Creating and managing flexibility in new product development

(1)

Creating and managing flexibility in new

product development

By

Marc Padding

University of Groningen

Faculty of Economics and Business

MSc BA Strategic Innovation Management

(2)

(3)

Abstract

(4)

Introduction

New product development (NPD) is seen as a crucial factor for the growth and survival of the firm (De Visser, de Weerd-Nederhof, Faems, Song, Van Looy, and Visscher, 2010). This factor is especially important in today’s strong economic competition and environmental uncertainty (Mumford and Licuanan, 2004). However, many studies carried out about how to effectively manage NPD projects were based on environments with target markets and technologies in products that were quite well understood (MacCormack, Verganti, and Iansiti, 2001). An example of this is the Stage-Gate model of Cooper (1990). The Stage-Gate model is widely acknowledged and Cooper (1990) states that the use of it will improve the effectiveness and efficiency of the project to reduce cycle time. The essence of this model is to structure the NPD process and minimize changes once the execution stage has begun (Cooper, 1990). In a more recent article of Cooper (2008), he refers to the traditional explanations of the Stage-Gate model and emphasizes the importance of concepts as: flexibility, simultaneous execution of development tasks, and the adaptability to changing conditions and fluid, unstable information.

This particular example of the evolution of the Stage-Gate model is in line with the call of scholars for the need for a more flexible approach for NPD processes to deal with today’s turbulent environment (Biazzo, 2009; Thomke, 1996). Fast changes in customer needs (and more critical consumers), rapidly evolving technologies, increased competition, changes in legal regulations and shortened product life-cycles are all examples of the turbulent environment a company faces. Due to this turbulent environment, there is a need for companies to respond to environmental changes during the NPD process. Companies without this ability are so to say inflexible, and face the risk of designing a product that does not fit to, for example the (changed) requirements of the customer (MacCormack et al., 2001).

(7)

Several scholars already investigated the concept of flexibility, but most studies did not focus on flexibility in the NPD process itself, but on other levels of analysis. On the manufacturing level of the firm, Sethi and Sethi (1990) operationalized manufacturing flexibility as ‘being able to reconfigure manufacturing resources so as to produce efficiently different products of acceptable quality’. Also, in order to achieve flexibility on the organizational level of the firm, it was found that organizations should attain a flexible structural design, to succeed under environmental turbulence (Englehardt and Simmons, 1997). Further, Sanchez (1995; 1997) focused on the industry level, by stating that the emergence of CAD/CIM systems have increased the potential flexibilities of product resources. However, these flexibilities found are more about the bigger picture, the preconditions of a flexible development process, but they do not really explain how this flexibility is created in the core of the NPD process.

When really focusing on the NPD process itself, a review of the relatively limited literature applicable to NPD projects brings different dimensions and underlying mechanisms of flexibility forwards. For instance, Thomke (1996) found that projects using flexible design technologies, are able to make changes in the product in a later stage of the project. An example of such a flexible design technology is modularity, and by creating a modular product architecture, the components of the product have interface characteristics that are within the architectural design of the project, thereby minimizing interdependence between those components (Sanchez and Mahoney, 1996). Verganti (1999) investigated whether to choose for an anticipation strategy (making decisions in the early phases of the project) or a reaction strategy (delaying decisions downstream). He found that neither anticipation of reaction may be considered as a best practice, but instead project teams should manage the early phases of the project in the right way. Moreover, Biazzo (2009) developed a conceptual framework based on much of the literature about flexibility in NPD. He separated three analytical dimensions, and the first dimension was the organizational dimension, which is about the degree of structuration in the process design. The second dimension he reconceptualized, was the informational dimension. This dimension deals with the degree of intersection between problem-formulation activities and problem-solving activities. The last dimension was the temporal dimension, and is about the execution strategies of development tasks and task scheduling.

(8)

choices regarding to use (a certain type of) flexibility. Secondly, besides the study of Biazzo (2009), there is little integration with regard to the different flexibilities dimensions found in the literature, and therefore the impact of these different dimensions on performance is unclear.

Concluding, although scholars emphasize the need for flexibility in NPD processes, to date there is no consensus about how exactly flexibility is created in NPD projects, and the impact the dimensions have on the performance of the project. Moreover, dimensions are studied on different levels of analysis. Therefore, this research will investigate how project teams create flexibility in NPD, and the impact it has on performance. The research question of this paper is as follows:

How do project teams create flexibility in NPD, and how does this flexibility influence performance?

(9)

Literature

review

In this section, the literature on NPD will be discussed and the flexibility dimensions found in the literature will be explained. At the end of this section, the dimensions found will be compared and shown in a framework.

New product development

‘Innovate or die’ is a popular message in business, and is in line with the statement in the introduction that new product development (NPD) is a crucial factor for the growth and survival of the firm. This implies both radically and incrementally new products (Garcia and Calantone, 2002). The literature on NPD is rather broad; it has been studied extensively from different angles, and therefore, new product development will be defined broadly as: ‘the transformation of a market opportunity and a set of assumptions about product technology into a product available for sale’ (Krishnan and Ulrich, 2001). The most popular approach for managing NPD is the Stage-Gate model of Cooper (1990). This model is a blueprint for managing the NPD process effectively and efficiently. The Stage-Gate model consists of a series of stages, starting with a product idea and ending with a successful new product. Each stage is followed by a go/kill decision point: at these points, the results of this stage are assessed and a decision will be made whether to continue or to abandon the project (Cooper, 1990, 1998). By using the Stage-Gate model, companies can manage risk in NPD and get ideas quicker and with fewer mistakes to the market (Cooper, 1990, 1998; van Oorschot, Sengupta, Akkermans, and van Wassenhove, 2010).

However, a more recent article of Cooper (2008) about the Stage-Gate model provides a more nuanced thought than his first article about it in 1990. Initially, the rationale behind the Stage-Gate model was that it was a one-size fits all model. The typically five stages of the model were suited to all NPD processes and companies. However, in his more recent article, Cooper says that not all new products are suitable for the whole Stage-Gate model, and some products only need a few stages in order to manage the process. Another evolution of the model is the addition of the concept of flexibility into it. In his first article of the Stage-Gate model, Cooper did not mention flexibility, however, in the more recent article Cooper introduced the concept of flexibility into the Stage-Gate model. Dimensions he identified are for example: the freedom of the project team to choose which activities they execute and which they do not execute, and the simultaneous execution of key activities or even entire stages (Cooper, 2008). In the next section, a closer look will be taken on flexibility.

Flexibility

(10)

2009; Cooper, 2008; MacCormack et al., 2001), most studies only viewed flexibility from a limited number of dimensions without any integration of these different dimensions, and on different levels of analysis. Having reviewed the literature on flexibility in NPD, a breakdown in different dimensions was made. Biazzo (2009) positioned his article at the beginning of the theory-building continuum and identified three analytical dimensions in which flexibility can be created: the organizational dimension (the degree of structuration in the process), the informational dimension (the degree of intersection between problem formulation and problem solving activities), and the temporal dimension (the degree of flexibility with regard to task scheduling). However, Biazzo (2009) mainly focused on the NPD process and did not elaborate much on flexibility in the design itself (e.g. by having a modular design or architectural flexibility) and did not name flexibility in the workforce; by having the possibility to adjust the number of workers, or level of worked hours assigned to a task (Coenen and Kok, 2014; Kok and Ligthart, 2013). Furthermore, a new dimension called strategic-decision making flexibility was found. Therefore, the existing literature on flexibility will be divided in six analytical dimensions, consisting of the three dimensions of Biazzo (2009), flexibility in the design of the product, flexibility in the workforce, and strategic-decision making flexibility. Sometimes the dimensions will be subdivided in sub dimensions, which will be further elaborated in the next section.

Organizational flexibility

The first dimension Biazzo (2009) introduced was ’organizational flexibility’, which is defined as: ’the degree of structuration in the process design’. This dimension entails the predefined decision points and the following stages in the NPD process, and the definition of activities that should be occurring in each stage. The structuration of the process can vary from an extremely limited ‘organic’ structure (large scope for changes in the structure, highly flexible) to an extremely detailed structure (little scope for changes in the structure, not flexible). Biazzo (2009) argues that the degree of organizational flexibility has to be driven by the structuration needs of managers; ‘the degree of control and standardization that managers judge to be necessary’.

Informational flexibility

The second dimension Biazzo (2009) points out in his study is the informational dimension of flexibility. In this dimension, the NPD process can be divided into two large categories: problem-formulation activities and problem-solving activities. The problem-problem-formulation activities include the definition of the problem/product and ‘a set of descriptive parameters covering target market segments and channels to reach those segments’ (Bacon, Beckman, Mowery, and Wilson, 1994; Biazzo, 2009), consisting of:

- Product price

(11)

- Technologies on which the product will rely

- An allocation of resources to complete product development

The problem-solving activities are the detailed design engineering activities in order to fulfill the problem-formulation activities. The flexibility in this informational dimension can be found in the degree of intersection between the problem-formulation tasks and the problem-solving activities. Whereby processes with a high degree of intersection between problem-formulation activities and problem-solving activities are flexible and processes with a low degree of intersection between problem-formulation activities and problem-solving activities are inflexible. In contrast to the organizational dimension, the choice for managing the product definition (having a low degree of intersection or having a high degree of intersection), should be based on the preference of using an anticipation strategy, or by using a reaction strategy. The anticipation strategy is characterized by an early and sharp approach to product definition (low degree of intersection between problem-formulation activities and problem-solving activities) and the reaction strategy is characterized by keeping the product definition open (to some extent) (high degree of intersection between problem-formulation activities and problem-solving activities). However, whichever strategy is chosen, Bacon et al. (1994) argue that the management (of change) of product definition (during intersection between problem-formulation activities and problem-solving activities ) is crucial to the successful outcome of a product development effort.

Concerning the allocation of resources to the product development process, Tatikonda and Rosenthal (2000) studied resource flexibility, which they define as: ‘flexibility in reallocation of project resources’. They found a positive relation between resource flexibility and project execution success. This means that project teams who are flexible in allocating resources as equipment, finance and personnel, will achieve a higher performance than teams who are not flexible in allocating those resources. Important to note is, however, that in this present study the resource allocation consists of equipment and financial resources. Personnel resource flexibility is placed in the workforce flexibility dimension, which will be discussed later.

Temporal flexibility

(12)

characterized by the simultaneous start and finish of both tasks (Joglekar, Yassine, and Eppinger, 2001). The degree of temporal flexibility can be placed on the continuum from a sequential execution strategy to a concurrent execution strategy, whereby projects with low temporal flexibility tend to a sequential execution strategy and projects with high temporal flexibility tend to a concurrent execution strategy.

Strategic-decision making flexibility

A relatively new dimension of flexibility is strategic-decision making flexibility, and Kandemir and Acur (2012) define it as: ‘the capability that enables firms to make effective strategic decisions by maintaining multiple simultaneous decision alternatives’. Their study (conducted on the firm level) shows that the direct effects of long-term orientation, strategic planning, internal commitment, an innovative climate, and especially champions and gatekeepers have a positive impact on strategic decision-making flexibility. Thus, firms having a high strategic-decision making flexibility, (have the capability to) maintain multiple simultaneous decision alternatives to make effective strategic decisions. On the other side, firms with a low strategic decision making flexibility do not (have the capability to) maintain multiple simultaneous decision alternatives.

Design flexibility

The fifth dimension of flexibility in NPD is the design dimension, and entails the extent to which a product concept can be kept open for change (MacCormack et al., 2001). Thomke (1996) investigated design flexibility on the project level, and argues that there is high design flexibility when the incremental cost and time of modifying a design is low, and that there is low design flexibility when the incremental cost and time of modifying a design is high. Design flexibility can be relevant because of several factors, for instance: with radically new products, customers often do not know how to use the product and the meaning of it, they are often not able to imagine the use of the product before they have seen it (the famous quote of Henry Ford underlines this phenomenon: ‘If I had asked my customers what they wanted, they would have said a faster horse’). The benefit of high design flexibility is that firms can make late changes in the design of the product in order to satisfy (customer) needs.

Design flexibility is the overarching flexibility of the design itself, and entails three sub dimensions that determine its degree of flexibility: architecture flexibility, modular flexibility, and resource (coordination) flexibility.

Architecture flexibility

(13)

definition of architecture flexibility says, it is about designing the product in such a way that by adding new functionalities, not the whole product has to be redesigned. Therefore, designs are characterized by low architectural flexibility when the addition of new functionalities to the design results in major changes to other parts of the design. For designs characterized by high architectural flexibility, the addition of new functionalities to the design will not lead to major other changes to the design.

Modular flexibility

The second sub dimension of design flexibility is modular flexibility (Sanchez and Mahoney, 1996). Product designs can be composed of highly integrated, tightly coupled component designs, and this is the traditional engineering design perspective, focused on the highest level of product performance within some cost constraint. However, a more flexible design methodology is to create modular components whose interface characteristics are within the architectural design of the product. This loose coupling of modular components gives the opportunity to create a large number of product variations by simply changing the modular components instead of the whole product. Therefore, a product architecture consisting of modular components is more flexible than a product architecture without modular components.

Resource (coordination) flexibility

Part of the product design are the resources (in terms of technologies) on which the product relies, and Sanchez (1995; 1997) studied resource flexibility on the industry level and divided it into three subtypes:

- The number of alternative uses to which a resource can be applied.

- The incremental cost and difficulties of switching from one use of the resource to another. - The incremental time required to switch from one use of the resource to another.

Thus, resource flexibility is determined by the properties of the resources. Each of these subtypes of resource flexibility increases with an increase in the number of alternative uses and a decrease in the cost, difficulties and time involved in switching from one use of the resource to another. However, only by having flexible resources, firms will not really benefit from it, they have to be able to exploit them. Sanchez (1995, 1997) named this the coordination flexibility, and can be divided in:

- The ability of the firm to define the number of uses the resources can be applied for.

- The ability of the firm to draw on a pool or network of resources and link those resources in a chain of resources capable of being applied in the uses targeted by an organization.

(14)

The flexibility of the coordination of the resources is, like the resource flexibility, determined by the possibilities, cost, difficulties and time involved. Therefore, firms that can relatively easy switch resources for alternative uses and are able to do so (coordination) have a high resource coordination flexibility. Firms that have difficulties with switching from one use of a resource to another have a low resource coordination flexibility.

Workforce flexibility

The last dimension of flexibility in NPD, and studied on the firm level, is the workforce dimension. Workforce flexibility will be defined as: ‘the ability to change the jobs and tasks assigned to workers, their working hours and their number’ (Kok and Ligthart, 2013). This dimension entails four sub dimensions investigated by different authors: functional flexibility, numerical flexibility, telework, and flexible work schedules.

Functional flexibility

Functional flexibility is the first sub dimension of workforce flexibility and involves the range of skills and the degree of reassigning of employees to different tasks (Kok and Ligthart, 2013). The degree of functional flexibility also includes the behavioral flexibility of employees; the willingness of employees to work on different tasks. Van de Ven (1986) links this functional flexibility to innovation by stating that employees should not only be working in their own functional area, but also have an understanding of what occurs in the other functional areas of the development process. Otherwise they will lose sight of the whole innovative effort and this will reduce the effectiveness of their contribution to the innovation. Firms with a high functional flexibility can relatively easy reassign employees to different tasks, and project teams with a low functional flexibility face more difficulties in reassigning employees to different tasks.

Numerical flexibility

(15)

Telework

The third flexibility with regard to the workforce is telework (Coenen and Kok, 2014). Telework means that employees perform tasks elsewhere that are normally done in a central workplace, using information technology to interact with each other. The ability of employees in NPD projects to telework, reflects the flexibility for project members that are dispersed in place and time, to continue cooperating.

Flexible work schedules

Flexible work schedules is the fourth sub dimension of workforce flexibility, and includes the freedom of choice for employees regarding to when to work and when not to work. Thereby, the degree of flexibility is determined by the possibility of employees to choose their own working hours. Connecting telework and flexible work schedules to NPD, Coenen and Kok (2014) state that the widened network of employees available to work on a project increases the variety of knowledge and team performance. Further, employees enjoy the freedom of flexible work schedules, which leads to higher job satisfaction and retention, ultimately having a positive impact on performance (McNall, Masuda, and Nicklin, 2009; Judge, Thoresen, Bono, and Patton, 2001).

Analysis of the literature on flexibility in NPD

As already mentioned at the beginning of this section, there is overlap between some dimensions and/or sub dimensions. Furthermore, not all dimensions are as isolated as what is claimed. The differences and overlap between the dimensions will be discussed in this section and in the end an overview of all dimensions will be given.

(16)

are dependent on each other. Lastly, strategic-decision making flexibility was identified as a new dimension, but also this dimension has overlap with other (sub) dimensions, and probably even is dependent on other (sub) dimensions. Kandemir and Acur (2012) argue that firms are able to ‘maintain multiple simultaneous decision alternatives’ by the direct effects of long-term orientation, strategic planning, internal commitment, an innovative climate, and especially champions and gatekeepers. However, they do not even mention design flexibility, which seems to be relevant too because of for instance working with modular components in order to keep open the multiple decision alternatives.

(17)

Flexibility dimension Description Flexibility sub dimension Description Level of analysis (studied on)

Organizational ‘The degree of structuration in the process design’

Project

Informational ‘The degree of

intersection between the problem-formulation tasks and problem-solving tasks’

Project

Temporal ‘The execution strategies

of development tasks, which refers to the task scheduling: sequential, overlap, or concurrent’ Project Strategic-decision making

‘The extent to which multiple simultaneous decision alternatives are maintained’

Firm

Design ‘The extent to which a

product design can be kept open for change’, and ‘the incremental cost and time of modifying a design’

Architecture The extent to which a product concept can accept the addition of new functionalities without requiring major changes to other parts of the system

Project

Modularity The extent to which a

product concept consists of modular components

Project

Resource coordination

The extent to which resources can be used for alternative uses

Industry

Workforce The degree of flexibility

with regard to the workforce

Functional The extent to which

employees can be reassigned to different tasks

Firm

Numerical The extent to which the

number of employees/hours assigned to a task can be adjusted

Firm

Telework The extent to which

employees perform tasks elsewhere

Firm

Flexible work schedules

The extent to which employees can choose their own working hours

Firm

(18)

Methodology

In this chapter the methodology of the study will be described. The aim of the methodology is to provide this research with a reliable and valid process of collecting and analyzing data.

Research Design

Given the limited theory about how project teams create flexibility in NPD, a multiple-case study was conducted to generate theory (Eisenhardt, 1989). A great advantage of doing a multiple-case study is its possibility of direct replication, resulting in more generalizable theory than single cases (Yin, 2014). Only qualitative data was used to answer the research question, mostly consisting of in-depth interviews with members of R&D projects about how they create flexibility in NPD projects. The interviews were supplemented by archival data, consisting of: design documents, meeting reports, project schedules, and e-mail correspondence. Yin (2014) argues that a number of six between ten cases is, in the aggregate, sufficient to satisfy the requirements of the replication strategy. According to Eisenhardt (1989), usually a number between 4 and 10 cases is sufficient for this type of study. In line with these two authors, this study consists of 6 cases.

Case selection

The focus of this study is on how project teams create flexibility in NPD. There is written about flexibility on the industry, organizational, firm, and manufacturing level, however, little is known about flexibility on the project level. Therefore, six innovation projects were studied at the company VMI Holland, located in Epe. This company employs around twelve hundred people to maintain their position as market leader in production machinery. The R&D department consists of approximately hundred employees. VMI has four business lines: tire (production technology for the manufacturing of tires), can (can washers, washer-ovens and wash coaters), rubber (designing and manufacturing equipment for processing technical rubber) and care (production technology for the manufacturing of cotton pads). VMI operates on the B2B market and make customer-specific products, and is highly suitable for this research, because a lot of different innovation projects, both successful and less successful, are carried out in the last years.

(19)

selection of the cases was carried out together with the vice-president R&D of VMI, who has good knowledge of all the innovations projects carried out at VMI. The selection of the cases was based on the expectations of the project at the beginning of it, in terms of project costs, profit margins, and estimated sales. Whereby the expectations of the two highly successful projects were surpassed, the expectations of the two successful projects came true, and the results of the less successful projects were below expectations. Because of the fact that these expectations were documented, it was relatively easy to select the cases by just comparing them with the final results, which were also documented. The six cases are displayed in the table below (the cases are anonymized).

Innovation project

Description

Highly successful projects Jupiter Duration: 2010-2012

Technology used was new to the company; radical Saturn Duration: 2011-2012

Product is derived from another product; incremental

Successful projects Neptune Duration: 2005-2008

Revolutionary product; radical Mercury Duration: 2014-2015

Retrofit on an existing machine; radical Less successful projects Pluto Duration: 2004-2007

Very radical; many new technologies used Venus Duration: 2007-*

Radical sequel to an existing product

Table 2. the six innovation projects

* The project is (temporarily) halted now, also during the project it was halted multiple times

Data collection

Two data sources were used in this study: (1) archives, including design documents, meeting reports, project schedules, and e-mail correspondence, and (2) semi-structured, in-depth interviews. After these interviews, there was e-mail correspondence with the respondents whereby the interview reports were checked for mistakes and ambiguous issues were further explained. These multiple sources of data resulted in triangulation, which results in greater accuracy (Jick, 1979). The semi-structured interviews were the most important data source for this research.

(20)

project (in most cases a (lead) engineer). Interviews lasted about one hour and topics like the total R&D process, flexibility and results were discussed.

Before the data collection started, an interview protocol was created (Appendix A). However, it should be noted that due to the semi-structured nature of the interview, not all questions were asked, but only served as guidance. All interviews were tape recorded to provide an accurate rendition (Yin, 2014). Also, field notes were made which consisted of overlapping data analysis with data collection resulting in flexible data collection (Eisenhardt, 1989).The first interview was a pilot interview, and the objectives of it were to test the interview protocol and to train the interview skills (van Teijlingen and Hundley, 2002).The archival data consisted of documents and e-mail correspondence about choices made about the design, and functions of new products. This archival analysis is important because some information may not be available in spoken form, but endures in the form of texts giving historical insights (Yin, 2014).

Data analysis

After an interview was conducted, it was written down as soon as possible. These case histories were about twelve pages per case and consisted of a detailed description of the interview, and were eventually supplemented by archival data. E-mails were sent to interviewees to add missing details. All transcribed responses were entered into a database. To cope with the enormous volume of data, within-case analysis was executed. The within-case analyses created familiarity with the cases, resulting in accelerated cross-case comparison (Eisenhardt, 1989). Also, member checks, consisting of asking for feedback on interview reports, were used to determine the accuracy of the findings and interpretations (Koelsch, 2013). With this info, mistakes could be corrected and additional information could be added. The next step was the cross-case analysis, in which was looked for similar constructs and themes in the cases (Eisenhardt, 1989). There were no a priori hypotheses; tables and flowcharts were used to make connections, and shape ideas out of the cases. When the incremental improvement to the theory from iterating between theory and data was minimal, theoretical saturation was reached (Eisenhardt, 1989).

Quality criteria for research

(21)

Validity

Blumberg, Cooper and Schindler (2011) operationalize validity as: ‘a characteristic of measurement concerned that a test measures what the researcher actually wishes to measure; that differences found with a measurement tool reflect true differences among respondents drawn from a population’. Van Aken et al. (2012) and Yin (2014) split validity into three different types: construct validity, internal validity, and external validity.

Construct validity is the extent to which a case study’s measures reflect the concepts being studied (van Aken et al., 2012). Yin (2014) proposes three tactics for increasing construct validity, which are all used in this research: (1) use of multiple sources of evidence (triangulation), (2) establish a chain of evidence, and (3) review the draft case study report with key informants (the vice-president R&D of VMI will review the (draft) case study report).

Internal validity is the ability of a research instrument to measure what it is purported to measure and when the conclusion of a relationship truly implies cause (Blumberg et al., 2011). As Yin (2014) argues, internal validity is mainly a concern for explanatory case studies, and is not very relevant for an exploratory research like this. Nevertheless, internal validity was ensured as much as possible by using pattern matching and explanation building in the data analysis phase.

External validity refers to the generalizability and/or transfer of research results and conclusions to other people, organizations, countries, and situations (van Aken et al., 2012). Since this study focuses on only one company in a specific industry, the generalizability of the results is relatively low. However, the aim of this study was not to provide results that are generalizable to other people, organizations, countries, and situations, and therefore it cannot be seen as a weakness.

Reliability

(22)

Results

In this section, an answer will be given on the research question ‘how do project teams create flexibility in NPD and how does this flexibility influence performance?’.

Design flexibility

When asking the respondents for their opinions on flexibility in NPD, respondents of all projects immediately stressed the importance of flexibility in the product itself. Several respondents appointed the importance of creating a product by which the customer can with more or less options customize his own product, so that the product is flexible in the range of options that customers have with it. Also found important was the flexibility for the project team to adjust (parts of) the product, to make it customer-specific, or to renew (parts of) the product. A major reason to consider this kind of flexibility so important for the project teams, was that almost all new products the project teams create are radical and customer-specific products. A drawback of developing radical products is that it is difficult for customers to clarify their opinions of it beforehand, because no reference products exist (Lettl, 2007). This means that when project teams want to design (radical) customer-specific products, it would be wise to create flexibility in the NPD process, which is in line with the literature on flexibility in NPD (Biazzo, 2009; Cooper, 2008; MacCormack et al., 2001). Taking into consideration the six flexibility dimensions identified in the literature section, the flexibility the respondents found most important, can be best placed in the design flexibility dimension, which deals with the flexibility in adjusting the product. It can be best placed here, because this dimension is really about the product, the other dimensions are more about the preconditions to create the product, for instance a flexible workforce (workforce flexibility), or flexibility in task scheduling (temporal flexibility). Next, the difficulties of the interviewees in separating the different sub dimensions of design flexibility will be discussed.

Difficulties in distinguishing the sub dimensions of design flexibility

(23)

Whereas architecture flexibility (the extent to which the product can accept the addition of new functionalities during the project without needing to changes other parts of the system), and modular flexibility (creating flexibility through the creation of high independence between standardized components) are independent dimensions in the literature, the cases provide a more nuanced thought. Several respondents argued that by creating a modular product architecture, there will automatically be architecture flexibility. Their line of reasoning is as follows: by creating a product, which is composed of multiple modules with interface characteristics that are within the architectural design of the product, adding new functionalities will not be troublesome, because there is independence between the modules. Therefore, by having a modular product architecture (not only the modules of the product, but also within those modules) automatically architecture flexibility will be achieved.

The same applies for resource (coordination) flexibility and architecture flexibility. Resource (coordination) flexibility (the ability to effectively exploit existing resources for other purposes) and architecture flexibility go hand in hand according to the cases analyzed. As a project leader said:

‘The greater the flexibility of the resources, the greater the flexibility will be in adding new functionalities without needing to change other elements of the design’.

Thus, by having resource (coordination) flexibility, also architecture flexibility will be obtained. However, in the literature on design flexibility, as mentioned, the various flexibilities are presented as independent dimensions. Although Thomke (1996) has argued that by having a modular product architecture, architecture flexibility will be higher, no single study has stated that both modular, architecture, and resource (coordination) flexibility are closely related to each other. The insights of the cases analyzed do make this connection, and imply that these sub dimensions cannot be seen as independent dimensions.

Modular flexibility as guiding design principle

(24)

number of product variations by simply changing the modular components. The reasons to take modular flexibility as a starting point for the NPD process to pursue the principle of flexibility in the product itself, are in line with the literature, who also argue for the easiness of making products customer-specific by simply substituting different modular components (Miller and Elgard, 1998; Sanchez and Mahoney, 1996). A quote of the lead engineer of the Mercury project highlights this importance of creating a modular design: ‘it is in our genes to create a modular design, we always do’.

Flexibility within the modules

Dividing the product into modules creates flexibility in the NPD process, but it says little about how the project teams react on environmental turbulence during the process. In order to understand this, it is necessary to look at how flexibility was created within the individual modules. In fact, two levels of modularity can be found when looking at the cases analyzed: first-level modularity, which deals with the partitioning of the product into modules, and second-level modularity, which is about modularity within the modules. An engineer of the Neptune project explains this clearly:

‘First you play with DUPLO bricks to determine the modules of the product, and after that you play with LEGO bricks to further shape the modules’.

By looking at the cases, it was found that the degree of technological and/or market uncertainty was decisive for the degree of flexibility in developing the modules. Like the citation of the engineer of the Neptune project, project teams tried to create modularity within the modules to be flexible and able to deal with the technological uncertainty. It was found that modules about which was uncertainty (radical modules), the project teams followed an iterative approach. In these instances, engineers started developing the module on basis of a ‘rough’ definition of requirements and specifications, and they made modifications and/or improvements in series, until the satisfactory solution emerged. Important to note is that the modular architecture, because of the high independency between the components, facilitated this iteration-based developing, which is in line with the thoughts of Biazzo (2009). The project leader of the Jupiter project said the following:

‘There were a number of requirements for this module, but the subsequent concept was set up through testing and trying things out, actually in a very organic manner’

(25)

‘There really was complete freedom, together with a colleague we divided the module into two parts,

and then we just started developing’.

‘In these cases, I don’t mind when an engineer fails to meet his deadline or budget. The only thing important is that he develops a really good module’.

At modules with less, or even no uncertainty (incremental modules), project teams did create the design by ensuring a modular product architecture, but the process itself was much more structured. In these cases, flexibility was only considered as important in the beginning of the project, the concept phase, resulting in hard deadlines during the project, and less freedom for engineers. A lead engineer of the incremental Saturn project said the following:

‘At a certain point the modules are defined, and you start developing. After the definition of the modules, you do not want to have flexibility in the process, it only costs money, and draws the attention away’.

Thus, the real core of flexibility in the modules of the product was determined by the degree of technological and/or market uncertainty. Both incremental and radical modules were modular composed, but at the incremental modules, the process was much more structured in terms of product definition, deadlines and costs. Regarding the radical modules, project teams followed an iterative-based approach to deal with this uncertainty. The modular architecture of the modules made it possible to follow an iteration-based strategy/reaction strategy, which also was characterized by organizational flexibility (organic process), and informational flexibility (intersection between problem-formulation activities and problem-solving activities, and no restrictions in time and costs). Briefly, the degree of flexibility within the modules can be placed on a continuum, starting with low organizational and informational flexibility at the incremental modules, and ending with a high level of organizational and informational flexibility at the radical modules.

The effects of modularity as guiding design principle

In the previous section the focus was on the modular product architecture itself, but how does having a modular product architecture affect the rest of the NPD process? In this section new insights to the theory will be discussed

The relation between modular flexibility and temporal flexibility

(26)

nuance. At all projects, the members of a project worked together on a central place, but the Pluto project started with a matrix-structure; engineers were working separately on different modules at different departments. After a period without any progression, the project leader decided to take the engineers to a new office and work together there. He stated that when creating a new product, it does not work when employees are spread over departments:

’It is possible to discuss with project members, but in the meantime his boss gives him other tasks. The only thing you can do is isolate the project team at a new place. This makes that you are able to work more closely together and act fast when problems or new information arises’.

Nevertheless, despite the fact that the Pluto team was working now on a centrally place together, the project was delayed because halfway the project, someone realized that some modules, which were worked out by different engineers independently, interfered with each other. The reason that some modules did not fit together was that some engineers exceeded the boundaries within they had to stay in terms of space. The designed modules did not fit together, and had to be redesigned. In contrast to this example, at the Neptune project, one engineer had the overarching task to monitor the development of all the separate modules. His task was to constantly track the whole process to ensure that no modules would interfere each other. This resulted in that during the process no costly or time-consuming adjustments had to be made because of modules not fitting together. This engineer with the overarching task said the following:

‘Engineers want more and more space, but they do not always take into account the rest of the

process, so you have to control them’.

(27)

Modular flexibility and strategic-decision making flexibility

The fourth dimension of flexibility is strategic-decision making flexibility, and this dimension entails ‘the capability that enables firms to make effective strategic decisions by maintaining multiple simultaneous decision alternatives’ (Kandemir and Acur, 2012). The study of Kandemir and Acur (2012) emphasizes the importance of long-term orientation, innovative climate, and strategic planning in order to achieve strategic-decision making flexibility, but they did not mention the importance of the design. However, by analyzing the six innovation projects, it became clear that the design of the product, and especially the modularity of it, also determines the degree of strategic-decision making flexibility (the factors identified by Kandemir and Acur (2012) were also found important). The reasoning behind this is that modularity enables the loose coupling of modular components, to create a large number of product variations by simply changing the modular components instead of the whole product. This provides the ability to maintain multiple decision alternatives during the NPD process, by creating multiple technological solutions (modules) for one problem (part of the machine). At the Mercury project, the project team effectively made use of strategic-making flexibility. The project leader said the following about it:

‘We were not really sure about one module, actually the most important module of the product, the hearth of the machine. Moreover, it was technologically completely new for us, and therefore we decided to develop parallel a module with another technology. We did this because this module of the machine was a real risk for us, if it did not work –it is the hearth of the machine – we must have a fallback’.

Modular flexibility and workforce flexibility

(28)

Looking at the numerical workforce flexibility, which can be split in internal numerical flexibility (adjusting the working hours of employees) and external numerical flexibility (adjusting the number of employees by changing the workforce); the workforce of the project teams analyzed consisted of a fixed core and a flexible layer around it. The fixed core consisted of a project leader and (lead) engineers, who were responsible for the overall design and the important decisions. Around this core was a flexible layer, consisting of temporary workers, and this flexible layer could be adjusted according to the workload. Important to note is that these temporary workers in most cases did ‘the easy tasks’, consisting of for instance the conversion from 2D drawings into 3D drawings. As the project leader of the Saturn project said:

‘The whole project can easily be divided in sub projects, because the product consists of modules which can be separately converted from 2D into 3D. This makes it easy to involve many temporary workers, because they do not need to have knowledge of the other parts of the product but just can do the converting work of their own module’.

An important advantage of the modular product designs was that the work could easily be divided. In the literature no connection was found between the numerical workforce flexibility and modularity, but from the cases emerged that having a modular product architecture, increases the numerical workforce flexibility, by being able to divide the modules into tasks (that can be done by people without needing to understand the other modules).

Flexibility and performance

Part of the research question was to investigate the impact of flexibility in NPD on the performance of the project. Two highly successful projects, two successful projects, and two less successful projects were studied to search for different impacts of flexibility on performance, whereby the performance of the project was based on the ratio between the expectations at the beginning of the project and the results at the end. Looking at the ingredients for a successful project, in particular the development costs, the reliability, and the degree of newness; in terms of functionalities, and accuracy were important. Taking into consideration the flexibility dimensions identified in the literature, the problem-formulation tasks of the informational dimension entails all these ingredients that determine the successfulness of the project (product price, product functionalities and features, technologies on which the product relies, and an allocation of the resources to complete product development).

(29)

a reaction strategy (keeping the product definition open (to some extent), high degree of intersection between problem-formulation activities and problem-solving activities). It was found that this choice for either an anticipation strategy (in situations with little uncertainty), or an reaction strategy (in situations with high uncertainty) was decisive for the ultimate success of the project. The successful Jupiter project followed a reaction strategy, and the lead engineer explained this by:

‘On the basis of feedback, more and more new functionalities were added, making it a very complete product in the end. In the beginning, users could not really imagine how the product would become, so their input was minimal, but during the process they were better able to give their opinion on it’.

In the case of Jupiter, the choice for a reaction strategy suited to the needs of the project, because the technology was really new. However, also examples exist where the reaction strategy did not fit to the needs of the project, for instance the less-successful Venus project, which was a sequel to an existing product, and in which the market needs and technologies used were relatively known. The choice for following a reaction strategy resulted in extremely high development costs, not worth its performance.

Looking at the effective use of the anticipation strategy, this strategy should be chosen when there is a relatively low level of market and/or technological uncertainty. However, at for example the Pluto project, the project team followed an anticipation strategy, and some modules were too ‘sharply defined’, despite the fact that the technologies involved were really new. This resulted in a product that was not fully developed and consequently less-successful. Therefore, the right choice of an anticipation strategy or a reaction strategy, is decisive for the ultimate performance of the project, and should fit to the needs of the project.

Behavioral flexibility

At the literature section, it was stated that besides the ability of employees to work on different tasks (functional flexibility), also the willingness to do so (behavioral flexibility) is of importance. However, no further attention is paid to behavioral flexibility in the NPD literature, but by looking at the cases, several insights emerged.

Interaction flexibility

(30)

At all projects review meetings were held at the early stages of the process, and at these meetings relevant specialists from several departments and projects were invited to give feedback on the design concepts and/or specifications. In some cases, also ‘raw’ design concepts and/or specifications were sent beforehand these meetings. The purpose of these review meetings was to minimize the risks of the forgetting of elements that should be incorporated or just left out, and to adjust practical difficulties in the product. At the Mercury project, the lead engineer presented the design concept to colleagues for relevant departments and said the following about it:

‘After this presentation, everybody was able to contribute to the design concept. I analyzed all the feedback and incorporated the relevant feedback into the design. Doing this is important, because otherwise there is a chance that you have to make costly adjustments halfway through the project’

The extensive presentation and involvement of employees resulted in that during the project no costly adjustments had to be made. In contrast to the Mercury project, at the Jupiter project, there were difficulties regarding the review meeting. As the lead engineer of Jupiter said:

‘I wrote the design specification and sent it to relevant colleagues for feedback, but I did not receive any feedback. After that, a review meeting was held but still nobody had read the design specification, so the input on it during this meeting was minimal. Nevertheless, we started developing and during the project several hurdles had to be overcome because other departments came up with new functionalities that had to be involved’.

The insufficient feedback on the design concept and specifications at this project resulted in an exceeding of the time and the costs budgeted.

These two cases are striking examples of the importance of interaction in NPD. Employees must be flexible in the sense of willing to interact with each other, and flexible in involving relevant colleagues in problem situations. Some employees have difficulties in asking colleagues for their opinion, they believe that they are doing it right and see it as a weakness to ask for help or opinion. On the other side, employees must be flexible in helping colleagues with feedback and should spend sufficient time on that. They must be willing and motivated to stop for a moment with their own work and spend sufficient time on evaluating the other project and give useful feedback on it.

Flexible thinking

(31)

case of the Mercury project, this led to having multiple back-up solutions for problems and thereby minimizing the risk of not incorporating the best solution for a problem. The message of this type of flexible thinking is that employees should not stick to one solution, but constantly have to keep in mind other possible solutions.

It was uniformly agreed that most engineers are not able to be really objective regarding their own design, and that they often are afraid of face loss when they have to quit with their solution. As a respondent said:

‘Engineers are always convinced of their own solution, even though colleagues say that they are doing it wrong, engineers can hardly admit’.

These situations often lead to wasting time; engineers are ‘muddling along’ with their solutions, and they only stop when their boss tells them to stop. Important to note is, however, that at the more organic projects, this ‘muddling along’ was more a problem than at the structured projects, in which was more control and less freedom to ‘muddle along’.

Another form of flexible thinking was found at the Mercury project; shortly before the prototype was finished, the design team decided to change the design. They replaced a component in the product, because they thought that this would make the product more customer-friendly. The lead engineer of the Mercury project said the following:

‘When you are designing, you constantly have to put yourself in the role of the customer, the pitfall is that you realize this just at the end’.

Thus, when engineers are able to constantly put themselves in the role of the customer, this can prevent expensive modifications at the end of the project. This is in line with the thoughts of Day and Moorman (2010), who also argue that customer benefits must be the criterion against which decisions are made.

Flexibility in using external resources

(32)

The cases provided the behavior of employees as an important factor for the inflexibility to use resources from others. Several respondents, including the vice-president R&D of VMI stated that there are two reasons for this:

- The first reason is that engineers do not always fully understand technological solutions invented by others, and thereafter, they do not ask for explanation to the inventor of the solution (which can be linked to interaction flexibility). Several respondents mentioned that occasionally the ‘wheel is re-invented’ in project teams. Obviously, this leads to costs that could have been avoided.

- The second reason and likely the most characteristic of behavior flexibility is that engineers do not like to use technological solutions invented by others, they believe that their own solutions are superior to the solutions of others. This behavior can be seen as the not-invented here syndrome (NIH), which can be defined as: ‘the tendency of a project group of stable composition to belief it possesses a monopoly of knowledge in its field, which leads it to reject new ideas from outsiders’ (Katz and Allen, 1982).

(33)

Discussion and conclusions

Flexibility in NPD is needed to deal with market and/or technological uncertainty. Extant research has identified dimensions of flexibility, but these were found at different levels of analysis, and little is known about how this flexibility exactly is created at the most important level of analysis: the project-level of analysis. This paper fills this gap and has several important contributions to the project flexibility literature.

Theoretical implications

The results show that in contrast to the flexibility literature, in which architectural, modular, and resource (coordination) flexibility are presented as independent dimensions (MacCormack et al., 2001; Sanchez, 1995, 1997; Sanchez and Mahoney, 1996), in practice these sub dimensions of design flexibility cannot be seen as independent dimensions. Respondents mixed these dimensions, and argued that they are all closely connected to each other, and they fully influence each other. This suggests the following proposition:

Proposition 1. The sub dimensions of design flexibility: architectural, modularity, and resource

(coordination) cannot be seen as independent dimensions, but should be viewed as a whole, because they are so closely connected to each other.

The results further show that the degree of technological and/or market uncertainty was decisive for the degree of flexibility in the NPD process. At all projects, project teams ensured flexibility in the design by creating a modular product architecture, consisting of first-level modularity (partitioning the product into modules) and second-level modularity (modularity within the individual modules). Whereas previous studies (Sanchez and Mahoney, 1996; Thomke, 1996) argue that modularity is mostly used in conditions of uncertainty, the results of this study imply that even in stable conditions with low uncertainty project teams emphasize the importance of modularity, for instance in order to create a flexible product for the customer. Nevertheless, it was found that at the radical modules, facing more uncertainty, a higher level of informational and organizational flexibility was used, resulting in an organic, iteration-based development process. At the incremental modules, the process was much more structured, with less freedom (e.g. in time and costs), implying a low degree of informational and organizational flexibility. This suggests the following proposition:

Proposition 2. Modular flexibility is beneficial regardless of the environmental turbulence, but at

the radical modules there was a higher degree of organizational and informational flexibility than at the incremental modules.

(34)

that modularity enables temporal flexibility, in which each module can be developed simultaneously, possibly geographically dispersed (Biazzo, 2009). Moreover, several authors (Sanchez and Mahoney, 1996; Spender and Grinyer, 1995) say that controlling the required output of component development processes results in an effective development process, without the need of continual exercise and managerial authority. However, the results nuance these thoughts, and imply that working geographically dispersed is totally not effective, and that there has to be an employee responsible for monitoring all the module development activities, so that modules will not interfere each other. Secondly, whereas the study of Kandemir and Acur (2012) emphasizes the importance of long-term orientation, innovative climate, and strategic planning in order to achieve strategic-decision making flexibility, from the cases became clear that modularity is also an important factor in determining the degree of strategic-decision making flexibility. And thirdly, it was found that modularity enables workforce flexibility. Other studies (e.g. Kok and Ligthart, 2013) focused on making the workforce more flexible, for instance by using training and education, in order to achieve flexibility in NPD. However, from the cases emerged that by creating a modular product architecture, the workforce (both functional and numerical) will become more flexible. Hence, it could be hypothesized that:

Proposition 3. Having a modular product architecture, positively influences temporal flexibility, but

project teams working geographically dispersed is detrimental, and there should be a team member responsible for monitoring all the module development activities.

Proposition 4. Having a modular product architecture, positively influences the degree of

strategic-decision making flexibility the project team has.

Proposition 5. Having a modular product architecture, positively influences the degree of workforce

flexibility the project team has.

The current literature on flexibility does not connect the different dimensions of flexibility to performance. Factually, the literature states that flexibility is beneficial in any case, but on basis of this research, it can be said that flexibility is absolutely not beneficial in any case, but should fit to the needs, in terms of technological and/or market uncertainty, to the project. It was found that the degree of informational flexibility, which is about the intersection between problem-formulation activities and problem-solving activities is decisive for the success of the project. This flexibility dimension deals with the extent of change in terms of specifications/functionalities of the product, the technologies used in the product, and the price of the product, which are all important

(35)

and/or technological uncertainty need low informational flexibility (anticipation strategy). This suggests the following proposition:

Proposition 6. The degree of informational flexibility is decisive for the ultimate success of the

project, and should fit to the needs of the project in terms of market and/or technological uncertainty.

The last contribution to the literature on flexibility in NPD is the addition of a new dimension: behavioral flexibility, which consists of three sub dimensions. The first sub dimension that was found is interaction flexibility, and deals with the willingness and openness of colleagues to interact with each other in problem situations, and to ask for feedback. On the other side, employees must spend enough time on giving feedback when colleagues ask for it. Flexible thinking is the second sub dimension, and means that employees should not stick to one solution, must be able to put

themselves in the role of the customer, and must be able to be objective regarding their own design. The last sub dimension of behavioral flexibility is flexibility in using external resources. Whereas the literature highlights the importance of the technological nature for resource (coordination) flexibility (Sanchez, 1995; 1997; Yuan, Zhongfeng, and Yi, 2010), the cases made clear that the behavior of employees also plays an important role in this. It was found that employees do not always fully understand technological solutions invented by others, and then refuse to ask their colleagues for explanation. Further, employees do not like to use technological solutions of colleagues, they believe that their own solutions are better, implying the not-invented here syndrome (Katz and Allen, 1982). Hence, it could be hypothesized that:

Proposition 7. Behavioral flexibility, consisting of: interaction flexibility, flexible thinking, and

flexibility in using external resources, is important in NPD in order to prevent unnecessary expenses.

Managerial implications

(36)

Also, despite the fact that behavioral flexibility was not identified as a dimension of flexibility in NPD in the literature, it emerged as an important dimension in the cases analyzed. Managers should be aware of the importance of behavioral flexibility of employees, for instance in terms of flexible thinking, which implies that employees should not only focus on their own solution, but also have to keep in mind other possible solutions. Management can encourage behavioral flexibility by for instance organize trainings in the use of different modes of thinking (alternatively using divergent and convergent thinking) (Georgsdottir and Getz, 2004).

Limitations and future research

As with any research, this study has several limitations that should be acknowledged. First, the Venus project is labeled as a less-successful project, but it is still running and consequently, there is always a chance that this label is not accurate anymore in future. Second, the data was collected through semi-structured, in-depth interviews. Deeper and additional insights would be acquired by also observing NPD projects for an extended period. This would particularly be beneficial for the behavioral dimension, as ‘live’ behavior is much more accurate than that described in interviews. Third, especially for the less-successful projects, respondents had difficulties to explain the weaknesses of these projects. A possible reason for this that due to their functional affiliations, these respondents could be predisposed toward their functions, and thereby giving biased answers. The effect of this limitation has been reduced by asking all respondents for their opinion on the less-successful projects.

Creating and managing flexibility in new product development