynamicsof distrust in an era of quality. In R. Kramer & T, Tyler(Eds.), Trust in organizations: 196-215, London: Sage,

Sitkin, S, B., Sutcliffe, K, M,, & Schroeder, R. G. 1994. Distin-guishing control from learning in total quality manage-ment A contingency perspective. Academy of Manage-ment Review. 19: 537-564.

Staw, B., & Epstein, L, 2000, What bandwagons bring: Effectsof popular management techniques on corporate perfor-mance, reputation and CEO pay. Administrative ScienceQuarterly, 45: 523-556,

Sterman, J, D.. Repenning, N, P,, & Kofman, F, 1997. Unantic-ipated side effects oi successful quality programs: Ex-ploring a paradox of organizational improvement. Man-agement Science, 43: 503-521,

Sull, D., Tedlow, R,, & Rosenbloom, R, 1997. Managerial com-mitments and technology change in the U,S. tire indus-try. Industrial and Corporate Change. 6: 461-500.

Sutcliffe, K,, Sitkin, S,. & Browning, L. 2000, Tailoring processmanagement to situational requirements. In R, Cole &W, Scott (Eds,), The quality movement & organizationtheory: 315-330. London: Sage.

Teece, D, I., Pisano, G,, 8E Shuen, A. 1997. Dynamic capabili-ties and strategic management. Strategic ManagementJournal 18: 509-533.

Tripsas, M. 1997. Surviving radical technological changethrough dynamic capability. Industrial and CorporateChange. 6: 341-377.

Tripsas, M,, & Gavetti, G, 2000. Capabilities, cognition, andinertia: Evidence from Digital Imaging. Strategic Man-agement Journal. 21: 1147-1161,

Tushman, M, L,, 8f Anderson, P, 1986. Technological discon-tinuities and organizational environments. Adminisfra-five Science Quarterly. 31: 439-465,

Tushman, M, L,, & Murmann, J, P, 1998. Dominant designs,technology cycles, and organizational outcomes. Re-search in Organizational Behavior, 20: 213-266.

Tushman, M,, & Nelson, R, 1990, Technology, organizations

and innovation: An introduction. Adminis(ra(ive ScienceQuarterly, 35: 1-8,

Tushman, M, L,, & O'Reilly, C, 1997, Winning fhrougii inno-vation, Boston: Harvard Business School Press.

Tushman, M,, & Romanelli, E. 1985. Organizational evolution:A metamorphosis model of convergence and reorienta-tion. Research in Organizational Behavior, 7: 171-222,

Tushman, M. L,, & Rosenkopf, L. 1992. Organizational deter-minants oi technological change: Toward a sociology oftechnological evolution. Besearch in Organizational Be-havior, 14: 311-347.

Tushman, M., & Smith. W. 2002, Organizational technology.In J, Baum (Ed.), Companion to organizations; 386-414.London: Blackwell.

Tyre, M., & Orlikowski, W. 1994, Windows of opportunity:Temporal patterns of technological adaptation in or-ganizations. Organization Science. 5: 98-109.

Utterback. J, 1994. Mastering the dynamics of innovation.Boston: Harvard Business School Press,

Van de Ven, A., Angle, H., & Poole, M. 1989, Research in themanagement of innovation. New York: Harper & Row,

von Hippel, E. 1988, The sources of innovation. New York:Oxford University Press,

Weick, K. E, 1969. The social psychology ol organizing. NewYork: McGraw-Hill,

Weick, K. 1976. Educational organizations as loosely coupledsystems. Administrative Science Quarterly. 21: 1-19.

Weick, K, 1995, SensemaJting in organizations. ThousandOaks. CA: Sage.

Westphal, J, D,, Gulati, R,, & Shortell, S, M, 1997, Customiza-tion or conformity? An institutional and network per-spective on the content and consequences of TQM adop-tion. Administrative Science Quarterly, 42: 366-394.

Williams, K,, & O'Reilly, C, 1998, Demography and diversityin organizations: A review of forty years of research.Research in Organizational Behavior. 20: 77-140.

Winter, S. 1994, Organizing ior continuous improvement:Evolutionary theory meets the quality revolution. InJ. Baum & I, Singh (Eds,}, Evolutionary dynamics o/organ-izations: 90-108. New York: Oxiord University Press.

Wruck, K,, & Jensen, M. 1994. Science, specific knowledge,and total quality management. Journal of Accountingand Economics, 18: 247-287,

Zbaracki, M, J, 1998. The rhetoric and reality of total quality man-agement. Administrative Science Quarterly, 43: 602-636.

Mary I. Benner is an assistant professor in management at The Wharton School of theUniversity o! Pennsylvania. She received her Ph,D. from the Graduate School oiBusiness, Columbia University, Her research focuses on technological innovation,organizational adaptation to technological change, and the efiects of process man-agement on innovation and adaptation.

Michael L. Tushman is the Paul R, Lawrence MBS Class of 1942 Proiessor of BusinessAdministration at Harvard Business School, He received his Ph,D. at MIT and waspreviously on the faculty of the Graduate School of Business, Columbia University, Hisresearch explores the relations among technological change, executive leadership,and organizational adaptation.

Documents

EXPLOITATION, EXPLORATION, AND PROCESS … · MARY I. BENNER University of Pennsylvania MICHAEL L. TUSHMAN ... ature with contingent theory and testable prop-ositions about process