Sie sind auf Seite 1von 16

Leveraging on systemic learning to manage the early phases of product innovation projects

Roberto Verganti

Dipartimento di Economia e Produzione, Politecnico di Milano, Piazza Leonardo da Vinci 32, 20133 Milano, Italy

A number of studies have pointed out the

importance of the early phases of new pro- duct development projects. In fact, these phases (addressed in the literature with

different names, such as pre-project activi- ties, concept generation, product planning, idea generation, investigation, product definition) are concerned with a number of critical decisions that have great impact on the performance of product development. Any fault occurring in these early phases in understanding the market needs, in choos- ing the product architecture and technology and in defining the product specifications would eventually deteriorate the innovation process, since adjustments in later stages imply reworks that are costly and time consuming. Although the relevance of early project phases has been empirically verified

in the literature, the mechanisms that allow

these phases to be properly managed are still largely unexplored. This paper investi- gates the articulated and coherent set of methods, organizational mechanisms and behavioural patterns that successful com- panies adopt to manage concept generation and product planning. Inferences are based on a field research concerning 19 in-depth case studies of Italian and Swedish com- panies in the vehicles, helicopters and white goods industries. The paper supports findings of other studies concerning the importance of teamworking and communi- cation. However, teamworking emerges as a necessary, but not a sufficient mechanism. Systemic learning from past experiences is the real keystone toward an effective management of the early phases of product development processes.

1. Introduction

Most authors maintain that the success of product innovation is rooted in the early phases of the product life cycle (Hayes et al., 1988; Gupta and Wilemon, 1990; Rosenthal, 1992; Wheelwright and Clark 1992a; Clark and Wheelwright, 1993; Brown and Eisen- hardt, 1995). These are the phases where the product concept is generated, the product specifications are defined and the basic project decisions are taken, concerning the product architecture, the major components, the pro- cess technology and the project organization. These phases are generally denoted with different names, such as pre-project activities, concept generation, product planning, idea generation, investigation, product definition (these terms will be used interchangeably as synonymous in the paper). The importance of pre-project activities has been empirically validated in well-known research studies based on extensive surveys (de Brentani, 1991; Cooper, 1994; Cooper and Kleinsch- midt, 1994, 1995). Also, the new organiz- ational paradigms of product development that have emerged in the last few years, give prominence to the early development phases. One of the basic principles of these paradigms is that the impact of design choices on market requirements, design feasibility, ease of manufacture, usability, reliability, maintain- ability, recyclability, and so on, should be analysed as early as possible in the product development process; this will prevent the need for late reworks that are costly and time consuming (Trygg, 1991). Authors assert that the anticipated analysis of the implications due to a new product concept and specification

R&D Management 27, 4, 1997. © Blackwell Publishers Ltd, 1997. Published by Blackwell Publishers Ltd, 108 Cowley Road, Oxford OX4 1JF, UK and 350 Main Street, Malden, MA 02148,USA.

377

Roberto Verganti

may be successful only if carried out at a cross-functional level, i.e. through multi- functional and multi-disciplinary teams (e.g., Hauser and Clausing, 1988; Gupta and Wilemon, 1990; Clark and Fujimoto, 1991; Fujimoto, 1993; Bacon et al., 1994). Clark and Fujimoto, for example, in their study of product innovation in the world auto industry, describe the preliminary exchange of informa- tion with downstream phases as the more complete, albeit complex, coordinating mechanism adopted by successful Japanese companies. Teamworking and early involve- ment of key development roles are therefore suggested as the main organizational princi- ples in concept generation and product planning. In particular, studies focused on the benefits provided by early involvement of manufacturing (Trygg, 1991; Jurgens, 1995; Ettlie, 1996; Haddad, 1997) and of major suppliers (Clark, 1989; Dyer and Ouchi 1993; Kamath and Liker, 1994; Liker et al., 1996). New managerial techniques have been also proposed to support teamworking in the early phases of product development, such as Qual- ity Function Deployment (Hauser and Clausing, 1988) and Life Cycle Costing (Blanchard, 1979). However, although com- panies are increasingly adopting multi- functional teams in concept generation and

product planning and dedicate great efforts to anticipate downstream information, they still face substantial problems. This means that reasons for failures in managing early phases cannot be simply traced back to a lack of communication between different departments or to time pressure that induces to skip product-planning activities. Hence, the need emerges to further understand the complex, articulated and coherent set of managerial criteria, organizational mechanisms and behavioural patterns that allow downstream information to be properly anticipated and that transform early involvement in a successful practice. The purpose of this paper is to identify the principles and mechanisms that appear to be effective when managing the early phases of product development projects. It reports the results of a field research based on nineteen in-depth case studies of companies operating in three different industries: vehicles (7 companies), white-goods (10 companies), and helicopters (2 companies). Companies are located in Italy (13 cases) and in Sweden (6 cases). Table 1 concisely illustrates the companies’ business area, their size and the content of innovation in their product range. The research method entailed to collect both qualitative and quantitative information at

Table 1.

Outline of the 19 companies involved in the case studies.

 

Revenues

New product lines launched on the market between 1992 and 1994

Percentage of 1994 revenues due to new product lines

(1994 –

 

Employees

million

Main products

Country

(1994)

dollars)

1.

Cars

Italy

125 000

30 000

10

75

2.

Cars

Sweden

7000

2475

1

60

3.

Trucks

Italy

32 000

6875

3

40

4.

Trucks

Sweden

500

122

4

60

5.

Trucks

Sweden

20 000

5250

4

40

6.

Buses

Sweden

500

130

4

20

7.

Tractors

Italy

2500

938

3

80

8.

Large cooking devices

Italy

1800

250

1

90

9.

Cooling, washing and cooking dev.

Italy

530

113

3

10

10.

Washing devices

Italy

350

82

3

15

11.

Cooling, washing and cooking dev.

Italy

1630

325

16

30

12.

Cooling, washing and cooking dev.

Italy

12 000

2373

11

50

13.

Large vacuum cleaners

Italy

62

13

2

50

14.

Vacuum cleaners

Italy

110

28

2

20

15.

Vacuum cleaners

Sweden

700

112

1

30

16.

Large sewing machines

Italy

596

64

1

30

17.

Sewing machines

Sweden

400

63

3

60

18.

Small helicopters

Italy

100

16

2

100

19.

Helicopters

Italy

5500

562

3

30

378 R&D Management 27, 4, 1997

© Blackwell Publishers Ltd 1997

Systemic learning in the early phases of product innovation

the company level and at a single project level (consisting of the most recent and finished innovation project of a product line or platform). Qualitative data allowed an understanding of the managerial approach and organizational mechanisms implemented in concept generation and product planning by each company. They have been gathered by interviewing four managers in each com- pany (marketing, research and development or product engineering, process engineering or manufacturing, project manager or plat- form manager). Quantitative data supported comparison of different approaches and investigation of their effectiveness. These data have been collected through a question- naire with more than 200 variables and structured managers’ subjective information. In the next sections, first the reason for the importance of pre-project activities is main- tained and the major problems and difficulties encountered when managing these activities are introduced. Next, section 3 discusses

successful approaches and mechanisms to manage pre-project activities, on the basis of the lessons drawn from observations of the 19 case studies. Finally, section 4 focuses on what appears as the most interesting and effective driver for managing the early phases of product development: the exploitation of past experiences and systemic learning. This latter detailed discussion will benefit by the quantitative and structured information col- lected through the questionnaire.

2. The early phases of product development projects: significance and management issues

The reason for the importance generally ascribed to early product development phases may be traced back to the structural nature of the product life cycle. As Figure 1 shows, the life cycle of a product may be seen as a pro- cess with different stages and several internal

as a pro- cess with different stages and several internal Figure 1. The information structure of

Figure 1.

The information structure of the product life cycle.

© Blackwell Publishers Ltd 1997

R&D Management 27, 4, 1997 379

Roberto Verganti

and external interactions (see also Clark and Fujimoto, 1990; Fujimoto, 1993). The main stages are product development (which articu- lates in concept generation and product planning — i.e., what we consider as early phases — product design and process design), production, consumption, disposal, and enhancement or evolution into a next generation product. These stages and phases are linked by reciprocal interdependencies (Thompson, 1967; de Weerd-Nederhof et al., 1994). First, reciprocal interdependencies characterize the internal activities of the pro- duct development process. For example: a new product concept must fit with available and prospected product technologies; a design solution must be feasible and should account for the constraints coming from process design; the different parts and components of the product must fit each other consistently. In other words, the new product concept, the new product package and the new production process mutually interact and should be con- tinuously integrated each other (Slack et al., 1995). For this reason, most authors look at product development as a series of problem solving cycles occurring among its internal activities (see for example Clark and Fujimoto 1991; Souder and Monaert, 1992). Second, reciprocal interdependencies occur between the product development process and the subsequent production and consumption processes. In fact, in order to develop a new product and to launch it into the market, customer expectations have to be anticipated and constraints due to manufacturability have to be accounted for; in this respect, Clark and Fujimoto (1991) represent the product development process as a simulation of con- sumption and production: developing a new solution means forecasting the future con- straints and opportunities that will arise during the life cycle of the product. Finally, recipro- cal interdependencies occur also with next generation products. In fact, new ideas and solutions may be carried over, especially when the current project develops a new platform to be re-used in other derivative products (Wheelwright and Sasser, 1989; Wheelwright and Clark, 1992b; De Maio et al., 1994; Corso et al., 1996). Constraints and opportunities coming from future production and consumption of the latter should therefore be anticipated in the former.

380 R&D Management 27, 4, 1997

Managing product innovation entails to effectively handle these reciprocal interde- pendencies between new solutions designed in the early phases and constraints oppor- tunities arising in the later phases of the product life cycle. This is a complex task, since downstream information is not com- pletely known when developing a new solution. In fact, on the one hand novelty of the product and on the other hand the con- tinuous changes in the market, technological, competitive, and regulatory environments, together with the company’s strategy, create new constraints and new opportunities that are unlikely to be predicted at the outset of a project. Uncertainty about downstream infor- mation therefore creates severe problems and eventually compromises the product develop- ment performance, since:

solutions have to be changed later to deal with unexpected events. This is typical of engineering changes due to unfeasibility or poor manufacturability emerging during pilot production. These late changes entail costly reworks and increase the time to market of the product; downstream phases have to put additional work in the project in order to handle incompatible solutions (e.g., special fixtures have to be developed in the pro- cess design phase, since product drawings did not contemplate constraints due to production technologies). This again implies higher development costs and delays; quality of the new product (defined as its fit with the production and consumption processes) is impaired, e.g. the actual market needs are not satisfied or new tech- nological opportunities are not incorpor- ated into the product.

In order to deal with reciprocal interdepen- dencies in the product life cycle companies have two different alternatives (see Figure 2):

Feed-back planning (or reactive appro- ach). The product is developed without intensively anticipating information in the early phases. Reasons behind this approach lay on the high uncertainty of downstream information when examined in the upstream phases of product development. Since this uncertainty decreases as the

© Blackwell Publishers Ltd 1997

Systemic learning in the early phases of product innovation

Systemic learning in the early phases of product innovation Figure 2. of the product life cycle.

Figure 2.

of the product life cycle.

Uncertainty and detrimental impacts of reciprocal interdependencies in the different stages

production and consumption processes become closer (see the dotted line in Figure 2), changes to design solutions due to unexpected constraints or opportunities are introduced later in the project, when downstream information is available. This approach is effective provided late correc- tive actions (i.e. engineering changes or additional work in downstream phases) have a minor impact on the product development performance. However, this is seldom true: as studies on Project Man- agement and Product Development unanimously assert (e.g., Meredith and Mantel, 1989; Hayes et al., 1988; Trygg, 1991), the later the downstream con- straints emerge, the higher the time and cost due to reworks and engineering changes (see the continuous line in Figure 2). In fact, late changes entail a greater number of already developed solutions to be re-designed, including also those most costly and time consuming such as detailed product design and process design. In some cases, unexpected technological opportunities or flaws in interpreting the market needs arise when the product is already on the market, and it is therefore too late to introduce any change; Feed-forward planning (or proactive

© Blackwell Publishers Ltd 1997

approach or front loading, Fujimoto, 1997). Information is anticipated as early as possible in the product development process, so that new solutions generated in the early phases already account for future constraints and opportunities. The manage- ment focus and attention is therefore concentrated in concept generation and product planning where, as Figure 2 shows, there is the highest ability to influence the outcome of the project. Hence, great efforts are devoted to prop- erly define the product concept and specifications in order to reduce possible problems and reworks in later phases. The drawback of feed-forward planning is the uncertainty that affects the early phases. This approach is effective provided a com- pany has great capabilities to anticipate highly uncertain information, otherwise early analysis of future constraints is a sterile exercise: time at the beginning of the project is wasted while reworks in later stages are not avoided.

Both reactive and proactive capabilities are necessary. However, studies previously reported in the introductory section, which demonstrate a strong relation between deci- sions taken in pre-project phases and success

R&D Management 27, 4, 1997 381

Roberto Verganti

of innovation, support the superiority of feed-forward planning. Indeed, this approach is also advocated by the new organizational paradigms of product development. In this perspective, feed-back planning is consid- ered as a compensation to feed-forward. It entails that any initially unforeseen event should be handled later by reacting at low cost and time. Anyway, since later changes have dramatic impacts on product develop- ment performance, this should be avoided as much as possible through a strong proactive orientation. Nevertheless, although there is a wide agreement on the effectiveness of feed-for- ward planning, this approach is quite complex. In fact, uncertainty about down- stream information is remarkable in the early phases of product development. This is a major problem especially in highly innova- tive projects or in fairly mutable environ- ments in terms of customer needs, competition, technologies and regulations. Implementing successfully the feed-forward approach therefore implies a need to under- stand which mechanisms allow the handling of uncertainty at the outset of a project and simultaneously boost innovation.

3. The principles of feed-forward planning

Managing the early phases of product development in a proactive perspective calls for an articulated and coherent set of organiz- ational mechanisms, managerial principles and methods. In particular, lessons learned from the, 19 case studies showed that 6 principles make feed-forward planning effective: teamworking and communication, harmonized objectives, encouraged and supported proactive thinking, integration with detailed design, planned flexibility, systemic learning. They are examined in the following sections.

3.1. Teamworking and communication

Anticipation of downstream constraints and opportunities entails the collection of a large amount of information. Bacon et al. (1994), for example, consider as information to be anticipated the customer and user needs, the

382 R&D Management 27, 4, 1997

competitive product offerings, the tech- nological risks and opportunities and the regulatory environment into which the pro- duct will be delivered. In some companies

this information is collected and analysed by

a specialized department (e.g., the product

planning department). In other cases, the product manager or platform manager are responsible for specifying the product concept; sometimes, a single function is involved in the early phases of product development, such as marketing (in market- driven companies) or research and develop- ment (in high-technology industries). However, a number of studies demonstrate that the most successful criterion for collect- ing and analysing such an articulated set of information is the early involvement of all the major departments which have direct contact with future constraints and oppor- tunities: marketing (and in some cases lead customers), research, product engineering, process engineering, manufacturing, pur- chasing (and or major suppliers). The effectiveness of teamworking and early involvement also emerges in this study (see section 4 for quantitative evidence). Indeed, all companies except one allocate concept generation and product planning to a multi- functional team. This preliminary cross- functional exchange of information among the main actors of product development provides four major benefits: first, it enlarges the knowledge base available in the early phases of product development and therefore reduces the uncertainty on future constraints and opportunities; second, it assures align- ment of product concept with company strategy and functional strategies (i.e., the technology, marketing and manufacturing strategies); third, it allows downstream phases (e.g., Process Engineering) to start earlier with some preliminary activities, thus encouraging parallel design phases; fourth, since the owners of downstream constraints are involved in concept generation, then their commitment to upstream decisions is fostered.

3.2. Harmonized objectives

In most case studies we noted that, although

different departments were gathered in a pre- project cross-functional team, this did not

© Blackwell Publishers Ltd 1997

Systemic learning in the early phases of product innovation

Without proper guidance and support, most pre-project teams eventually only anticipate superficially. Then, given the uncertainty on future implications of an early decision, they quickly proceed straight into detailed design, where the necessary information eventually emerges. Proactive thinking should on the contrary be stimulated, nurtured and sup- ported as much as possible in the early phases. The problem is how to encourage uncertain information to be gathered in the early phases, without entering into detailed design. The most successful companies in our sample apply three methods to this end.

Formal pre-project stage. Most companies use a formal procedure that regulates the activities and documents to be completed in the concept generation and product planning phases, which usually end with top manage- ment’s formal approval. Some of them also use check-lists to screen the product concept in the light of typical downstream constraints and opportunities (in this respect, see also the study of de Brentani, 1986). This does not mean that these companies define detailed specifications, nor that they restrict their feed-forward effort only in the early phases of product development (indeed, proactive thinking is necessary during all the product development process). They simply try to focus the initial phases of the project on the sole effort of anticipating information. They clearly define the objectives, the basic concept, the architecture and parts of the product and of the production process, and develop a plan for the following downstream phases. Detailed analyses are only performed on a few critical issues, with the aim of anticipating possible constraints and oppor- tunities rather than providing thorough and complete drawings. Any other detailed design is instead pushed further in subse- quent phases in order not to divert the pre- project team from its feed-forward effort. The only focus is proactively thinking on the consequences and opportunities that will emerge in the product life cycle.

Managerial techniques and tools. In recent years several methods have been proposed to support the feed-forward effort of the pre project team. Examples are Quality Func- tion Deployment, Target Costing, Failure

assure communication and anticipation of information from the product life cycle. One of the reasons for this ineffectiveness of feed- forward planning was that team members did not share common objectives while involved in concept generation. Hence, they implicitly or even explicitly paid prevailing attention to their own functional performance. In these cases, for example, the Product Engineering representative often hindered any early deci- sion that could place additional constraints to product design and increase the time needed to provide detailed drawings. In this situa- tion, with a lack of a shared system of objectives, integration in the pre-project team is unlikely to occur and early involvement may result in a fruitless exercise. Upstream departments, exposed to time pressure, do not see any advantage in anticipating and communicating their likely product design to downstream phases, while Process Engineer- ing and Manufacturing bring to pre-project only their constraints and feasibility prob- lems, restraining the creativity and innovation typical of concept development. Of course, a new culture and distinctive capabilities are necessary when applying the feed-forward approach. Upstream phases have to master the development of new solutions in the presence of a higher number of constraints, while downstream phases should learn to be creative. However, such a culture and capabilities are ineffective (and moreover are unlikely to be nourished) if pre-project team members do not have com- mon objectives. This means, for example, all members being measured on the overall time to market of the project, or on the break- even-time of the product (House and Price, 1991). A harmonized performance measure- ment system fosters early sharing of information, since all pre-project team mem- bers are measured both on the newness and integrity of product concept and on their ability to avoid the detrimental impacts of late, unexpected events.

3.3. Encouraged and supported proactive thinking

Anticipating information in the early phases of a product development project is a difficult task, since plenty of uncertain and unstruc- tured information should be analysed.

© Blackwell Publishers Ltd 1997

R&D Management 27, 4, 1997 383

Roberto Verganti

Mode Failure Effect Analysis, Value Analy- sis, Life Cycle Costing and Analysis. Target Costing (Sakuray, 1989) was the most applied technique in our sample while QFD was used only partially successfully in 6 cases. Indeed, as also other studies demon- strate (Akao, 1990; Govers, 1996), QFD stimulates cross-functional proactive think- ing, but it also bounds the pre-project team to thorough and long early analyses, which in turbulent environments may be useless and counterproductive.

Early prototyping. One of the most effective mechanisms to foster early discussion and enhancing the feed-forward exercise is to simulate as early as possible the physical characteristics of the new product. To this

purpose a useful method is early prototyping, i.e. the early and rapid development of a prototype that reflects the expected features, functions, and characteristics of the product

in a rough and approximated manner. Again,

early prototyping should only aim at stimu- lating proactive thinking and not at testing detailed solutions; hence, it may be easily

and quickly realized by the pre-project team.

A white-goods company, for instance, sim-

ply modifies a product of the previous generation in order to simulate the functions and characteristics of the new product. A car manufacturer has decided to anticipate the development of prototypes in the first phase of product development, although these prototypes may not be an accurate and detailed reproduction of the new car. A company in the helicopters industry uses Electronic Mock-Up Systems to detect any geometric interference of parts during concept generation. For further insights on advanced technologies that are currently accelerating the fabrication of early proto- types see Bullinger et al., 1995.

3.4. Integration with detailed design

Efforts of anticipating information in pre- project activities are fruitless if this informa- tion is not transferred and exploited in the following phases of product development (Fujimoto, 1989). A white-goods company, for example, designated its R&D, marketing and manufacturing directors as responsible for developing the concept and specifications of a

384 R&D Management 27, 4, 1997

new refrigerator. These highly experienced people were able to identify and anticipate information on a number of critical areas of the new product. However the refrigerator faced several problems and modifications during production ramp-up since product and process engineers had only a minor knowl- edge of the criticalities that were identified by this high-level pre-project team. We observed several ways of integrating pre-project with downstream detailed design. Most companies rely on reports that describe product specifications. Albeit necessary, these reports in themselves are rarely able to transfer the richness of the unstructured information discussed in the pre-project team. The most effective solution appears to provide continuity between people involved in the early phases and the design teams: first of all, by assuring that the project manager also belongs to the pre-project team and participates in concept generation and pro- duct planning; second, by keeping the pre- project team in action during product and process design (e.g., as a product planning council); third, by involving working-level members early in product planning, at least in some meetings or in the preliminary col- lection of downstream information.

3.5. Planned flexibility

As seen in section 3.3, encouraging proactive thinking does not necessarily mean that the pre-project team must anticipate and analyse in detail any possible constraint and opportun- ity of the product life cycle. In some cases, too much information is more detrimental than effective. Consider the following example. A white-goods company of our sample adopts a stage-gate structure for the product development process. This means that product development is split into sequential phases, separated by check points (see Coo- per, 1990). In particular, after investigation and product planning, a strategic committee has to certify the coherence and completeness of product specifications before proceeding any further. Product specifications thereafter are not allowed to change. An extreme and too strict application of this approach created severe problems in a project for developing a new washing-machine. The concept develop- ment team was forced to define in detail the

© Blackwell Publishers Ltd 1997

Systemic learning in the early phases of product innovation

product specifications before starting product design. The consequence was that concept generation and product planning took twice the time of product and process design. More- over, the quality of the product was compromised, since the pre-project phase consumed such a large amount of time that customer’s needs had already changed when the product reached the market. Indeed, complete and detailed product specifications are seldom necessary, especially when the environment faced by the company is turbulent and variable, or in highly innovative trials (see Iansiti, 1995). Here, initial uncertainty is considerable. Downstream information therefore is not completely available during product planning and risks of implementing late corrective actions cannot be unconditionally avoided. In the helicopters industry, for example, defining detailed specifications for avionics systems is useless, since technology is con- tinuously changing and new unpredictable opportunities often emerge. In a white goods company, the pre-project team was unable to certify the reliability of a device for controll- ing the temperature of a new oven. In fact, this device was radically innovative and its inclusion in the new product was only poss- ible by completing its refinement simul- taneously with oven development. In these cases, where the information needed for defining product specifications does not become available until the team gets into design, detailed feed-forward planning does not provide significant insights on future constraints opportunities and eventually results in a waste of precious time. A feed- back approach therefore seems to be the only viable alternative (Smith and Reinertsen, 1991; Von Hippel and Tyre, 1995; Ward et al., 1995; Thomke, 1996). However, feed- forward planning again plays a central role in controlling the detrimental impacts of a reactive policy. In fact, the cost and time of late corrective actions may be greatly reduced if preventive measures are taken from the early development phases. For example, the product may be conceived with a modular structure so that any change in a single part will not affect the overall system. Protective and alternative solutions may also be pre- viously arranged to overcome any infeasibil- ity problem due to a new technology. Finally,

© Blackwell Publishers Ltd 1997

additional resources may be allocated to the most uncertain activities. These preventive measures confer flexibility to the product development process (on the concept of development flexibility see Thomke, 1996). Without this flexibility the feed-back appro- ach would provide poor performance. Note that preventive measures are effective pro- vided they are arranged in the early phases of the project. For this reason, we call this prin- ciple planned flexibility. It entails that the flexibility of the product development process is planned early so that late adjustments and innovations may be accepted with marginal consequences. In other words, referring back to Figure 2, it means that feed-forward plan- ning should aim both at: (1) reducing uncertainty about constraints and opportuni- ties by anticipating downstream information, thus decreasing the probability of implement- ing late corrective actions; and (2) reducing the cost and time of these corrective actions, thanks to planned flexibility. The principle of planned flexibility also shows that feed-forward and feed-back plan- ning are not reciprocally incompatible. On the contrary, they may be jointly applied with outstanding results and mutual support. This synergy is evident in those companies in the sample that use an overlapped structure in product development. ‘Overlapping’ means to carry on simultaneously different phases of the product development process (Imai et al., 1988). It may provide superior performance, as demonstrated in several studies. Clark and Fujimoto (1991), for example, analysed the benefits of overlap- ping product design and process design in car development, while Iansiti (1995) investi- gated the practice of overlapping concept generation and detailed design in the main- frame and multimedia industries. Overlap- ping is a feed-back approach. In fact, it implies that upstream phases progressively refine a new solution on the basis of the rapid downstream feed-backs ensuing from the actual attempts to implement the innovative solution (Fujimoto, 1989; Ward et al., 1995). In addition, overlapping entails that, with two or more phases simultaneously in pro- gress, late changes may be accepted with low impact on cost and time. For example, insofar as product specifications are not frozen, the product concept may be adjusted

R&D Management 27, 4, 1997 385

Roberto Verganti

to deal with sudden changes in market needs or in the competitive environment and pro- duct design may be updated according to new technological developments, especially those concerning the manufacturing process. In other words, overlapping enlarges the window for reacting to late constraints and opportunities. Being a reactive approach, overlapping implies the embedding of substantial flexibility into the product development process. This leads again to the pre-project policy of a company and to the planned flexibility principle. Indeed, several com- panies in the sample followed an overlapped path for developing their new product. How- ever, successful implementations were observed only in those cases where overlap- ping was an explicit approach, i.e. where the flexibility it needs (e.g., cross-functional teams and enhanced communications means) was properly planned and activated since the early phases of the project. A car manufac- turer, for example, noted that overlapping product design and process design was effective only when product designers, pro- cess designers and suppliers had frequent and direct interactions, avoiding endorse- ment from project manager and functional managers. This increased level of delegation to working-level teams was achieved by fostering the proactive orientation of mana- gers during concept generation and product planning. Top managers and leaders devoted great attention in the early phases to analyse the coherence between product concept and future constraints (including functional constraints due to a lack of resources and functional strategies). Apart from a few critical issues that were specified in details, their major purpose was to agree upon allow- ances within which designers could have margins for decision. Detailed design teams were then able to interact directly and had a higher degree of freedom to seize quickly any unexpected opportunity. This example shows that overlapping and feed-forward planning may provide superior performance when jointly applied. In fact, on the one hand overlapping becomes effective especially when supported by proactive thinking and planned flexibility. On the other hand, overlapping, with its increased reactive capabilities, may provide a useful support to

386 R&D Management 27, 4, 1997

feed-forward planning, since the proactive effort may focus only on few critical ele- ments, avoiding detailed and time consuming analyses of every downstream constraint and opportunity.

4. Systemic learning: the keystone of feed-forward planning

As emerged in the above discussion, the proactive capability of a company does not depend on the time and effort spent in concept generation and product planning. What actually makes the difference in feed- forward planning is the capability of the pre- project team to identify early on any critical areas and to forecast their influence on pro- ject performance. Critical areas have two major characteristics: (1) they are the most uncertain as well as relevant elements in the product life cycle, and (2) they call for costly and time consuming corrective actions if detected and handled late in the project. Critical areas may concern the market needs, the competitive environment, the product and process technologies, the regulations or the project resources. They should receive pri- mary attention and anticipation effort in the pre-project team. For example, in helicopter development, the distribution of weights has to be defined and thoroughly specified as early as possible, since any subsequent change could generate a number of modifications in several components and in the architecture of the product. A white- goods company recognized as a critical area in a new refrigerator project the high variabil- ity in the characteristics of the polyurethane insulation provided by its prospective sup- plier; a detailed analysis in product planning was devoted to define the proper thickness of insulation material and, therefore, of the body of the refrigerator. The product design teams were therefore able to provide solutions which already accounted for constraints in raw materials and in the supply process. The basic issue in feed-forward planning is therefore how to identify critical areas in advance and how to understand their impacts, given the initial project uncertainty. In this respect, the previous section emphas- ized the benefits of teamworking, communi- cation, common objectives, and tools, such

© Blackwell Publishers Ltd 1997

Systemic learning in the early phases of product innovation

Process Engineering to anticipate down- stream constraints strongly depends on the information provided by the product engineer and, therefore, on his understanding of the process technologies. This example shows that feed-forward planning is made effective by the systemic knowledge held by each member of the pre- project team, i.e. by his capability to detect early which elements in his own decisions may have significant consequences on the other phases of the product life cycle. This systemic knowledge is not built through a learning by developing mechanism, such as in feed-back planning. Being in the initial phases, systemic knowledge may only be drawn from previous projects, i.e. through learning from experience mechanisms. Hence, the capability of a company to learn from past projects and to incorporate this systemic experience in the pre-project team appears as the keystone of feed-forward planning. We call this principle systemic learning. The importance of systemic learning mechanisms clearly emerges from our case studies. In particular, Figures 3–5 compare the contributions of systemic learning and teamworking to feed-forward planning. Measures are drawn by data in the structured questionnaires, which are aggregated by means of fuzzy functions and fuzzy opera- tors. Details on metrics are provided in the appendix. Here, two major points are high- lighted. First, all measures are relative, thus making it possible to compare companies operating in different industries. Second, we do not analyse the impacts of the above principles on the overall performance of the product development project (e.g., product success in terms of sales), since overall performance is affected by several parameters and phenomena (company A for example, ranks third in terms of product functional features; however, because of a lack of feed- forward capabilities, it achieved this perfor- mance to the detriment of extremely high product costs and time to market). On the contrary, we restrict the analysis on the sole parameters that reveal the capability to antici- pate uncertain downstream information in the early phases of product development. This feed-forward effectiveness was measured through a fuzzy function depending on the

as Quality Function Deployment and early prototyping. However, our case studies demonstrated that feed-forward capabilities also depend on ‘soft’ rules and behavioural patterns. Consider the following example. A white goods company should develop a new, very small refrigerator. New technological solutions are proposed and, from a technical point of view, the new product is feasible. However, the concept development team wants to detect early any problem that could emerge in the assembly line that could make the product unmanufacturable (unless at very high costs). To deal with this issue, a feed- back or a feed-forward approach could be implemented. A purely reactive approach would entail that Product Engineering starts developing the new refrigerator and provides information and drawings to Process Engin- eering as soon as possible; then Process Engineering begins to develop dies, tools and fixtures and verifies any assembly problems. Any feed-back concerning manufacturing constraints would reflect the actual practice of Process Engineering in attempting to implement the upstream design. In other words, feed-back is based on a learning by developing mechanism. Feed-forward plan- ning, on the contrary, is based on a completely different and far more complex procedure. This entails that in the concept generation phase Product Engineering and Process Engineering, both belonging to the pre-project team, exchange uncertain infor- mation on future constraints and opportuni- ties. In particular, about every possible information need that could be anticipated, Product Engineering selects only those parts of the new technological solution that could generate problems during process design and manufacturing. Note that these critical areas are not necessarily concerned with the most complex parts of the product; in other words, the product engineer has not to anticipate criticalities on his own future tasks (usually these criticalities are easily known by a skilled designer). The difficult thing is that Product Engineering should identify, without entering into detailed design, any critical areas that could create problems in other phases. Process Engineering would then analyse this preliminary information in order to detect possible manufacturing constraints and opportunities. Hence, the capability of

© Blackwell Publishers Ltd 1997

R&D Management 27, 4, 1997 387

Roberto Verganti

occurrence of the typical drawbacks of inef- fective pre-project phase (such as reworks, engineering changes, unexpected events that call for additional work in later stages of product development, higher product costs and time to market than foreseen). Figure 3 shows the relation between feed- forward effectiveness and the degree of teamworking and communication in early phases (a fuzzy function that accounts for involvement in the pre-project team of the major actors of the product life cycle, for their teamworking attitude, and for their communication patterns). Company A, that appoints concept generation to its Product Engineering Manager, has poor proactive capabilities compared to other companies, all using a team approach to pre-project. This highlights how communication in the early phases is necessary in order to anticipate downstream information. It also confirms results of other studies on early involvement, with the note that teamworking in early phases is becoming a diffuse practice. How- ever, if we focus the analysis on 18 companies excepting A, teamworking does not appear as a sufficient condition for suc- cessful feed-forward planning. Some companies (see for example company B) have considerable degrees of teamworking and communication. However, this does not directly reflect in high feed-forward effec-

this does not directly reflect in high feed-forward effec- Figure 3. communication and proactive capabilities. Relation

Figure 3.

communication and proactive capabilities.

Relation between teamworking

388 R&D Management 27, 4, 1997

tiveness, especially if compared to company

C that is remarkably proactive.

The above results were therefore further examined in the light of the systemic knowl- edge embodied in pre-project teams. As said above, systemic knowledge is the under- standing of each pre-project team member about the impacts of his specific decisions on the product life cycle as a whole. Direct measures of systemic knowledge were not

available, because of the intangible nature of this variable. We therefore investigated the mechanisms that each company uses to create this knowledge, i.e. the systemic learning mechanisms. In particular, systemic learning only occurs if people that developed the product concept and specifications also receive and conceptualize the feed-backs related to their original choices. Project termination meetings and project audit at the break-even time of the product appear as extremely useful in this regard (for studies specifically investigating learning mechan- isms in new product development, see Maidique and Zirger, 1985; McKee, 1992; Clark and Wheelwright, 1993; Caffyn, 1996; Hughes and Chafin, 1996; Bartezzaghi et al., 1997). Indeed, the most proactive companies have a common policy in introducing systemic knowledge to the early project phases. Their pre-project team members have

a major role also during project termination

and audit in order to learn from variations between their prospected early solutions and actual final outcomes. In companies D and E, for example, the pre-project team is also appointed to the project termination phase; in these project termination meetings priority is given to discussion of unexpected problems and to understanding reasons for variations in project results. Job rotation of team mem- bers, on the contrary, seems to be less capable of generating systemic learning, since it usually implies upstream solutions and downstream constraints being experi- enced in different development projects. The learning cycle is therefore interrupted and, in addition, specialized knowledge is likely to be compromised. In other words, the pre- project phase does not necessarily need generalists, but specialists with knowledge of the systemic impacts of their choices. These considerations are supported by the structured data collected in the questionnaires.

© Blackwell Publishers Ltd 1997

Systemic learning in the early phases of product innovation

C shows considerable proactive capabilities

with only a moderate orientation toward

teamworking (its pre-project team consists of the Marketing, Product Engineering and Process Engineering functional managers). This means that systemic knowledge and understanding of the impacts of early choices

on the entire product life cycle is often more important than enlarging the number of people involved in early phases.

Systemic learning was measured through a fuzzy function depending on: existence of a formal project termination or project audit phase; relative importance, in that phase, of the analysis of mismatching between initial solutions and resulting constraints oppor- tunities; presence in the project termination team of the same members that belonged to the pre-project team. One company has been skipped in this analysis since it has been using project termination procedures only recently (less than 4 projects). Its learning path is therefore not identifiable. Figure 4 demonstrates how systemic learn- ing emerges as a keystone for feed-forward planning (the correlation coefficient between degree of systemic learning and feed-forward planning is 0.76, with a significance level of more than 99%). In addition, Figure 5 points out the superior performance provided by teamworking when supported by systemic learning. In particular, the two most effective companies (D and E) are characterized by high teamworking and systemic learning, while the unsuccessful project in company A lacks both these principles. Companies with low and intermediate performance mainly locate in the lower-right area of the matrix:

this confirms that teamworking is not a sufficient condition for effective anticipation of information, unless it is sustained by systemic learning. Most interesting, company

is sustained by systemic learning. Most interesting, company Figure 4. past development projects and proactive

Figure 4.

past development projects and proactive capabilities.

Relation between systemic learning from

capabilities. Relation between systemic learning from Figure 5. Joint analysis of teamworking communication and

Figure 5.

Joint analysis of teamworking communication and systemic learning.

© Blackwell Publishers Ltd 1997

R&D Management 27, 4, 1997 389

Roberto Verganti

5. Conclusions

Managing the product development process implies handling reciprocal interdependencies between upstream solutions and downstream constraints opportunities in the product life cycle. The early phases of a development project play a central role in this regard. In fact, by anticipating as early as possible the analysis of implications due to a new product concept, or to a new product and process technology, late corrective actions may be avoided and high levels of quality and inno- vation may be achieved. However, proactive policies are extremely complex, because of the high uncertainty inherently induced by inno- vation and environmental turbulence. In this paper the basic mechanisms that make early project activities and feed-forward planning successful have been investigated. In particu- lar the following principles have been drawn by observing successes and failures in 19 case studies: teamworking and communication, harmonized objectives, encouraged and sup- ported proactive thinking, integration with detailed design, planned flexibility, systemic learning. All these principles are necessary and are effective only if jointly applied. The major challenge is that they are only partially encoded in managerial techniques and tools. Instruments, such as checklists, mapping tools, structured data-bases or early prototyp- ing may be useful. However, ‘soft rules’, behavioural patterns and, most of all, the skills of pre-project team members emerge as essen- tial. In particular, the crucial capability is what we call systemic knowledge, i.e. the capability to detect the impacts of specific early decisions on the downstream phases of the product life cycle. A lack of this capability makes working in teams fruitless and causes sterile feed-for- ward efforts. The paper shows how systemic knowledge and proactive skills are nurtured by continu- ously striving to learn from experience in product development. Companies in the sam- ple that involved pre-project team members also in project termination and that stimulated the analysis of variances between early solu- tions and actual final outcomes demonstrated the highest proactive capabilities. Systemic learning is therefore the driving force behind feed-forward planning. It is the keystone that makes the other proactive principles effective.

390

R&D Management 27, 4, 1997

Systemic learning entails that there is a subtle link between the current proactive effort and previous similar projects. This shows that learning from experience is not marginal, even for managing innovation processes where previous and established assumptions must be overcome. Radical developments of new products and continuous improvements of capabilities can co-exist and, indeed, are the self-supporting leverages of the same successful mechanism.

Appendix

In order to compare the proactive capabilities of the nineteen companies, structural data where collected by means of a questionnaire with more than 200 parameters. By aggregat- ing some of these parameters we obtain measures for the 3 synthetic variables reported in Figures 3–5 (feed-forward effectiveness, teamworking and communication, systemic learning). Aggregation is implemented thanks to fuzzy set theory (Zimmermann, 1993). This is necessary since each synthetic variable has two major characteristics: (1) it is multidimensional, being dependent on several constituting parameters through a hierarchical structure (for example, systemic learning depends on the main focus of project termina- tion, on the main roles involved in project termination and on the main roles involved in pre-project); (2) its constituting parameters are measured in the questionnaire by heterogeneous indexes (for example the focus of Project Termination is a Likert’s-like scale, while the involvement in pre-project of specific roles such as Manufacturing or Pro- duct Engineering is a logical variable). In this case, the ‘crisp’ fuzzy logic is adopted, that is:

heterogeneous parameters are transformed into homogeneous fuzzy functions whose value varies between 0 and 1; then, these constituting fuzzy functions are aggregated through fuzzy operators. In particular, two compensatory operators are used in the study:

X.Fuzzyand.Y = α min(X, Y) + (1 − α)(X + Y) 2, X.Fuzzyor.Y = α max(X, Y) + (1 − α)(X + Y) 2 , where α = 0.2.

Table 2 reports the aggregation operators

© Blackwell Publishers Ltd 1997

Systemic learning in the early phases of product innovation

Table 2.

Measures for the synthetic variables analysed in Figures 3–5.

Synthetic

variable

Aggregation operator

Notes on constituting parameters

Feed-forward

1 – [(Reworks .Fuzzyor. Unexpectedevents). Fuzzyor.(Costvariance.Fuzzyor.Timevariance)]

Reworks = occurrence of reworks during detailed design, relative to other flaws in the project not depending on feed- forward capabilities (e.g., lack of top management support); Unexpectedevents = occurrence of events initially unforeseen, relative to other flaws in the project; Costvariance = variance of final product costs with respect to the target cost established in the early phases. This variance is considered to be extremely critical if product cost is a competitive priority for the company; Timevariance = variance of the actual time to market with respect to the expected time to market established in the initial plan. This variance is considered to be extremely critical if time is a competitive priority for the company.

effectiveness

Teamworking and

communication

Departments .Fuzzyand.Attitude .Fuzzyand. Communication

Departments = number of major roles involved in the pre-project team (Platform Manager or Product Planning, Marketing, Research or Product Engineering, Process Engineering, Manufacturing, Purchasing, lead customers, major suppliers); Attitude = importance of attitude towards integration when selecting pre-project team members relative to other skill parameters (e.g., technical competencies); Communication = % of co-located team members.

Systemic learning

Team .Fuzzyand. Focus

Team = percentage of pre-project team members involved in project termination of previous similar projects; Focus = importance of analysis of variances during project termination of past projects, relative to other purposes of project termination (execution of production rump-up, late problem fighting, resource evaluation).

for the 3 synthetic variables analysed in the study. A brief explanation of the constitut- ing parameters from which synthetic measures have been derived is also pro- vided. It is important to point out that all measures are relative, thus making it poss- ible to compare companies operating in different industries.

© Blackwell Publishers Ltd 1997

References

Akao, Y. (ed.) (1990) Quality Function Deployment. Integrating Customer Requirements into Product Design. Cambridge MA:

Productivity Press. Bacon, G., Beckman, S., Mowery, D. and Wilson, E. (1994) Managing product definition in high-technology industries: a pilot study. California Management Review, Spring, 32–57. Bartezzaghi, E., Corso, M. and Verganti, R. (1997) Continuous improvement and inter-project learning in new product development. International Journal of Technology Management, 14, 116–138.

R&D Management 27, 4, 1997 391

Roberto Verganti

Blanchard, B.S. (1979) Design and Manage to Life Cycle Cost.

Portland: M A Press.

Brentani, U. de (1986) Do firms need a custom-designed new product screening model? Journal of Product Innovation Man- agement, 3, 108–119. Brentani, U. de (1991) Success factors in developing new business services. European Journal of Marketing, 15, 2, 33–59. Brown, S.L. and Eisenhardt, K.M. (1995) Product development:

past research, present findings, and future directions. Academy of Management Review, 20, 2, 343–378. Bullinger, H.J., Fischer, D. and Worner, K. (1995) Rapid prototyp- ing — a method for the development of innovative products. The 13th International Conference on Production Research, Jer- usalem, 6–10 August, pp. 426–429. Caffyn, S. (1996) Extending continuous improvement to the new product development process. R&D Management, 27, 3,

253–267.

Clark, K.B. (1989) Project scope and project performance: the effect of parts strategy and supplier involvement on product development. Management Science, 35, 1247–263. Clark, K.B. and Fujimoto, T. (1990) The power of product integrity. Harvard Business Review, November–December,

107–118.

Clark, K.B. and Fujimoto, T. (1991) Product Development Performance. Cambridge MA: Harvard Business School Press. Clark, K.B. and Wheelwright, S.C. (1993) Managing New Product and Process Development. New York, NY: The Free Press. Cooper, R.G. (1990) Stage-gate systems: a new tool for managing new products. Business Horizons, 33, 3, May–June. Cooper, R.G. (1994) Third-generation new product process. Journal of Product Innovation Management, 11, 3–14. Cooper, R.G. and Kleinschmidt, E.J. (1994) Determinants of timeliness in product development. Journal of Product Innovation Management, 11, 381–396. Cooper, R.G. and Kleinschmidt, E.J. (1995) Benchmarking the firm’s critical success factors in new product development. Journal of Product Innovation Management, 12, 374–391. Corso, M., Muffatto, M. and Verganti, R. (1996) Multi-product innovation: emerging policies in automotive, motorcycle, and earthmoving machinery industries. 3rd International Product Development Management Conference, EIASM, Fontainebleau, April. De Maio, A., Verganti, R. and Corso, M. (1994) A multi-project management framework for new product development. European Journal of Operational Research, 78, 178–191. Dyer, J.H. and Ouchi, W.G. (1993) Japanese-style partnerships:

giving companies the competitive edge. Sloan Management Review, Fall, 51–63. Ettlie, J. E. (1996) Global comparisons of early manufacturing involvement (EMI) in new product development. EIASM 3rd International Product Development Conference, Fontainebleau, France, 15–16 April, pp. 275–298. Fujimoto, T. (1989) Organization for Effective Product Develop- ment: the Case of the Global Automobile Industry. PhD Dissertation, Harvard University, Graduate School of Business Administration, Boston MA. Fujimoto, T. (1993) Information asset map and cumulative concept translation in product development. Design Management Journal, Fall, 34–42. Fujimoto, T., (1997) Shortening lead time through early problem solving — a new round of capability building competition in the auto industry. International Conference on New Product Develop- ment and Production Networks, WZB, Berlin, 20–22 March. Govers, C.P.M. (1996) QFD as a way of management. R&D Management Conference on Quality and R&D, Enschede, The Netherlands, 6–8 March, pp. 144–153. Gupta, A.K. and Wilemon, D.L (1990) Accelerating the develop- ment of technology-based new products. California Management Review, 32, 2, Winter, 24–44. Haddad, C. (1997) Involving manufacturing in the early stages of product development: a case study from the North American automotive industry. International Conference on New Product Development and Production Networks, WZB, Berlin, 20–22 March. Hayes, R., Wheelwright, S.C. and Clark, K.B. (1988) Dynamic

392 R&D Management 27, 4, 1997

Manufacturing. Creating the Learning Organization. New York NY: Free. Hippel, E. von and Tyre, M.J. (1995) How learning by doing is done: problem identification in novel process equipment. Research Policy, 24, 1–12.

House, C.H. and Price, R.L. (1991) The return map: tracking product teams. Harvard Business Review, January–February, 92–101. Houser, D.E. and Clausing, D. (1988) The house of quality. Harvard Business Review, May–June, 63–73. Hughes, G.D. and Chafin, D.C. (1996) Turning new product development into a continuous learning process. Journal of Product Innovation Management, 13, 89–104. Iansiti, M. (1995) Shooting the rapids: system focused product development in the computer and multimedia environment. Working Paper 95-026, Boston: Harvard Business School. Imai, K., Nonaka, I. and Takeuchi, H. (1988) Managing the new product development process: how Japanese companies learn and unlearn. In Clark, K.B., Hayes, R.H. and Lorenz, C. (eds.), The Uneasy Alliance: Managing the Productivity-Technology Dilemma. Boston MA: Harvard Business School Press, pp.

337–375.

J urgens, U. (1995) Anticipating problems with assembly in the early stages of product development. Third International Work- shop on Assembly Automation, Venice, October 12–14. Kamath, R.R. and Liker, J.K. (1994) A second look at Japanese product development. Harvard Business Review, November–December, 154–173. Liker, J.K., Kamath, R.R., Wasti, S.N. and Nagamachi, M. (1996) Supplier involvement in automotive component design: are there really large US Japan differences?’ Research Policy, 25, 59–89. Maidique, M.A. and Zirger, B.J. (1985) The new product learning cycle. Research Policy, December, 299–313. McKee, D. (1992) An organizational learning approach to product innovation. Journal of Product Innovation Management, 9,

232–245.

Meredith, J. R. and Mantel, S. J. (1989) Project Management. A Managerial Approach, New York NY: John Wiley. Rosenthal, S.R. (1992) Effective Product Design and Development:

How to Cut the Lead-Time and Increase Customer Satisfaction, Homewood IL: Business One Irwin. Sakuray, M. (1989) Target costing and how to use it. Journal of Cost Management, Summer. Slack, N., Chambers, S., Harland, C., Harrison, A. and Johnston, R. (1995) Operations Management. London: Pitman Publishing. Smith, P.G. and Reinertsen, D.G. (1991) Developing Products in Half the Time, New York NY: Van Nostrand Reinhold. Souder, W.E. and Monaert, R.K. (1992) Integrating marketing and R&D project personnel within innovation projects: an information uncertainty model. Journal of Management Studies, 29, 4, July,

485–512.

Trygg, L. (1991) The need for simultaneous engineering in time-

(eds.), Tech-

nology Management: Technology Strategies and Integrated Information Systems in Manufacturing. Helsinki: VTT. Thomke, S.H. (1996) The role of flexibility in the development of new products: an empirical study. Working paper 96-066, Boston:

Harvard Business School. Thompson, J.D. (1967) Organizations in Action, New York NY:

McGraw-Hill. Ward, A., Liker, J.K., Cristiano, J.J. and Durward, K.S.II (1995)

The second Toyota paradox: how delaying decisions can make better cars faster. Sloan Management Review, Spring, 43–61. Weerd-Nederhof, P.C. de, Kerssens-Van Drongelen, I.C. and Verganti, R. (eds.) (1994) Managing the R&D Process. Enschede, The Netherlands: TQC. Wheelwright, S.C. and Clark, K.B. (1992a) Revolutionizing Product Development: Quantum Leaps in Speed, Efficiency and Quality. New York NY: Free. Wheelwright, S.C. and Clark, K.B. (1992b) Creating project plans to focus product development. Harvard Business Review, March–April, 70–92. Wheelwright, S.C. and Sasser, W.E. (1989) The new product development map. Harvard Business Review, May–July, 112–123. Zimmermann, H.J. (1993) Fuzzy Set Theory and its Application, 6th ed. London: Kluwer Academic.

based competition. In Ranta, J. and Horte,

S. A.

© Blackwell Publishers Ltd 1997