Inter-project learning: processes and outcomes of...

Research Policy 30 (2001) 1373–1394

Inter-project learning: processes and outcomes of knowledgecodification in project-based firms

Andrea Prencipea,∗, Fredrik Tellba CoPS Innovation Centre, SPRU, University of Sussex, Mantell Building, Falmer, Brighton BN1 9RF, UK

b Department of Management and Economics, Linköping University, SE-58183 Linköping, Sweden

Abstract

In this paper, we argue that the literature on knowledge codification has been overly concerned with the economic propertiesof its outcomes, neglecting the importance of its underlying learning processes. Following Zollo and Winter [OrganisationScience, 2001, in press], the paper distinguishes three learning processes: experience accumulation, knowledge articulationand knowledge codification and suggests a framework to analyse the learning abilities of project-based firms. We proposethat mechanisms for inter-project learning draw upon these learning processes and can be found at various levels of theproject-based firm. Using empirical evidence from six case studies, we discern three empirical patterns, that we definedlearning landscapes, of such mechanisms. Implications for the literature and practice of knowledge codification are discussed.© 2001 Elsevier Science B.V. All rights reserved.

Keywords:Project-based firm; Organisational learning; Firm; Codified knowledge

1. Introduction

The literature on the knowledge-based theoryof the firm suggests that the capabilities of a firmare learning-based (Grant, 1996; Lazonick andO’Sullivan, 1996; Dosi et al., 2000). Accordingly,the core activities of the firm, what the firm choosesto produce and sell, as well as the boundaries ofthe firm, are determined by the knowledge a firmpossesses. As observed by Levitt and March (1988),learning in the firm tends to be local. Interpretationof experience is difficult, as lessons must be drawnfrom a relatively small number of observations in acomplex and changing environment. This makes itlaborious to identify causality and draw correct in-

∗ Corresponding author. Tel.:+44-1273-678936;fax: +44-1273-685865.E-mail addresses:[email protected] (A. Prencipe),[email protected] (F. Tell).

ferences. Organisations and organisational membersexhibit systematic biases in interpretation, since theyconcentrate overwhelmingly on recent and salientevents. Also, they may be insensitive to sample size,attribute too much importance to intentionality, andmay use simple and linear algorithms.

The aim of this paper is to discuss the learning abili-ties of project-based firms. We study whether and howproject-based firms are able to capitalise on knowledgethat is acquired during the execution of one project andtheir ability to transfer it to other projects or parts of theorganisation. In its ideal form, the project-based firmis organised solely around projects. Galbraith (1969)suggested that there is a continuum of organisationalforms ranging from the pure functional form throughthe matrix form to the pure product- or project-basedform. While in the functional form, a firm’s businessactivities are organised according to functional spe-cialisation (e.g. marketing, R&D); in the matrix formthese activities are organised both within projects and

0048-7333/01/$ – see front matter © 2001 Elsevier Science B.V. All rights reserved.PII: S0048-7333(01)00157-3

1374 A. Prencipe, F. Tell / Research Policy 30 (2001) 1373–1394

along functional lines. Within a pure project-basedfirm, projects “embody most, if not all, of the businessfunctions normally carried out within departments offunctional or matrix organisations” and act as the mainmechanisms for co-ordinating and integrating them(Hobday, 2000, p. 874). The project-based form has,for instance, been studied in consultancy and mar-keting firms (Alvesson, 1995), the film industry (De-Fillippi and Arthur, 1998), in architectural practice(Winch and Schneider, 1993), and construction firms(Gann and Salter, 1998).

The project-based firm can also be found in the pro-duction of so-called complex product systems (CoPS)industries (Davies and Brady, 2000; Hobday, 1998,2000; Prencipe, 2000). CoPS are capital-, engineering-and IT-intensive, business-to-business products, net-works, constructs and systems. They are often pro-duced in multi-firm alliances, as a one-off or in smallcustomised batches for specific customers and mar-kets. Examples include global business networks, air-craft engines, civil airliners, power stations, off-shoreoil platforms, mobile telephone systems and large civilengineering projects. The focus of this paper is on thiscategory of products.

The strong focus on projects displayed by firms de-veloping CoPS suggests that there might be problemsassociated with organisation-wide learning (Lind-kvist et al., 1998). While in a functionally-basedfirm, departments act as knowledge silos, the pureproject-based firms lack the organisational mech-anisms for the knowledge acquired in one projectto be transferred and used by other projects. Twofurther issues impair organisation-wide learning inproject-based firms: the unique and the temporarynature of projects (Brusoni et al., 1998). With re-gard to the former, projects differ from each other inseveral, critical aspects. They entail heterogeneousactivities that may well not be repeated in successiveprojects. If projects exhibit one-off characteristics,the project-based firm confronts the difficult task of“learning from samples of one or fewer” (March etal., 1991). In addition, projects may be characterisedby relatively long life cycles, requiring similar projectactivities to be retrieved and repeated after long timeintervals. With regard to the temporary nature ofprojects, projects can be characterised by the tem-porary constellation of people they entail (DeFillippiand Arthur, 1998; Tell and Söderlund, 2001). This

feature implies that new human encounters and rela-tionships take place whenever a new project is started,which may increase the barriers to learning from theprevious experience of others.

Warglien (2000) argued that project-based firms canbe regarded as apopulation of projects.

If one puts on Darwinian spectacles and looks at theprojects portfolio of such firm, he will see popula-tions of entities (the single projects) that are born,die, compete for limited resources (skilled labour,equipment, and financial means), inherit clients,technologies and other traits from preceding gener-ations, and re-transmit them (sometimes modifiedor innovated) to next generations (Warglien, 2000,p. 5).

Based on Warglien, we submit that projects canbe characterised byquasi-genetictraits (Cohen et al.,1996) that can be technological (i.e. technologies,components) and organisational (reporting systems,project management). These quasi-genetic traits em-body the firm’s capabilities and routines and “can usu-ally be inherited from project to project, and they aresometimes modified by project development efforts”(Warglien, 2000, p. 5). Following Davies and Brady(2000), who found that firms developing CoPS try toeconomise by the repetition of similar categories ofprojects, this paper analyses the mechanisms adoptedby project-based firms to make these quasi-genetictraits inheritable from project to project.1 In otherwords, we look at how these firms are able to learnbetween projects.

We build on the literature on knowledge manage-ment and, in particular, on the debate on knowledgecodification (Steinmueller, 2000). It has been sug-gested that the ability to codify tacit knowledge car-ries some important implications for the efficiencyproperties of economic organisation. In fact, if knowl-edge can be codified and commodified, the ease ofknowledge transfer will increase and costs associatedwith such transfer will decrease (Cowan and Foray,1997). Given the features of a project-based firm

1 We are aware, however, of the limits of the biological metaphor.As pointed out by several authors (e.g. Penrose, 1952), in organi-sations such as business firms, on the one hand there is no sexualreproduction, but on the other hand there is deliberate action andstrategy.

A. Prencipe, F. Tell / Research Policy 30 (2001) 1373–1394 1375

highlighted above, knowledge codification could haveprofound implications for organisation-wide learning.Although the impact of codification for the efficiencyof organisational processes warrants further study, weargue that the debate on codification focuses mainly(and wrongly so) on the importance of theoutcomesof codification leaving the cognitiveprocessunderly-ing it unexplored. We base the analytical frameworkof this paper on Zollo and Winter (2001) who em-phasised the relevance of theprocessdimension ofknowledge codification.

This article is structured as follows. Section 2briefly introduces the debate on knowledge codifica-tion. Based on Zollo and Winter (2001), Section 3suggests a simple model that outlines three learningprocesses that occur in the firm (experience accumu-lation, knowledge articulation, and knowledge codi-fication). We then identify the different mechanismsused by the project-based firms investigated in rela-tion to their inter-project learning processes. Thesemechanisms are categorised intolearning landscapesof an ideal form (Section 4). We conclude the articleby discussing the implications for firm capabilitiesand the processes and outcomes of knowledge codifi-cation, as well as identifying future research issues.

2. A note on the codification debate

Increasingly, the nature and processes of organisa-tional knowledge have become the focus for theoristsin their attempts to delineate a viable theory of thefirm. A number of definitions of organisational knowl-edge has been suggested in the literature (see, e.g.Tell, 1997; Baumard, 1999; Eisenhardt and Santos,2001, for literature reviews). The aim of this sectionis to briefly introduce the literature on the codificationdebate.

In reviewing current perspectives on knowledgecodification, we are going to employ two assump-tions. The first is that information and knowledge aretwo different concepts. Dosi et al. (1996) consideredinformation as “well stated and codified propositionsabout (i) states-of-the-world (e.g. “it is raining”; (ii)properties of nature (e.g. “A causes B”); (iii) iden-tities of other agents (e.g. “I know Mr. X and he isa crook”) and (iv) explicit algorithms on how to dothings”. Knowledge is instead understood as includ-

ing “(i) cognitive categories; (ii) codes of interpreta-tions of the information itself; (iii) tacit skills and (iv)search and problem-solving heuristics irreducible towell-defined algorithms”. We submit that informationis symbol, code: a sign without meaning (cf. Dretske,1981), whereas knowledge is meaning, meaning thatcan only be given by an observing system.

This distinction between the two concepts of in-formation and knowledge leads to the second as-sumption. Organisations, and particularly businessorganisations, are viewed here as knowledge-basedrather than simple information-processing systems(Dosi and Marengo, 1994; Fransman, 1994; Nonakaand Takeuchi, 1995). Firms do not merely processinformation, but elaborate and interpret it accordingto beliefs developed and accumulated by them overtime. New knowledge is acquired via firm-specificprocesses, in turn sedimented over time, which filterand constrain a firm’s current behaviour.

Cowan and Foray (1997) defined the codificationprocess as “The process of conversion of knowledgeinto messages that can be processed as information”(p. 596). Hence, codification refers to the processof knowledge being transformed into information,where information is in the form of messages, orsets of identifiable rules and relationships, that canbe transmitted to decision agents (Kogut and Zan-der, 1992; Ancori et al., 2000). Codified knowledge,thus, can be characterised as information-like andobjectified (Kogut and Zander, 1992; Spender, 1996).In business practice, knowledge codification seemsto play an important role in what organisations andbusiness consultants labelknowledge management.Through the development of new technologies moreinformation can be stored and retrieved. A number oftechnological applications developed for knowledgemanagement aim at substituting human activity andknowledge with rule-based systems that can aid, orperform, problem-solving (Davenport and Prusak,1998).

Two antagonistic positions in the codification de-bate on the pros and cons of codification of knowledgecan be discerned. Ancori et al. (2000, p. 257) labelledthese the ‘absolutist position on codification’ (empha-sising that all knowledge can be codified) versus the‘absolutist position on tacit knowledge’ (for whomall codified knowledge requires tacit knowledge to beuseful).


Advocates of codification argued that it reducesthe cost of knowledge acquisition due to ease ofstorage, retrieval and reliability. As an outcome, theeconomic properties of codified knowledge are verymuch akin to the economics of information (Arrow,1984). There are high initial fixed costs, but sincemessages become reproducible successive operationscan be carried out at very low marginal cost. Throughcodification complexity can be reduced when confin-ing the description and analysis of a domain into whatis expressible in codes (Boisot, 1995). This meansthat codified knowledge can instruct machines, as isthe case with computers, which then can substitutefor people in certain situations. Because of the publicnature of codified knowledge, codification has the po-tential to reduce asymmetric information in markets.Moreover, codification may transform knowledge intoa commodity that can be bought and sold in markets,thereby allowing firms to purchase knowledge insteadof developing it internally (which may lower costsfor knowledge acquisition, e.g. through outsourcing).Through the creation of a memory external to individ-uals, codification may make organisations less vulner-able to loss of tacit knowledge stored in individuals.

Advocates of tacit knowledge, however, pointed tothe drawbacks of organisations’ relying too much oncodification of knowledge. For instance, the inher-ent flexibility in the skilled use of tacit knowledgeseems to suggest that the exercise of such skills in-volves more than the internalisation of codified rules(cf. Nightingale, 2001). If followed accurately and pre-cisely, codified knowledge, understood as rules, mayinstead imply rigidity. Hence, in the context of changeexcess codification can stifle the development of newknowledge and inflict stability and inertia on systems.Moreover, the production of codified knowledge im-plies the production of new forms of tacit knowl-edge, due to the contextual nature of economicallyvaluable knowledge. In addition, and implicit in this,non-algorithmic judgement is necessary even in con-texts where it would seem that complete codificationis possible.

In this paper, we will refrain from taking either ofthese positions. In line with Hansen et al. (1999), whodiscussed knowledge management strategies in termsof codification and personalisation (contingent uponthe strategy of the firm), our position is that severalstrategies for knowledge management are viable for

the firm. Also, we argue that what is lacking in sucha simplistic model of the benefits and drawbacks ofknowledge codification is the importance of aknowingsubject being involved in most learning processes (cf.Polanyi, 1962). We want to emphasise the cognitiveelement of knowledge in organisations by pointing tothe activities of theknowerin project-based firms. Asrecognised by several authors (e.g. Brown and Duguid,1991; Lave and Wenger, 1991; Weick, 1995; Spender,1996; Baumard, 1999), suchknowersmay be individ-uals, but are often participants in a collective commu-nity that develop their knowledge by drawing uponseveral sources of knowledge in their actions. In aproject-based context, both individual and collectiveknowledge processes may be hypothesised, due to thecomplexity of the problems involved that implies bothindividual and collective effort. In conducting this re-search on knowledge codification, therefore, we tryto ask questions not only to establish the outcomesof codification, but also as a way of informing our-selves about what the possible cognitive and socialmechanisms involved may be. What are the processesin project-based firms involved in the management ofknowledge? What are the collective mechanisms usedfor developing knowledge that can be reused in otheractivities and projects?

3. Managing knowledge in project-based firms: atentative framework

Since we consider learning between projects as,at least in part, a collective endeavour, we followZollo and Winter’s (2001) argument on the evolutionof collective knowledge within organisations. As hasbeen suggested by several authors (e.g. Nonaka andTakeuchi, 1995; Boisot, 1998; Baumard, 1999; Ancoriet al., 2000), the dynamics of learning in organisationscan be depicted as a cyclical pattern. We focus on theknowledge processes associated with learning fromexperience and the possibility for the organisation toarticulate, elaborate and draw inferences from experi-ence (Levinthal and March, 1993; Levitt and March,1988; March et al., 1991).

Zollo and Winter (2001) identified three learningprocesses, namelyexperience accumulation, knowl-edge articulationandknowledge codification. They ar-gued that the effectiveness of these processes depends


on the characteristics of the tasks that the organisa-tion attempts to learn. In particular, the frequency,heterogeneity and causal-ambiguity of the task renderexperience accumulation and knowledge articulationand codification more or less effective. The lowerthe frequency (temporary nature) and the higher theheterogeneity (uniqueness) of the task, the more ef-fective are knowledge articulation and knowledgecodification. Their argument rests on the concept ofthe learning investment function, according to whichknowledge articulation and knowledge codificationinvolve more resources (e.g. time) and cognitive ef-forts than the mere accumulation and routinisationof knowledge and experience. Also, Zollo and Win-ter (2001) emphasised the processes of knowledgearticulation and codification as compared to their out-comes. In line with their argument, we argue that theprocesses of knowledge articulation and codificationserve an important purpose in the dynamics of learn-ing between projects as the project intensity in CoPSproduction implies unique (and therefore heteroge-neous) and temporary (and therefore infrequent) tasks.

We argue that the distinction between the dimen-sions of outcome and process is fundamental to un-derstanding the knowledge management strategies ofproject-based firms. When conceptualising codifica-tion, many writers drew on Kogut and Zander’s (1992)useful insight that knowledge codification facilitatesthe transfer of knowledge and, thus, contributes tothe firm’s combinative capability. In line with this,knowledge codification is important because of thegains that can be made through new combinations ofstocks of codified knowledge. Hence, incentive struc-tures should be established, to encourage the codifica-tion of knowledge due to this outcome property. Whileacquiescing with this argument, this paper acknowl-edges both the stock (outcome) and flow (process)characteristics of knowledge as an economic resource(Dierickx and Cool, 1989), with particular emphasison the latter. This process aspect of knowledge artic-ulation and codification seems to have been neglectedin the existing research, which tends to focus moreon the outcome and the economic benefits related tosuch outcomes.2 Fig. 1 is an attempt to single out thelearning typologies, outcomes and economic benefits

2 The distinction between knowledge as process and knowledgeas outcome is also discussed in Paoli and Prencipe (1999).

related to experience accumulation, knowledge articu-lation and knowledge codification. The figure also at-tempts to frame and highlight the focus of the currentdebate on knowledge codification.

3.1. Experience accumulation

Levitt and March (1988) argued that organisationallearning is target-oriented, is based on historical ex-perience and stored in routines. As emphasised byNelson and Winter (1982), routines can be charac-terised by their tacit and programmatic nature. Suchroutines are an outcome of trial-and-error and reflectthe accumulation of experiential wisdom (Gavetti andLevinthal, 2000). Experience-based learning tends tobe local, i.e. closely related to existing routines.Learn-ing by doingandlearning by usingare based on experi-ence from actions where actors either have difficultiesin drawing or ignore inferences to causality (Levinthaland March, 1993). The knowledge developed by or-ganisations in such situations, thus exhibits certainelements of procedural rationality, lacking consciousvolition (Nelson and Winter, 1982). In such situations,actors learn fairly passively through what Argyris andSchön (1978) labelled single-loop learning.

Viewing the organisation as a collection oflearning-based routines, Levinthal and March (1993)suggested that organisations run the risk of myopia,exemplified by capability traps and superstitiouslearning. Such learning disabilities stem from thetendency of organisations to execute existing opera-tional routines in response to problems encountered,and the restricted range of alternatives that searchroutines may select from (Nelson and Winter, 1982).As discussed by Zollo and Winter (2001), experi-ence accumulation and organisational routines arefundamental for firms’ capability development. First,the smooth functioning of routines creates the pos-sibility for automatic behaviour, which requires lessattention and effort on behalf of the skilled worker(cf. Penrose, 1959). Second, organisational routinesallow for efficient specialisation and coordination.As discussed by Nelson and Winter (1982, p. 105):“While each organisation member must know hisjob, there is no need for anyone to know anyoneelse’s job. Neither is there any need for anyoneto articulate or conceptualise the procedures em-ployed by the organisation as a whole” (emphasis


Fig. 1. Learning typologies, outcomes and economic benefits.

in original). Skilled labour that performs routinisedactivities can be partitioned using division of labour,hence giving rise to benefits related toeconomics ofspecialisation.

3.2. Knowledge articulation

The behavioural focus on learning as routine-based,however, neglects some of the deliberative processesinvolved in organisational learning (Nonaka andTakeuchi, 1995; Witt, 1998; Zollo and Winter, 2001).Apart from the procedural nature of organisational

learning, there are arguably elements of substantiverationality or logic of consequence involved (Marchand Olsen, 1989). Through agents’ abilities to ex-press opinions and beliefs (Zollo and Winter, 2001),the ability to develop visions (Fransman, 1994) andthe creation of metaphors and analogies (Nonaka andTakeuchi (1995), cognitive processes drawing moreglobal inferences and determining causalities are trig-gered. Therefore, Zollo and Winter (2001) submitthat articulation processes form a second importantdevice of the cognitive dimensions of organisationallearning processes.


Being able to articulate practical knowledge is animportant facet of the reflective practitioner (Schön,1983). Schön (1983) discussed how the practitioner byexperimenting can obtain reflection-in-action, that is,an ability to attend to a situation where he/she is exe-cuting routine-based knowledge. Hence, in the frame-work of Schön, articulation of knowledge may occurin the practical context of experiential learning. At theindividual level, the learning typologies at the basis ofthe articulation process arelearning by reflectingandlearning by thinking.

Zollo and Winter (2001) pointed out that articula-tion of knowledge performs two roles. First, it consti-tutes a context for justification (cf. Tell, 1997, 2000;Grand and Von Krogh, 2000). Second, it is a cognitiveprocess that implies deliberation and carries the pos-sibility that individuals and groups can come to gripswith causality and feasibility in performing differenttasks (cf. Nonaka and Takeuchi, 1995; Witt, 1998).By discussing the role of conversation in the articula-tion of knowledge through reflection, both Zollo andWinter (2001) and Schön (1987) introduce a collec-tive element of knowledge articulation. By dialogueand discussion knowledge can be articulated by or-ganisational members and an arena can be created fordouble-loop learning (Argyris and Schön, 1978). Be-sideslearning by reflectingand learning by thinking,the articulation process subsumes alsolearning by dis-cussingand learning by confronting.

The articulation process improves the understandingof action–performance relationships and enables thecreation of agreed upon representations. This creationof agreed upon representations facilitates communi-cation amongst the actors using the concepts embed-ded in such representations (Foray and Steinmueller,2001). Common representations, in turn, allow for bet-ter task co-ordination (economics of co-ordination).

3.3. Knowledge codification

Zollo and Winter (2001) further argued that an or-ganisation may produce even higher cognitive effortsin the process of codification. In a sense codification isan extension of articulation. Even more so than artic-ulation, the ability to codify knowledge allows for thecreation of externalised knowledge, brought forwardin linguistic and symbolic representations. Moreover,codification processes are associated with much ef-

fort and high costs. When individuals in organisationscodify articulated knowledge into codebooks (Cowanet al., 2000), the aim may be to reveal even strongerlinks between actions and outcomes. In such cases, ef-forts may involve, e.g. the screening of multiple sce-narios, different explanatory frameworks, or the test-ing of different organising principles. Codified rules ascontained in manuals and procedures can also merelyserve to provide guidelines for repetitive actions. Insuch instances, codification (as an outcome) primar-ily serves the purpose of facilitating routine replica-tion. The economic benefits of codification lie primar-ily in the re-use and diffusion of codified knowledge(economics of information).

As emphasised by Zollo and Winter (2001),however, the cognitive efforts of creating codifiedknowledge from what has been (perhaps) tacitlyknown involves creative elements (cf. Nonaka andTakeuchi, 1995) as well as the establishment of inter-nal selection processes. Along the same lines, Ancoriet al. (2000, p. 258) suggested that the cognitive and or-ganisational mechanisms mobilised by codification areparticularly important. These authors submitted thatknowledge is dependent upon the cognitive abilities ofactors and cannot be separated from the communica-tion process through which it is exchanged. FollowingForay and Steinmueller’s (2001) discussion, one candistinguish between two functions of codification. Thefirst function is that codified systems of symbols allowfor storage and transfer across time and space. Thesecond function of codification is to allow humans torearrange, manipulate and examine symbols and sym-bolic relationships in order to transform the underly-ing knowledge represented in such systems. Hence,not only is there an aspect of inscribing what is tacitlyknown involved in codification, but also, as Foray andSteinmueller (2001) emphasised, a higher effect ofchanging knowledge structures by the potential andactual transpositions implied by such a literate form ofknowledge representation. This line of reasoning sug-gests that besides the substantial cognitive investmentin the learning by writing and re-writing suggestedby Zollo and Winter (2001), organisationslearnby implementing, replicating and adapting codifiedknowledge.

The processes of experience accumulation, knowl-edge articulation and knowledge codification are cer-tainly not unidirectional. Moreover, there are overlaps


between these types of learning and any seeminglydefinite distinction between them may sometimes blur.What we have tried to emphasise in the preceding dis-cussion, however, is an increasing element of delib-eration and, in one sense, rationality, as articulationand codification processes are included in an under-standing of the creation of organisational knowledge.To borrow Weick’s (1995) terms, one could say thatsense-making processes become an explicit elementof the learning processes involved when individualsand communities of individuals create representationswhich they can use to interpret and elaborate on ex-periences encountered (Choo, 1998).

4. Project-to-project learning and learninglandscapes: some empirical evidence

This section illustrates the approaches of project-based firms to managing knowledge and the mainmechanisms adopted by them to accumulate experi-ence, and to articulate and codify knowledge at the in-dividual, project (or group) and organisational levels.The empirical evidence reported here originates fromsix case studies of project-based firms (see AppendixA for the research method). We found that firms investin a variety of tools and mechanisms to try to capitaliseon the knowledge developed during the execution ofone project and transfer it across the organisation. Wedeveloped a 3× 3 matrix to categorise the variousproject-to-project learning mechanisms (Fig. 2). Thehorizontal dimension of the matrix refers to experienceaccumulation, knowledge articulation and knowledgecodification, shown in the first row of the matrix pre-sented in Fig. 1, while the vertical refers to the levelof analysis (i.e. individual, project — or group — andorganisational).3

The project-to-project learning mechanisms thatfirms may use vary according to both the hori-

3 Whereas the majority of the learning mechanisms we focus onin this paper may hint at a conceptualisation of learning internalto the firm, we point out that the sources of such learning maybe both internal and external. In fact, given their multitechnology,multicomponent nature, complex product systems are increasinglydeveloped via collaborative partnerships with suppliers as wellas with customers. This renders externally stimulated learningas important as (and sometimes more important than) internallystimulated learning.

zontal and vertical dimensions. The analysis ofproject-to-project learning mechanisms, based on acombination of the two proposed dimensions, enablesa detailed assessment of the type of approach thatfirms adopt to project-to-project learning. In particu-lar, the analysis of the horizontal dimension enablesan assessment in relation to the identified learningprocesses (experience accumulation, knowledge ar-ticulation and knowledge codification) on which afirm focuses. Along the vertical dimension, the ma-trix maps the project-to-project mechanisms on to theindividual, project, and organisational levels. Boisot(1995), Nonaka and Takeuchi (1995), and Brownand Duguid (1991) studied the mechanisms throughwhich knowledge becomes diffused throughout anorganisation. The analysis of the vertical dimensionenables an assessment of the approach that firmsuse to diffuse knowledge across projects and withinthe organisation. Firms may be more or less focusedon the replication and diffusion of routines (or bestpractices) across the organisation.

The analysis of the horizontal and vertical dimen-sions combined enables the identification of what wehave termed a firm’slearning landscapein relation toproject-to-project learning. We define a firm’slearn-ing landscapeas the mix of project-to-project learn-ing mechanisms adopted and implemented.4 We ar-gue that the concept of learning landscape reflectsthe multidimensional nature of a firm’s approach toproject-to-project learning. Our approach to a firm’sknowledge management strategy builds upon Hansenet al. (1999), who maintained that firms follow twoknowledge management strategies, namely the per-sonalisation strategy and the codification strategy. Infirms that pursue a personalisation strategy, “knowl-edge is closely tied to the person who developed itand is shared mainly through direct person-to-personcontact” (Hansen et al., 1999, p. 107). A codifica-tion strategy, on the other hand, would revolve aroundICT-based technologies, “knowledge is carefully cod-ified and stored in databases, where it can be accessedand used easily by anyone in the company” (p. 107).Hansen et al. (1999) also argued that to use knowl-

4 Learning mechanisms are empirical instances such as, e.g.lessons learnt meetings, databases or informal encounters. Thelearning landscapethen refers to the collection, or portfolio, ofsuch mechanisms, here clustered into three distinct patterns.


Fig. 2. Inter-project learning mechanisms.


edge effectively, firms should focus on one dominantstrategy and use the other in support of the dominantone. They maintained that firms that try to pursue bothstrategies at the same time run the risk of failing atboth, and proposed an 80–20% split between the dom-inant and the supporting strategy.

We argue that by proposing the concept of a firm’slearning landscape our approach extends and enrichesthat of Hansen et al. (1999). The personalisationand codification strategies proposed by Hansen et al.(1999) focused on only one dimension, defined by thecontinuum whose extreme poles are tacit and explicitknowledge. These two types of knowledge manage-ment strategy would find their place in the top-leftand bottom right quadrants of the matrix, respec-tively, as shown in Fig. 2. Relying on the discussionabove, however, our research shows that a contin-uum of approaches defined by personalisation andcodification strategies is hardly a good approxima-tion of a firm’s knowledge management strategies. Amono-dimensional interpretation of a firm’s approachto knowledge management is reductivist to the pointthat it may obscure the identification and the relatedunderstanding of important elements that characterisesuch an approach.

The concept of learning landscape takes into ac-count the multidimensional character of a firm’s ap-proach to managing knowledge. It broadens the focusand emphasises the process dimension of a firm’sknowledge management approaches to experienceaccumulation, knowledge articulation and knowl-edge codification and relates them to the individual,project, and organisational levels. Based on the em-pirical evidence gathered during our field study, weidentified three main types of learning landscape (thatis, theexploreror L-shaped landscape, thenavigatoror T-shaped landscape, and theexploiter or stair-case landscape) and position them within the matrixshown in Fig. 2.

4.1. The explorer (or L-shaped) landscape

Firms that rely to a great extent on people-embeddedknowledge are characterised by L-shaped land-scapes. These firms emphasise experience accumu-lation processes and knowledge transfer throughpeople-to-people communication, and are charac-terised by a strong and receptive culture. We class

these firms as explorers, since we found that their rel-atively small size in combination with the features oftheir corporate culture, constitutes a strong enablingspringboard from which to explore different routesto adopt and implement ‘project-to-project learning’mechanisms based on knowledge articulation andcodification processes (Fig. 3).

The first firm is part of a larger defence conglom-erate and develops and manufactures complex, hightechnology products for the defence industry. Al-though it has had two changes in ownership duringthe last decade, the firm has maintained a corpo-rate culture and identity. A positive and co-operativeorganisational culture is, in fact, one of the mainfeatures of the firm. This friendly environment ispartly due to its relatively small size and the geog-raphy of its main plant (i.e. open-plan offices), “weare in walking distance”. The top managers of thefirm stated that their main knowledge resources liein people’s heads. Employees (design and softwareengineers and project managers) are equipped withpaper-based tools that should help them articulateand record their day-to-day activities so that can bere-used in different projects.

The firm lacked formal ‘project-to-project learning’mechanisms. Employees relied to a great extent onpersonal and informal contacts for project staffing andknowledge transfer purposes. The project managersinterviewed underlined that the lack of formal toolshampered the re-use of project knowledge, especiallyin those phases where knowledge does not become em-bedded in drawings and/or products. One good exam-ple of this is the bidding phase. Although the outcomeof this phase is eventually incorporated in a document,interviewees argued that such a document captures lit-tle or nothing of the process that underlies the biddingprocedure. They therefore disapproved of the lack offormal tools that could be considered as enablers forthe re-use of project knowledge.

The second firm operates in the aerospace sector.Following its acquisition by a larger firm, it metamor-phosed from a functional to a matrix organisation.The firm is now organised around Customer FacingBusiness Units that focus on a particular group of cus-tomers. The internal business units are called Operat-ing Business Units and are organised around the mainproduct components. Each of the Operating BusinessUnits includes technologists, buyers, and the manufac-


Fig. 3. The L-shaped learning landscape.


turers of the specific component for which they haveresponsibility. For a new project, Integrated ProductTeams are constituted of engineers from the Operat-ing Business Units and a team leader is appointed.This team leader role changes according to the stageof the project. At the start a design engineer would beleader. A supply chain manager would take over oncethe design is completed. A service manager wouldassume the lead to sustain delivery. There is a projectmanager to oversee the work of the Integrated ProductTeams.

The firm is characterised by an informal, people-based culture. Personal contacts are paramount for thestaffing of projects. The transfer of project knowl-edge occurs through people-to-people communication.One of the directors stated that although project re-views are held throughout the projects, they are aimed95% at reviewing costs and schedules, and only 5%to lessons learnt. He added that this 5% is insufficientand that the firm had as far as inter-project learningwas concerned a long way. What usually happens isthat at the end of a project, the financial resourcesare exhausted and there is no time to look at thelessons learnt so knowledge continues to be transferredfrom project to project via people. Another example ofthe informal, people-based approach to inter-projectlearning is the Project Management Forum initiative.This is led by project managers and consists of bi-monthly events to talk about tools and the share ofinformation.

The shift to a matrix form is being accompaniedby a shift in the principles that inform the manage-ment of organisational learning. In particular, the firmis trying to engender a procedure-oriented culture. Ineffect, the firm’s ad hoc project management proce-dure is being replaced by a more formal one involvingtraining courses and university-level education. TheProject Management Forum has evolved into an In-tegrated Program Management Team (composed offormer senior project managers) whose aim is to de-velop and implement new formal procedures (mainlyICT-based) for project managers. This team is sup-posed to establish a Centre of Excellence. Accordingto the interviewees, however, the firm is strugglingto adopt and adapt to this procedure-oriented culture.One project manager, for instance, argued that the newIntegrated Program Management Team is not workingwell because it is stifling the real learning.

4.2. The navigator (or T-shaped) learning landscape

This landscape characterises firms that started im-plementing mechanisms for project-to-project learn-ing based on a knowledge articulation process. Theirfocus was on the implementation and appreciationof these mechanisms not only at the individual andproject levels but also, and mainly, at the organisa-tional level. These firms navigate through a few evolv-ing routes to improve their project-to-project learningthat may become established at some point (Fig. 4).

The first case is from a firm that develops flightsimulators for both civilian and military purposes.The firm has gone through a series of organisa-tional changes following a period of mergers andacquisitions. Here the focus is on the organisationof lessons learnt meetings. The project managementguide at one time considered this mechanism onlyat the end of each project. Because of the lengthof the development process and the ensuing phe-nomenon of selective memory, the firm decided toadopt a life-cycle approach. According to this ap-proach, the entire project is divided into phases(from bidding, to delivery to customer) and at theend of each phase a lessons-learnt meeting is held.Project managers usually lead these end-of-phasemeetings.

(We) have a process called lessons learnt: as you gothrough all the phases of a life cycle that a lessonslearnt process could be applied. Traditionally, it hasbeen applied retrospectively at the end of the pro-gramme, so when you have just about handed overto the customer you reflect upon lessons learnt. Thelife cycle of our products varies between 2 and 7years. People tend to forget the issues if you leaveit for 2 years, so what we try to do is to get theselessons for processes in earlier. We try to introducea policy which says that if we have a life cycle, webreak the life cycle up into partitions and we try tohave a lessons learnt (meeting) at the end of eachphase which benefits future programmes or projectswhich are about to embark on that same process.But because of the diversity of the programmes. . . it depends on how clever you are in applyingwhat lessons you have learnt because only somethings are consistent. . . the environment in whichwe work is the same, the financial reporting, the


Fig. 4. The T-shaped learning landscape.


procedures we follow are all standard. So, we tryto enhance the process.

Except of the phenomenon of meeting overload, wefound that interviewees (and in particular project man-agers) valued lessons learnt meetings because theyoffered a context for reflecting on past actions andfor identifying what could be carried over to the nextphase or project. It is in the course of lessons learntmeetings that project managers and/or practitionershave the opportunity to reflect upon their actions, to tryto improve their understanding and, eventually, artic-ulate the relations between actions and performance.Project managers also found the preparation for thesemeetings extremely valuable for drawing lessons fromtheir previous actions.

Interestingly, we found that this firm does not linksuch meetings with its appraisal system. In otherwords, there are no incentive systems in terms ofpromotion or higher payments linked to the actualimplementation of lessons learnt meetings. “The onlymotivation is in improving the process, in improvingour software process in order to make more moneyif you like, to build it [the product] more cheaply.That is the only motivation and obviously if you canimprove the process. . . that is a motivation, isn’t it?. . . But there is no other incentive to do that.”

The articulation of project knowledge does not re-duce the importance of people-to-people knowledgetransfer, however. According to our interviewees, infact, inter-project learning still occurs via people.“When you go to work on another project, you learnyourself obviously, well it may not be obvious, butyou learn yourself what you did last time that wasn’tso good or something you did last time that was goodand you bring this to your next project. But I amnot sure that you transfer it to somebody else’s nextproject.” Knowledge codified (information) in projectdocuments does not get diffused properly. “We havedone the lessons learnt meeting, that goes to yourengineering manager and he makes sure its signedand then puts it in the file because it has been done.And although it is available, it is not actually used byany other programme.” Also, project knowledge getstransferred probably more easily via informal chan-nels. “If you know that another project is 6 monthsahead of you and they might have had the same sort ofproblems or it might be the same sort of programme,

you go and talk to them or picking the phone.” How-ever, “If you do not know the person concerned, youwill not go and talk to him”.

The second case is a firm operating in the de-fence industry. The firm is characterised by aproject-based organisational structure and it collabo-rates in multi-firm projects usually with internationalpartners. It is organised as a matrix with developmentdepartments on one axis and projects on the other.During recent years, the firm has changed its organi-sation from a sequentially partitioned organisation toone relying more on cross-functional and concurrentways of working. It is also characterised by a formalproject management process that, among other things,produces a document (a project history file) in orderto capture the lessons learnt. The firm has a processdevelopment director who is in charge of developingand maintaining the project management process.Project managers are supposed to prepare this docu-ment at the end of the project. This procedure has beenmandatory for only a year, however, interviewees ad-mitted that so far “we haven’t been very good at it”.Moreover, four times a year the company has a reviewmeeting with all programme managers to update theprocedures in the handbook. Another recent initiativehas been to invite an external reviewer to examineproject progress and organise de-briefing sessions.

The firm has tried to implement an Intranet to fa-cilitate the re-use of knowledge developed in previousprojects. One problem, however, is that much of theinformation cannot be stored in this Intranet becauseof its sensitive and classified nature of the information.There was also divergence among project practitionersand the process development director about the degreeto which the Intranet was implemented. Project man-agers do have access to lessons learnt reports, however:“If you start a new project and are a project manager,you can have access to a number of lessons learned re-ports. But it is not automatically distributed to projectmanagers: it is only on a ‘need-to-know’ basis”. Theinterviewees were very aware of the reinventing thewheel problem, but were coming to recognise that thisis inevitable. Some other important mechanisms forinter-project learning can be found elsewhere, how-ever. As stated by a project manager interviewed:

One can then ask the question: we have learned a lotin this project. But how do we get this information


to other projects? (1) Informal ways: people sittingclose to each other, discussions in the corridor, etc.(2) Our process development manager is invited toprogramme and team managers’ ‘gatherings’ wherethey sit down and discuss progress and lessonslearnt over 2–3 days at some location off the com-pany site. This takes place approximately oncea year. (3) Departments: since we have a matrixorganisation all technical responsibility rests withthe departments. This means that a department isresponsible when we start a new program, e.g. forestimating costs, assessing risks, etc.

While not an unusual practice, the interesting fea-ture of this firm was its adoption of what we mightcall an indirect incentive system for governing use ofthis information or the social networks that it involves.The firm rewards project manager in relation to per-formance indicators of projects (e.g. time and use ofresources) but does not assess project managers in re-lation to inter-project learning performance indicators.In other words, project managers are not assessed inrelation to the extent of knowledge re-used from pre-vious projects and/or project knowledge made avail-able to other projects (for instance via inter-projectmeetings). As highlighted by the project managers in-terviewed, however, it is paramount for the good exe-cution of the project to create an informal network ofcontacts by referring to other project managers or teammembers and getting access to their knowledge. Fur-ther evidence of this firm’s approach to learning be-tween projects is the initiative of one project managerin calling for a start-up meeting in the early phasesof a project. In the initial phase, the project and teammanagers invite someone from another recently com-pleted project to sit down with them and give them thebenefit of his experience. This was an informal meet-ing and not a requirement in the project managementhandbook of the firm.

4.3. The exploiter (or staircase) learning landscape

Firms already involved in the advanced de-velopment of ICT-based tools to support theirproject-to-project learning are characterised by whatwe define the exploiter (or staircase) learning land-scape. Their emphasis is on deliberate attempts tocodify and store knowledge developed during the exe-

cution of a project and document it so that it becomesmore easily accessible and exploitable for the rest ofthe organisation’s members. These firms are involvedin the advanced development of ICT-based tools totransfer and exploit project knowledge (Fig. 5).

The first case study focuses on a firm operating inthe power generation industry. In the mid-1990s, thefirm was taken over by a large conglomerate. The firmunder scrutiny is very much rooted in the local contextsince most, if not all, employees are from the samecity (or nearby villages) and, in any case, from thesame region. Although the acquirer is well known topursue a policy of cultural diversity emphasising lo-cal values as well as global values (the so-called glob-alisation), the take over entailed a number of signif-icant organisational and cultural changes in the firm.The acquirer is in fact also well known for instill-ing in the acquired firms its way of managing thebusiness.

Before the take-over, the firm was a functionallystructured organisation. With the take-over, the firmmoved first to a light matrix, then to a medium–heavymatrix form. The take-over also impacted on the or-ganisation of the business processes. A formal projectmanagement process has been put in place. Such pro-cess is software-based so that the project documenta-tion is organised in databases (available on the corpo-rate Intranet). The workload per person is formalisedand computer-based. The project process is structuredin disciplined, measurable, and repeatable phases (alsoknown as tollgates). In addition to the project manager,the project management team includes a support teamcomprised of a contract administrator, a completionmanager, a project controller and a technical secretary.With the take-over, the project manager emerged as akey figure in the firm.

There are four corporate-wide initiatives dealingwith knowledge management in this firm: globalisa-tion, service solutions, TQM, and e-business. The em-phasis of all these initiatives is on the business (andproject) processes. Process improvement is understoodas the main source of business performance improve-ment. A Quality Team is in charge of the revisions andupdates to the project management process. Revisionsand updates are accomplished with a top-down ap-proach. The Quality Team also carries out Root CauseAnalyses of problems that emerge during the executionof a project. Projects are assessed at in-depth meetings


Fig. 5. The staircase learning landscape.


held monthly to assess the status of the project usingspecific metrics (cost, resources, etc.).

Notwithstanding this emphasis on process, people-to-people communication is still a valuable and usefulway of exchanging ideas and transferring knowledge.For instance, although there is a database on previousexperience of personnel which is regularly updated,project staffing is done on the basis of personal con-tacts of the director of the specific division. Lessonslearnt meetings are ad hoc: i.e. only held when prob-lems arise. The project manager, the director of thedivision, or also the client can call them. These re-views are minuted, but the outcomes are not shared ormade available on the Intranet to other projects or di-visions. Culture is considered open, customer-orientedand productivity-oriented.

The second case study focuses on a firm engagedin the development of large software systems for bothcivilian and defence purposes. The firm is organisedaround divisions focused on a specific customer group,e.g. infrastructure, defence, aerospace. The divisionsare project-based. Software systems are developed inlarge projects. Some functional disciplines are ma-trixed, e.g. finance, human resources, and safety.

As a result of the acquisitions in the last decade thefirm operates in nearly 50 locations worldwide. Themulti-site feature of the firm prompted top manage-ment to try to leverage the knowledge base of eachsite to improve business processes across the differ-ent divisions. A corporate function labelled Engineer-ing and Technology, is in charge of developing, up-dating, and documenting the firm’s generic businessprocesses. These are available on the firm’s Intranet.Given the one-off characteristics of the projects car-ried out by the firm, each project is seen in termsof problematic issue, such as risk management, andextrapolated and defined generic policies and proce-dures that can then be tailored to the project’s spe-cific needs. The firm has developed a generic processframework that encompasses all its business processes.The process framework is organised in five-tiers. Thefirst three tiers contain organisational-level policiesand procedures, which are stored on the corporateIntranet. The organisational policies and proceduresare a general road map that broadly defines “what todo”. The last two tiers contain the project-level poli-cies and procedures and are stored on the specificproject’s web page or disk space (with restricted ac-

cess). The corporate Intranet contains a lessons learntdatabase.

Project policies and procedures are based on or-ganisational ones and tailored to the project. Thetailoring is carried out according to the guidelinescontained in a guide provided by the firm that de-fines what, how, whether and why to tailor. As oneinterviewee said “The generic feature of the processas enforced by the firm is essential for the complex-ity of the project manager job (in terms of control,costs, resources) as it is the tailoring for the specificproject”. Organisational policies and procedures areupdated regularly (every 18–24 months). Anyone inthe firm can initiate a change process. A new pro-cess is reviewed and put on the web. Before beingenforced, however, the new process goes througha certification process composed of four steps: (1)project sample (raw process); (2) generic projectsample; (3) certified project sample; (4) standardproject sample (process enforced). At the projectlevel, the policies and procedures are changed morefrequently.

The company has institutionalised a number of dif-ferent types of project reviews that are embedded andformalised in the business process. These formal re-views can be categorised in two main groups. (1)Reviews led by auditors external to the project. Theaim of these reviews is to assess performance sta-tus, risks, costs, schedule of the project and lessonslearned from previous projects. The diffusion of theoutcome of these reviews is limited to those respon-sible for conducting the reviews and to the corporatemanagers who assess the project’s merits. (2) Reviewsled by a project team member (usually the projectmanager). These are software reviews, design reviews,customer reviews (held monthly), and quality reviews(held quarterly). The outcomes of these reviews aredocumented and stored electronically and are avail-able only to the project team members only.

Although project reviews contain a learning el-ement and the firm encourages informal ways ofknowledge sharing via seminars and staff meetings,the procedure-oriented culture permeates the firmand is the major driver for organisational and projectperformance improvements. According to one direc-tor interviewed, “A regular update is the best wayto improve the business process, much better andmore efficient than lessons learned databases, because


nobody looks at them, my line is ‘You haven’t learnedthe lessons, change the process’ ”.

Informal ways of transferring knowledge (and par-ticularly people-to-people communication) are stillimportant. In fact, there are informal project re-views. The project manager or other team memberson specific issues may call these reviews. A divi-sional director meets with his project managers oncea month to help them re-use knowledge developed ineach project. Also, inter-project informal reviews areheld in the form of staff meetings between all projectmanagers. As a consequence of this an informal net-work of project managers has been created. Thereis also a project manager electronic symposium ondelivery mechanisms for a web-based programme.Informal ways of knowledge sharing have been in-stitutionalised via weeklybrown bagseminars givenby internal or external speakers on specific (usuallytechnical) subjects.

5. Discussion and conclusions

Nelson and Winter (1982) suggested that acapability-based theory of the firm should scruti-nise organisational routines. From an evolutionaryperspective, such routines could be treated as thequasi-genetic traits of a firm, causing persistent dif-ferentials in firm behaviour and performance. In thisstudy, we have been particularly concerned with un-derstanding the project-based firm as a population ofprojects, where specific project traits may get trans-ferred via various mechanisms from one project toanother. In particular, this paper proposed a frame-work to analyse and interpret firms’ approaches toproject-to-project learning. We discussed the impor-tance of the process dimension of firms’ approaches toproject-to-project learning. Following Zollo and Win-ter (2001), we have argued that it is not sufficient toconsider knowledge codification only as an outcome.

The literature on knowledge codification is charac-terised by a tendency to think that the costs of codifi-cation activities are justified by their outcomes ratherthan by the cognitive implications of the codificationprocess as such (Zollo and Winter, 2001). In the em-pirical findings reported in this study, it seems thatit is not only researchers that have focused primarilyon the outcomes of knowledge codification, however.

Project-based firms also seemed to focus their effortsonoutcomesrather than on theprocessof codificationin developing technical devices and organisationalmechanisms for learning between projects. Interest-ingly, in the firms we studied we found little evidenceof direct incentives, such as monetary benefits, as-sociated with the codification of knowledge. Rather,the codification of knowledge into reports, minutes,lessons learned, etc. is based on a presumption ofgood behaviour among members in the organisation.In the light of the emphasis put on incentive issuesin research into internal knowledge transfer throughcodification in organisations (Cowan and Foray, 1997;Eisenhardt and Santos, 2001), these findings indicatethat determining the empirical properties of incen-tives for knowledge codification may be difficult.That is, although in a theoretical discourse it may bepossible to distinguish what incentives should prevailfor knowledge codification by determining its costsand benefits, the way such incentives work practicallyin project-based firms may be of a character that issubtle and difficult to measure, and assess.

Appreciating the knowledge codification process asa cognitive effort led us to adopt the model suggestedby Zollo and Winter (2001) based on the processesof experience accumulation, knowledge articulationand knowledge codification. By relating these threeprocesses to the individual, project and organisationallevels we developed a 3× 3 matrix that enabled us tocategorise a variety of mechanisms that project-basedfirms use. We defined a firm’slearning landscapeas the mix of project-to-project learning mechanismsadopted and implemented. The concept of a learninglandscape reflects the multidimensional nature of afirm’s approach to project-to-project learning. Basedon the empirical evidence gathered during our fieldstudy, we identified three main types of learning land-scape, namely theexplorer landscape(or L-shaped),the navigator landscape(or T-shaped), and theex-ploiter landscape(or staircase). Each of these learninglandscapes is characterised by a different emphasison specific inter-project learning mechanisms, whichin turn are based on different learning processes(experience accumulation, knowledge articulation andknowledge codification), both at the individual andorganisational levels.

Firms that rely primarily on people-embeddedknowledge are characterised by the explorer


landscape. These firms emphasise experience accu-mulation processes and knowledge transfer throughpeople-to-people communication. The navigator land-scape characterises firms that started implementingmechanisms for project-to-project learning based onthe knowledge articulation process. Firms alreadyinvolved in the advanced development of ICT-basedtools to support their project-to-project learning arecharacterised by the exploiter learning landscape.These results point to two interesting features con-cerning the learning dynamics of firms and the de-velopment of the firm’s organisational capabilities.First, despite facing quite similar industry charac-teristics (for one thing they were all operating inso-called CoPS industries), we observed substantialvariation with respect to the mechanisms used for thetransfer of vital knowledge from project to project.Second, notwithstanding the temporary and one-offnature of the tasks performed by project-based firms,these firms do develop a set of routines that definehow to approach projects. The learning landscapeas conceptualised in this paper represents an em-pirical example of the set of routines employed byproject-based firms to manage inter-project learning.These results also have a more pragmatic implicationfor managers, that is, a firm’s knowledge manage-ment strategy and the tools supporting it should becarefully selected and customised to the specific firmcontext.

How then do our findings fit into the codificationdebate? Cowan et al. (2000) proposed that the artic-ulation and the communication associated with suchprocesses presuppose some kind of codified knowl-edge. The properties of codified knowledge should,thus predominate over those of articulated knowl-edge. While we acknowledge that the outcomes andeconomic benefits of articulation and codification toa large extent are similar, the framework proposedby Zollo and Winter (2001) provides insight into aprocess in which articulation is an intermediate stepin the codification of knowledge. Based on our em-pirical investigation of six firms trying to establishmechanisms for inter-project learning we observe thatthe process through which knowledge is accumulatedinto tacit, articulated and codified knowledge is highlycomplementary. Informal processes are sometimesimportant; some knowledge is really hard to codify,but articulation and codification of knowledge also

seem to help firms in their pursuit of better knowl-edge about why some projects succeed and othersfail.

One suggestion for further research is to investigatethe contexts in which different inter-project learningmechanisms are useful and effective. This calls forsome kind of contingency analysis where variables,such as, e.g. size, strategy, task complexity, uncer-tainty, rate of technological change, market conditions,etc. are related to the effectiveness of inter-projectlearning mechanisms such as those discussed here.While authors such as Zander and Kogut (1995), stud-ied the impact of codification on knowledge transfer,we have not focused on this issue. This delimitationof this study relates to our explicit interest in the pro-cess, rather than the outcome, of knowledge codifica-tion. Hence, we, at least at this stage, have little to sayabout the efficiency properties of codification strate-gies for knowledge transfer between projects. More-over, as pointed out by Eisenhardt and Santos (2001),from a strategic perspective the relationship betweenlearning effectiveness and overall firm performancehas yet to be established empirically. Our empiricalfindings do, however, square with what Eisenhardt andSantos (2001) found in their review of empirical stud-ies of internal knowledge transfer: (1) the relationshipbetween the sender and recipient in the knowledgetransfer processes is paramount; (2) integrative mech-anisms, both formal and less formal, facilitate suchlearning. One interesting avenue to explore, therefore,is the emergence of various communities of practicesand how such communities contribute to, or impair,more formal or technology-based initiatives for learn-ing between projects (e.g. our exploiter landscape)(Brown and Duguid, 1991, 2000; Lave and Wenger,1991; Wenger, 2000).

Acknowledgements

The paper has greatly benefited from discussionswith and comments from Tim Brady, Anna Grandori,Mike Hobday, Keith Pavitt, and Edward Steinmueller.We are grateful to the Editor of this Special Issueand one anonymous reviewer. We would like to thankcompanies’ managers and engineers who gave theirvaluable time for interviews and comments and theOffice of the Scientific Attaché of the British Embassy


in Washington, DC for its help in setting up interviewsin the USA. Earlier drafts of the paper were presentedat workshops and conferences held at the Universityof Leicester, Victoria University of Wellington, CNRS(Nice), and Imperial College of Science and Technol-ogy. Conference participants offered invaluable sug-gestions. The authors gratefully acknowledge financialsupport from the British Engineering, Physical andScience Research Council (EPSRC) and the SwedishResearch Council. Fredrik Tell wants to express spe-cial gratitude to Handelsbanken for a scholarship fromJan Wallander and Tom Hedelius’ Foundation for So-cial Science, which facilitated his visiting fellowshipat SPRU, University of Sussex, during 2001. Usualdisclaimers apply. The names of the authors are or-dered alphabetically.

Appendix A. Research method

The paper draws on qualitative data collected dur-ing a field study of six firms. The study is part of alarger 3-year research project designed to explore themanagement of inter-project learning in a convenientsample of 50 project-based firms developing CoPS.We selected six case studies as the most representa-tive to highlight the relevant features of the identifiedlearning landscapes. The selected firms were based inItaly, Sweden, the UK, and the USA and belonged tothe software, aerospace, defence, flight simulation andpower generation sectors.

We conducted three interviews per firm. Intervie-wees covered a corporate manager, a project manager,and a practitioner. This three-layer interview designwas explicitly planned to gather data on the same top-ics from different hierarchical levels in the firm. Also,we chose to interview representatives at three differ-ent hierarchical levels in order to analyse and identifythe typology of learning mechanisms that charac-terised each level. The analysis of the typology of thelearning mechanisms at each level enabled an assess-ment of thelearning landscapethat characterised thefirms. We did not develop performance measure toassess the effectiveness of the different mechanismsadopted by the firm for inter-project learning. We re-lied therefore on interviewees’ perceptions (Lawrenceand Lorsch, 1967). Interviews were semi-structuredand lasted about 90 min.

Data from these interviews were elaborated toproduce mini case studies of the firms involved inthe study enabling the identification, evaluation andmatching of patterns as they emerged fromwithinindividual cases(Eisenhardt, 1989, p. 540). We thencross-compared the cases to identify common pat-terns. This tactic goes under the name ofpatternmatching(Yin, 1994, p. 106), “Such a logic. . . com-pares an empirically-based pattern with a predictedone (or with several alternative predictions)”. The useof this tactic assumes the development of rival ex-planations involving mutually exclusive independentvariables. As Yin (1994, p. 108) argued, “the presenceof certain independent variables (predicted by oneexplanation) precludes the presence of other indepen-dent variables (predicted by a rival explanation)”. Theuse of the pattern-matching tactic strengthens thein-ternal validityof the research. Also, as we followed amultiple case research design, we adopted a strategyof replication in multiple empirical settings in orderto establishexternal validity(Yin, 1994).

References

Alvesson M., 1995. Management of Knowledge-IntensiveCompanies. De Gruyter, New York.

Ancori, B., Bureth, A., Cohendet, P., 2000. The economics ofknowledge: the debate about codification and tacit knowledge.Industrial and Corporate Change 9 (2), 255–287.

Argyris, C., Schön, D.A., 1978. Organisational Learning: A Theoryof Action Perspective. Addison-Wesley, Reading, MA.

Baumard, P., 1999. Tacit Knowledge in Organisations. Sage,London.

Boisot, M., 1995. Information Space: A Framework for Learningin Organisations. Routledge, London.

Boisot, M., 1998. Knowledge Assets: Securing CompetitiveAdvantage in the Information Economy. Oxford UniversityPress, Oxford.

Brown, J.S., Duguid, P., 1991. Organisational learning andcommunities-of-practice: toward a unified view of working,learning, and innovation. Organisation Science 2 (1), 40–57.

Brown, J.S., Duguid, P., 2000. The Social Life of Information.Harvard Business School Press, Boston, MA.

Brusoni, S., Prencipe, A., Salter, A., 1998. Mapping and measuringinnovation in project-based firms, CoPS Working Paper No. 46,SPRU, University of Sussex.

Cohen, M., Burkhart, R., Dosi, G., Egidi, M., Marengo, L.,Warglien, M., Winter, S.G., 1996. Routines and other recurringaction patterns of organisations. Industrial and CorporateChange 5 (3), 653–698.

Cowan, R., Foray, D., 1997. The economics of codification andthe diffusion of knowledge. Industrial and Corporate Change6 (3), 595–622.


Cowan, R., David, P., Foray, D., 2000. The explicit economics ofknowledge codification and tacititness. Industrial and CorporateChange 9 (2), 211–253.

Davenport T.H., L. Prusak, 1998. Working Knowledge: HowOrganisations Manage What they Know. Harvard BusinessSchool Press, Boston, MA.

Davies, A., Brady, T., 2000. Organisational capabilities andlearning in complex product systems: towards repeatablesolutions. Research Policy 29, 931–953.

DeFillippi, R., Arthur, M., 1998. Paradox in project-basedenterprise: the case of film making. California ManagementReview 40 (2), 125–139.

Dierickx, I., Cool, K., 1989. Asset stock accumulation andsustainability of competitive advantage. Management Science35 (12), 1504–1513.

Dosi, G., L. Marengo, 1994. Some elements of an evolutionarytheory of organisational competences. In: England, R.W.(Ed.), Evolutionary Concepts in Contemporary Economics. TheUniversity of Michigan Press, Ann Arbor, MI.

Dosi, G., Marengo, L., Fagiolo, G., 1996. Learning in EvolutionaryEnvironments. IIASA Working Paper No. 96–124.

Dosi, G., Nelson, R.R., Winter, S.G. (Eds.) 2000. The Nature andDynamics of Organisational Capabilities. Oxford UniversityPress, Oxford.

Dretske, F.I., 1981. Knowledge and the Flow of Information. BasilBlackwell, Oxford.

Eisenhardt, K.M., 1989. Building theories from case studyresearch. Academy of Management Review 14 (4), 532–550.

Eisenhardt, K.M., Santos, F.M., 2001. Knowledge-based view:a new theory of strategy? In: Pettigrew, A., Thomas, H.,Whittington, R. (Eds.), Handbook of Strategy and Management.Sage, London.

Foray, D., Steinmueller, W.E., 2001. Replication of routine,the domestication of tacit knowledge and the economics ofinscription technology: a brave new world? Paper presented atthe DRUID Nelson and Winter Conference, Aalborg, Denmark.

Fransman, M., 1994. Information, knowledge, vision and theoriesof the firm. Industrial and Corporate Change 3 (3), 713–757.

Galbraith, J., 1969. Organisation design: an information processingview. Working Paper No. 425–69. Alfred P. Sloan School ofManagement, MIT Press, Boston, MA.

Gann, D., Salter, A., 1998. Learning and innovation managementin project-based, service-enhanced firms. International Journalof Innovation Management 2 (4), 431–454.

Gavetti, G., Levinthal, D., 2000. Looking forward and lookingbackward: cognitive and experiential search. AdministrativeScience Quarterly 45, 113–137.

Grand, S., Von Krogh, G., 2000. Justification in knowledgecreation: dominant logic in management discourses. In: vonKrogh, G., Nonaka, I., Nishigushi, T. (Eds.), KnowledgeCreation: A Source of Value. Macmillan, New York.

Grant, R.M., 1996. Toward a knowledge-based theory of thefirm. Strategic Management Journal 17 (Winter Special Issue),109–122.

Hansen, M.T., Nohria, N., Tierney, T., 1999. What’s your strategyfor managing knowledge? Harvard Business Review, 106–116.

Hobday, M., 1998. Product complexity, innovation and industrialorganisation. Research Policy 26, 689–710.

Hobday, M., 2000. The project-based organisation: an ideal formfor managing complex products and systems? Research Policy29, 871–893.

Kogut, B., Zander, U., 1992. Knowledge of the firm, combinativecapabilities, and the replication of technology. OrganisationScience 3 (3), 383–397.

Lave,J., Wenger, E., 1991. Situated Learning: Legitimate PeripheralParticipation. Cambridge University Press, Cambridge, MA.

Lawrence, P.R., Lorsch, J.W., 1967. Organisation and Environment:Managing Differentiation and Integration. Harvard UniversityPress, Cambridge, MA.

Levinthal, D.A., March, J.G., 1993. The myopia of learning.Strategic Management Journal 14 (Winter Special Issue),95–112.

Levitt, B., March, J.G., 1988. Organisational learning. AnnualReview of Sociology 14, 319–340.

Lindkvist, L., Söderlund, J., Tell, F., 1998. Managing productdevelopment projects — on the significance of fountains anddeadlines. Organisation Studies 19 (6), 931–951.

March, J.G., Olsen, J.P., 1989. Rediscovering Institutions: TheOrganisational Basis of Politics. Free Press, New York.

March, J.G., Sproull, L.G., Tamuz, M., 1991. Learning fromsamples of one or fewer. Organisation Science 2 (1), 1–13.

Nelson, R.R., Winter, S.G., 1982. An Evolutionary Theory ofEconomic Change. The Belknap Press of Harvard UniversityPress, Cambridge, MA.

Nightingale, P., 2001. If Nelson and Winter are only half rightabout tacit knowledge, which half? A reply to David, Forayand Cowan. Paper presented at the DRUID Nelson and WinterConference, Aalborg, Denmark.

Nonaka, I., Takeuchi, H., 1995. The Knowledge Creating Company.Oxford University Press, Oxford.

Paoli, M., Prencipe, A., 1999. The role of knowledge basesin complex systems: some empirical evidence from theaero-engine industry. Journal of Management and Governance3, 137–160.

Penrose, E., 1952. Biological analogies in the theory of the firm.American Economic Review XLII (5), 804–819.

Penrose, E., 1959. The Theory of the Growth of the Firm. OxfordUniversity Press, Oxford.

Polanyi, M., 1962. Personal Knowledge: Towards a Post-criticalPhilosophy. Routledge & Kegan Paul, London.

Prencipe, A., 2000. Breadth and depth of technological capabilitiesin CoPS: the case of the aircraft engine control system. ResearchPolicy 29, 895–911.

Schön, D.A., 1983. The Reflective Practitioner: How ProfessionalsThink in Action. Basic Books, New York.

Schön, D.A., 1987. Educating the Reflective Practitioner.Jossey-Bass, San Francisco, CA.

Spender, J.-C., 1996. Making knowledge the basis of a dynamictheory of the firm. Strategic Management Journal 17 (WinterSpecial Issue), 45–62.

Steinmueller, W.E., 2000. Will new information and communi-cation technologies improve the codification of knowledge?Industrial and Corporate Change 9 (2), 361–376.


Tell, F., 1997. Knowledge and justification: exploring theknowledge-based firm. Dissertations from IMIE, LinköpingUniversity, Sweden.

Tell, F., 2000. What do organisations know? Different types ofknowledge and the dynamics of justification contexts. Paperpresented at the EGOS Colloquium, Helsinki, Finland.

Tell, F., Söderlund, J., 2001. Lärande mellan projekt. In:Berggren, C., Lindkvist, L. (Eds.), Projekt: Organisation forMålorientering och Lärande, Studentlitteratur.

Warglien, M., 2000. The evolution of competences in a populationof projects: a case study. University of Venice, Mimeo, Italy.

Weick, K.E., 1995. Sensemaking in Organisations. Sage,London.

Wenger, E., 2000. Communities of practice and social learningsystems. Organisation 7 (2), 225–246.

Winch, G., Schneider, E., 1993. Managing the knowledge-basedorganisation: the case of architectural practice. Journal ofManagement Studies 30 (6), 923–937.

Zollo, M., Winter, S.G., 2001. Deliberate learning and the evolutionof dynamic capabilities. Organisation Science, in press.

Yin, R.K., 1994. Case Study Research: Design and Methods. Sage,London.

Inter-project learning: processes and outcomes of...

Documents

Transcript of Inter-project learning: processes and outcomes of...