Managing Complex NPD Projects with Strict Time Constraints: a Case Study

Managing Complex NPD Projects with

Strict Time Constraints: a Case Study

By: Annemiek Sterks

S1839721

Powered by:

KXA Project Professionals

University of Groningen

Faculty of Economics and Business

MSc BA – Business Development

C. Reezigt

J.D. van der Bij

Managing Complex NPD Projects with

Strict Time Constraints: a Case Study

Annemiek H. Sterks

University of Groningen, Faculty of Economics and Business, Specialization Business Development 2 September 2010 Abstract Keywords Project Management; Complexity; Uncertainty; Time-to-Market; Centralization; Decentralization; Process models; NPD.

Worldwide there is an increasing interest in new product development (NPD). These NPD projects are becoming more and more complex. See for example inter-firm partnerships and the ongoing introduction of new techniques. Besides, the environment is changing fast and competition intensifies which asks for accelerated product development processes. In current literature there are discussions about how to manage complex NPD projects on the one hand and NPD projects with strict time constraints on the other hand. In this study we research how a project manager has to manage a NPD project that is both complex and subject to strict time constraints. Especially we will focus on the sequence of the activities in a formal process model and on the type of decision-making process. Propositions are formulated and tested in a case study. We conclude that a combination of a centralized and decentralized decision-making structure, and a partly-parallel process model is required to manage such NPD projects.

1. Introduction

Over the last few decades, new product development (NPD) has become more and more important. Nowadays, it is the driving force for attaining competitive advantage and sustained growth of the firm (Chang & Cho, 2008, p. 13). At the same time competition is intensifying, new technologies are being developed quickly, and markets are shifting radically. To handle this fast changing environment, an accelerated NPD process is necessary in which speed becomes a key factor in a firm’s degree of competitiveness (Sun, Zhao & Yau, 2009).

Moreover, projects become more complex. Geographical borders are disappearing which leads to expanding markets to be served. Furthermore,

there is a high need for expertise and experience because techniques become more complex. To handle the demands of the global market and the more complex techniques, firms cooperate more and more to achieve success in NPD projects.

Complex projects and projects in which time-to-market is important, are both well known in literature. There are plenty of management implications for managing a project in a timely manner; furthermore there are many management implications for successfully managing a complex project. However, what to do if you have to manage a very complex NPD project with strict time constraints? This is not simply the sum of all implications given in literature. This is potentially a completely different type of project.

In this article we focus on the coordination of complex NPD projects with strict time constraints. We try to find an answer to the questions which decision-making approach will be best and how to structure such a project. Before we are going into answering these questions we provide definitions of both complexity and time-to-market in section 2. Especially complexity is a term with many different interpretations, and therefore it is treated rather extensively in order to derive a proper definition within the context of this research.

In section 3, we look for contradiction in literature about managing complex NPD projects on the one hand and NPD projects with strict time constraints on the other hand. Two of these contradictions are further investigated in section 3.1 and 3.2, where after propositions are set about how to deal with these contradictions in complex NPD projects with strict time constraints. In section 4, these propositions are tested in a case study in which we use practical observations and interviews. We consider a case in which a project is executed by a special created network of different small and medium sized enterprises (SMEs) that are developing a radical innovation.

2. Explanation of the concepts

This section provides the theoretical background for the definitions used. First we present a review of the literature regarding ‘complexity’; next we derive a definition of ‘time-to-market’.

2.1 Complexity

As indicated in the introduction, there are many different definitions of complexity used in different fields of studies; physics, chemistry, cybernetic research, biology and organizational research (Alhadeff-Jones,

2008). In this research, we focus on the definition of complexity in organizations, and more specifically in NPD projects. Alhadeff-Jones (2008) offers a chronological presentation of three generations of theories that are shaped around the term complexity. In the 14th century the term complexity was born in the Latin expression complexus, meaning embracing or comprehending several elements (Simpson, 2005). This term of complexity was, and still is, used in many different fields, like a complex fraction in mathematics or a complex note in music. The term complexity was first used especially in physiology in the 18th century, later it migrated to economy, chemistry, geometry, biology, medicine, psychology and psychoanalysis (Institut National de la Langue Française, 2005). Each field of study uses its own definition, which leads to a great variety in definitions of complexity.

At first, complexity was related to the components of a system. Later, the complexity is seen as a fabric of relations (Bachelard, 1934/2003). Weaver (1948) identified the terms ‘disorganised’ and ‘organised’ complexity. Disorganised complexity is the complexity of a process which cannot be reduced by separating it into smaller parts. With organised complexity Weaver (1948) means ‘all problems which involve dealing simultaneously with a sizeable number of factors which are interrelated into an organic whole’ (p. 5). To reduce this organised complexity, the components were divided into smaller manageable parts (Beer, 1959; Churchmann, Ackoff & Arnoff, 1957). In addition, the use of mixed-teams was a good way to handle organised complexity (Weaver, 1948).

Around 1945 the complexity ‘theory’ separated into two directions. One direction focussed on relationships and the possibility to adapt a system based on the input. The other direction takes into account all different components of a process (Alhadeff-Jones, 2008). These two directions differ in dynamism, in which the first direction is more dynamic than the second. Another, even more dynamic view on complexity was born in the early 1970s, called the chaos theory (Gleick, 1987). ‘Chaos is an a-periodic, unpredictable behaviour to variations in initial conditions’ (Singh & Singh, 2002, p. 23). A-periodic means that the same state is never repeated twice (Wilding, 1998). This makes a chaotic system very instable (Ruelle, 1991) and impossible to predict (Wilding, 1998). This unpredictability has everything to do with the lack of a complete system of equations of causes and effects, to describe the behaviour of systems. If these equations were known, the behaviour of the system could be predicted. Although there is a relationship between causes and effects, science has not yet developed to the level to determine it (Singh & Singh, 2002).

Years passed by in which the complexity theory becomes more supported. However, nowadays the definition of complexity is discussed once again, because some authors see complexity, described as the number of components and relationships, as ‘complicated’ or ‘hyper-complicated’ (Rogers, 2008).

Glouberman and Zimmerman (2002) introduce a distinction between simple tasks, complicated tasks, and complex tasks. An example of a simple task is following a recipe. Following a recipe could cause some basic issues of technique or terminology. However, if these issues are mastered, following a recipe has a very high success rate.

Sending a rocket to the moon is a complicated task. Complicated tasks contain subsets of simple components but are not merely reducible to them (Glouberman & Zimmerman, 2002, p. 1). Complicated tasks also contain issues of coordination and they do have a need for specialized expertise.

Raising a child is a complex task. Complex tasks do have unique conditions, interdependencies (Holland, 1995) with the added attribute of non-linearity (Lorenz, 1993), and the capacity to adapt to changing conditions (Kauffman, 1995; Kelly, 1994). Moreover, complex tasks have large elements of ambiguity and uncertainty (Wheatley, 1992).

These terms and definitions are not the only descriptions used in the world of science. Many researchers use their own definition of complexity and also use their own term; aggregate complexity (Manson, 2001); static complexity (Deshmukh, Talavage & Barash, 1998; Wood, 1986), dynamic complexity (Bozarth et al., 2009; Deshmukh, Talavage & Barash, 1998; Kellen & Stefanczyk, 2002; Wood, 1986), institutional complexity (Heywood, Spungin & Turnbull, 2007), individual complexity (Heywood, Spungin & Turnbull, 2007), task complexity (Wood, 1986), system complexity (Simon, 1996), detail complexity (Bozarth et al., 2009; Senge, 1990). However, it is not that important how you name your type of complexity but it is more important how you define this term. Complicated, static complexity, detail complexity, organised complexity, institutional complexity, do all have similarities in their definitions. Even as dynamic complexity, aggregate complexity, system complexity, and chaos theory have more or less the same definition.

Two clusters can be selected from these definitions. This leads to a distinction between two types of complexity: dynamic complexity and detail complexity.

In dynamic complexity there is a general consensus that the system is nonlinear, a-periodic, and unpredictable (Bozarth et al., 2009; Casti, 1979; Daft & Lewin, 1990; Johnson & Burton, 1994; Singh & Singh, 2002;

Yates, 1978). A dynamic complex system is a system in which input leads to unexpected outcomes. So the behaviour is surprising and hard to predict because of its nonlinearity (Daft & Lewin, 1990). In a nonlinear system, changing one or two parameters only with a small amount can drastically change the behaviour of the entire system, in which the whole can be very different from the sum of its parts (Daft & Lewin, 1990).

Detail complexity, also known as static complexity, is about the number of distinct components or parts that make up a system and the relationships (Bozarth et al., 2009). It involves situations where cause-and-effect are subtle, and time-effects are not obvious (Senge, 1990). Daft (1992) has applied this type of complexity at an organizational level, where he equates complexity with the number of subsystems or activities within the organization and the relationships among them.

Both types of complexity lead to uncertainty. Uncertainty is also a term with different definitions and types, like market uncertainty (Little, 2005), technical uncertainty (Little, 2005), unforeseeable uncertainty (Sommer & Loch, 2004; Sommer, Loch & Dong, 2009), internal uncertainty (Kolltveit, Karlsen & GrØnhaug, 2004; Perminova, Gustafsson & WikstrÖm, 2008; Sicotte & Bourgault, 2008), external uncertainty (Kolltveit, Karlsen & GrØnhaug, 2004; Perminova, Gustafsson & WikstrÖm, 2008), task uncertainty (Mathiassen & Pedersen, 2008; Galbraith, 1973) and so on. However, the general definition of uncertainty as a whole is, ‘the difference between the amount of information required to perform the task and the information already possessed (Galbraith, 1973, p. 5).’ Yet, a distinction could be made regarding this definition, because just like with complexity, uncertainty can be predictable or unpredictable.

Unpredictable uncertainty, also named unforeseeable uncertainty, could be defined as ‘the inability to recognize and articulate all relevant variables affecting performance’ (Sommer, Loch & Dong, 2009, p. 118). So, ‘uncertainty refers to events for which it is impossible to specify numerical probabilities’ (Perminova, Gustafsson & WikstrÖm, 2008, p. 75). Also external uncertainty is in line with this explanation of uncertainty. Sicotte and Bourgault (2008) claim that the external environment is unpredictable and that it may change and affect the organization randomly. Predictable uncertainty refers to situations where a team could have foreseen events by doing its homework (Sommer & Loch, 2004). Courtney, Kirkland and Viguerie (1997) describe four different levels of uncertainty, which can also be divided into predictable uncertainty on the one hand and unpredictable uncertainty on the other hand (figure 1).

Figure 1: Four levels of uncertainty (Courtney, Kirkland & Viguerie, 1997, p. 70-71)

The world is either certain, and therefore open to precise predictions about the future (level 1 in figure 1), or uncertain, and therefore completely unpredictable (level 4). Besides, a range of potential outcomes could be possible (level 3) or even a set of scenarios (level 2). Level 2 and 3 are also in a certain way predictable, only level 4 ‘true ambiguity’ is a scenario which is completely unpredictable. This scenario, however, is fortunately also scarce. According to the four levels of uncertainty presented by Courtney, Kirkland and Viguerie (1997), a distinction could be made between dynamic complexity which leads to unforeseeable uncertainty (level 4) and detail complexity which leads to predictable uncertainty (level 2 and 3). Dynamic complex systems are non-linear, a-periodic, and unpredictable, in which input lead to unexpected outcomes. Therefore it is impossible to predict the future and so the system has a high level of unpredictable uncertainty. Detail complex systems have many distinct components and relationships. Nevertheless, these components and relationships are known or could be known by doing some homework. The goal is clear, and the options to achieve that goal are clear within a certain range. This leads to the possibility to ‘predict’ the future if the homework is done. However, also in detail complexity you will have to keep in mind the possibility of dynamic relationships. We simply cannot predict everything correctly, even not the simplest task.

All NPD projects are more or less unique and complex undertakings (Perminova, Gustafsson & WikstrÖm, 2008). In this paper we will focus on detail complexity and uncertainty caused by a lack of information which could be reduced by doing the necessary homework. Actually it is rather impossible to make a clear distinction between detail complexity and dynamic complexity due to the limitations of the human brain to determine if a system is more dynamic or static. This is due to the non-linearity and unpredictability of a dynamic complex system. As a consequence, focussing

on detail complexity does not mean we could totally forget dynamic complexity. A highly static complex system could be perceived as having no dynamic relationships, but you always have to keep in mind the possibility of dynamic relationships. Moreover, because dynamic complex systems are non-linear and therefore unpredictable it is for example impossible to make it manageable by separating the system into smaller parts. The only way to handle dynamic complexity is to be flexible and have an outward-facing perspective on the environment to try to manage the unexpected. Because detail complexity is more tangible and projects with detail complexity are more common, we will focus on detail complexity in NPD projects.

In conclusion, detail complex NPD projects consist of many different parts and relations between those parts. These parts are not only the number of parts the product consist of and its required actions and information cues (Wood, 1986), but also the amount of stakeholders, the team size, the team location and team maturity, the domain of knowledge gaps and the dependencies among each other (Little, 2005).

2.2 Time-to-Market

As claimed before, time-to-market is very important in current NPD projects. This is due to the competitive advantage a company could have by reducing their development time and by taking the first mover advantage (Kessler & Chakrabarti, 1999). Other reasons are that product life cycles are shrinking, and competition has intensified (Griffin, 1997).

It seems that a project manager has four project objectives namely ‘1) schedule time, 2) development project expense, 3) unit manufacturing cost of the resulting product, and 4) product performance’ (Smith, 1999, p. 225). Of course all these objectives are really important, however in reality project managers trade-off these four objectives against each other. According to Smith (1999) focussing on cycle time could lead to the most beneficial outcome. In this, employing resources effectively is very important to shorten the cycle time, so there is a need for high productivity. ‘In addition, a fast system does not allow for unnecessary rework and redesign, thus poor quality’ (Smith, 1999, p. 229). Everything has to work flawlessly.

In this research we define time-to-market, also known as speed-to-market (Griffin, 1997) or innovation speed (Kessler & Chakrabarti, 1999) as ‘the time elapsed between a) initial development efforts, including the conception and definition of an innovation, and b) ultimate commercialization which is the introduction of a new product into the market place’ (Kessler & Chakrabarti, 1999, p. 231-232). This is also known

as ‘concept to customer’ (CTC), the time period between the approval of the innovative idea through the introduction of the product onto the market (Griffin, 1993).

3. Contradictions in literature

Having defined complexity and time-to-market, the question arises how to manage such projects? Besides, what makes it so difficult to manage both complexity and time-to-market in NPD projects? First of all, a project can be defined as a unique interrelated set of tasks with a beginning, an end, and a well defined outcome (PMI Standard Committee, 1996). Here we focus on NPD projects. There are various reasons to develop a new product, and thus to undertake a NPD project, 1) to meet market or customers demands, 2) as a reaction to a competitors offer, 3) to take advantage of a technical development, 4) because of a strategic decision of management (Griffin, 1993). Moreover, most projects have restrictions in time, cost and scope and also have certain quality demands. Additionally, projects do have high levels of uncertainty which can have both positive and negative effects (Perminova, Gustafsson & WikstrØm, 2008).

Mahmoud-Jouini, Midler and Garel (2004) characterize a project by two curves; a learning curve which represents the increase in knowledge about the project and a decision-making curve, representing a reduction in the possibilities of actions during a project. These curves are indicated in figure 2. According to Mahmoud-Jouini, Midler and Garel (2004) it is necessary, if you want to accelerate a project, to take time at the beginning of a project to explore and prepare project options as thoroughly as possible before deciding on them and putting them into practice (p. 360-361).

The beginning of a project is very uncertain because of the lack of necessary information and the many possible options to complete the project. Especially NPD projects deal with this lack of necessary information due to the newness of materials or techniques which are used during the development of a new product (Kim & Wilemon, 2003). During the project the required knowledge is obtained, and actions are taken, to eventually complete the project (Mahmoud-Jouini, Midler & Garel, 2004). Projects with such high uncertainty could better be described as ‘journeys of exploration in a given direction, rather than strict plan-following endeavours’ (Perminova, Gustafsson & WikstrØm, 2008, p. 74).

Managing requirements of such complex NPD projects that also have strict time constraints are not simply determined by the sum of the managerial implications already given for complex NPD projects on the one hand, and

Figure 2: Knowledge curve and action curve of a project (Mahmoud-Jouini, Midler & Garel, 2004, p. 361).

NPD projects with strict time constraints on the other hand. This is caused by some contradictions between the managerial implications of both phenomena. Two of these contradictions are selected for further investigation. These two contradictions are chosen because they are indicated as important success factors in NPD.

The first contradiction is in the domain of the overall NPD process. To manage a NPD project there is a need for some kind of process. A distinction is made among three process models with different sequences of executing activities; parallel, partly-parallel, or sequential. In the next section (3.1), we try to find out which type of process model will be best for complex NPD projects with strict time constraints.

Besides, it is also important how you cooperate within your team, which partly depends on the type of process model you use. To look closer at this cooperation within the team, we focus on the decision-making process in section 3.2. The decision-making process shows how participative a team is and how interactions flow.

3.1 Process models

Each project must deal with a certain degree of uncertainty. One way to handle uncertainty in NPD projects is hierarchical decomposition (Simon & Cilliers, 2005). Hierarchical decomposition is the process of decomposing a complex project into simpler and more manageable tasks that are processed separately and later integrated into a finished product (Duimering et al., 2006). In this way the team is also broken into sub teams. Each sub team

will focus on one or a few sub problems (Leenders, van Engelen & Kratzer, 2007). This is what makes it possible to reduce the number of components you have to deal with all at once and thus reduces detail complexity. However, there are different ways how to process these simpler and more manageable tasks.

The NPD process consists of many activities that could be conducted simultaneously or sequentially (Song, Thieme & Xie, 1998). Therefore, there are various process models that differ in the sequence of activities. Nevertheless, another option is to have no formal process at all. In a survey that was completed in 1995, 36% of the questioned firms, in a broad mix of product industries, did not use a formal process, even though these firms were successful in product development (Hustad, 1996). Besides, there are many firms that do have a formal process but do not follow it. So maybe a formal process is not necessary to be successful. However, in a survey among 343 projects in 11 companies that operated in five different industries it seems that ‘formal NPD processes interact with product complexity to affect product development cycle time. Specifically, using a formal process in more complex projects will decrease cycle time more than in less complex projects’ (Griffin, 1997, p. 27). So, a formal process is expected to be a success factor in complex projects with strict time constraints.

In section 3.1.1 a process model is described to manage complex NPD projects. Next, in section 3.1.2, a process model is described for managing NPD projects with strict time constraints. In section 3.1.3 a proposition is stated about which process model is best to manage a complex NPD project with strict time constraints.

3.1.1 Process models in complex NPD projects

In NPD projects with high complexity a sequential process could be used (figure 3). This sequential process, also known as the phase-review-process (Hughes & Chafin, 1996), was developed by NASA in the 1960s to deal with the complexity of the project to put humans in space (Griffin, 1997). A sequential process is a ‘process in which one phase of effort is completed by one function and the results are then passed on to the next function to complete the next phase of effort’ (Griffin, 1997, p. 27). NASA did need this sequential process to keep track of all interrelated tasks that need to be organized for completing such a complex project in a timely manner. Also Hewlett Packard is positive about the phase-review-process because it brings discipline into an otherwise chaotic, ad-hoc activity. In this way technical risks were reduced and it did ensure completion of the tasks

Figure 3: Phase-review-process (Hughes & Chafin, 1996, p. 92)

(Cooper, 1994). Sequential processes are very engineering driven (Verworn & Herstatt, 2002). ‘The sequential process applied strictly to the physical design and development of the product’ (Cooper, 1994, p. 4). Marketing people were not included in the schema (Cooper, 1994). However, a lot of companies did improve the effectiveness and efficiency of their product development cycle thanks to this formal process model (Cooper, 1994). Nevertheless, there are also some drawbacks. First of all, activities are put on hold to wait for other activities to be completed, before the next management review (Verworn & Herstatt, 2002). In addition, R&D engineers fail to communicate directly with manufacturing engineers during phase 1 and 2. This causes that the product design proceeds without considering manufacturing requirements (Schilling & Hill, 1998). Besides, a sequential process does not have an early warning system to indicate that some features are impossible to be manufactured (Schilling & Hill, 1998). Consequently, this can lead to delays. Furthermore, the sequential process only dealt with the development of a product and not with the whole process from idea to launch (Verworn & Herstatt, 2002).

3.1.2 Process models in NPD projects with strict time constraints

A well known project management method for projects where speed-to-market is most important, is concurrent engineering (Cordero, 1991; Duffy & Salvendy, 1999; Mahmoud-Jouini, Midler & Garel, 2004; Verworn & Herstatt, 2002). This method reduces project delay by planning parallel activities on the same project and, by using cross-functional teams (Mahmoud-Jouini, Midler & Garel, 2004). In a parallel process model (figure 4) the distinction between the phases that have to be completed traditionally, are totally blurred (Cordero, 1991). In addition, ‘concurrent engineering is concerned with the timely availability of critical design information to all development participants’ (Yassine & Braha, 2003, p. 165). It is very difficult to collect all this critical design information that is required at the start of a task. Besides, it is a risk using early upstream information, because this could cause rework if information changes (Yassine & Braha, 2003). Moreover, a parallel process will lead to an

Figure 4: Simultaneous development phases (Crawford, 1994, p. 27)

increase of complexity compared to the sequential process model because more activities have to occur in a specific period of time (Kim & Wilemon, 2003). So speeding up the NPD process will lead to higher complexity.

3.1.3 Process models in complex NPD projects with strict time constraints

Sequential process models are well suited to deal with complex projects, and parallel process models are well suited to deal with projects with strict time constraints. Nevertheless, sequential process models take a lot of time, and parallel process models entail greater complexity in managing the stages and activities. This leads to the fact that both process models are not suitable for managing complex NPD projects with strict time constraints. However, there are process models that manage activities partly-parallel.

Because of the increasing importance of speed-to-market in NPD, and because of the shift from technology-push towards market-pull, which resulted in the greater importance of a combination between the engineering perspective and the marketing perspective (Rothwell, 1994), the second generation process model (Cooper & Kleinschmidt, 1990) was born (figure 5). In this model the innovation process is also broken into more manageable stages. Between the stages, go/kill decisions are made. The first big difference compared to the phase-review-process is using cross-functional teams (Cooper, 1994). The go/kill decisions are also made by these cross- functional teams. This will reduce uncertainty and improve quality of NPD decisions (Moenaert & Souder, 1990). In addition, the second generation process model integrates marketing and manufacturing, so there are no

Figure 5: Typical second generation stage-gate-model (Cooper & Kleinschmidt, 1990, p. 46)

stages which are ‘owned’ by one function. R&D, sales and marketing, and manufacturing people are all full-time players on the project team. Besides that, this model covers the whole innovation process from idea to launch. For example, up-front homework is very important. While in the phase- review-process it was just assumed that the up-front homework did occur, it is a separate topic in the second generation model. This leads to a much stronger market orientation, where customers become an integral facet of the NPD process (Cooper, 1994). Also, more activities are undertaken in one phase compared to the review-process (Cooper, 1994). The phase-review-process divides the development process into 4 stages while the second generation process puts all these activities in only one stage. Therefore more parallel activities are permitted to speed up the process (Verworn & Herstatt, 2002). But also this process has its drawbacks. As in the phase-review-process also in the second generation process projects must wait at each gate until all tasks have been completed. Of course this ensures the quality and satisfactory of critical tasks. However, this could also lead to expensive delays. Besides, projects must go through all gates and stages. This also is a control feature to reduce product failures, however, for some projects it is unnecessary to blindly follow all stages. As a result, this second generation model is transformed into a third generation model (Cooper, 1996). This model is even more partly-sequential, in that the stages also overlap and not only the activities (figure 6). The third generation model is more focussed on efficiency. Cooper (1994, p. 9) represents the third generation model by four fundamental F’s; 1) Fluidity, 2) Fuzzy Gates 3) Focused, 4) Flexible. In which fluidity is referred to as a fluid and adaptable process, with overlapping and fluid stages for

greater speed. Fuzzy gates stands for the go-decisions which are dependent on the situation. Focused, refers to the fact that too few resources are dissipated across to many projects. Focussing on the entire portfolio will place the resources on the ‘best bets’ (Cooper, 1994, p. 9). And lastly, it is more flexible than the second generation stage-gate-model. ‘Each project is unique and has its own routing through the process’ (Cooper, 1994, p. 9). It is not strictly sequential at all. The rules of the second generation stage-gate-model are changed into guidelines. The borders between the stages become unclear (Verworn & Herstatt, 2002).

Consequently, to deal with the contradictions between managing speed and complexity in NPD projects it is expected that a partly-parallel process model is required. In this way cycle time will be accelerated compared with the sequential process model. However, it will be slower than the parallel process model. Nevertheless, because the parallel process model makes the project even more complex, it is expected that this could also lead to more rework and redesign and thus time delays. That is why a partly-parallel process is chosen to deal with both time and complexity. It is important to know that this process model needs careful management, and a thoughtful game plan to complete the project successfully (Cooper, 1990).

There is a distinction between the second generation stage-gate-model and the third generation stage-gate-model. In NPD it is all about innovation. We focus on more radical innovations. In radical innovations, the newness of the product is high which leads to a need for a high level of creativity. To be creative you need flexibility. In addition, in NPD projects that develop a radical innovation, uncertainty is even higher which demands more flexibility because new information could ask for

changes in the process. However, the process becomes also more complex because of this flexibility in activities and relations. That is why management becomes even more important in a complex NPD project with strict time constraints. Another advantage of the usage of the third generation process model, in comparison to the second generation model, is that the development time will be even less. Thus we formulate:

Proposition 1: Complex NPD projects with strict time constraints require a

formal partly-parallel process equivalent to the third generation stage-gate-model.

3.2 Decision-Making

There are different kinds or groups of decisions, like business decisions, technical decisions, strategic decisions, et cetera (Chin, 2004). The decision-making structure is also dependent on the process model. We expect that the formal partly-parallel process will be best suited for complex NPD projects with strict time constraints. This process asks for a cross-functional team. So the development of new products needs different specialists from different functional areas. However, the decision-making process could be structured in numerous ways.

We will investigate the two extreme types of decision-making, 1) centralization and 2) decentralization. In centralized decision-making, the project managers, who are at the top of the control structure, make the decisions (Jin & Levitt, 1996), while in decentralized decision-making, ‘the decision-making authority is delegated to the worker with no interference from the manager’ (Zábojník, 2002, p. 5). There are three elements in the making process which could detect if the decision-making process is more centralized or more decentralized. First of all it is important where a decision is taken; this is referred to as the locus of decision-making. Second, the focus is on the degree of autonomy of decision-making at a specific level. Lastly, it is important to know if the one making the decision is also concerned with that decision (INES OECD, 2000).

In section 3.2.1, we consult the literature to investigate which type of decision-making is best for complex NPD projects. In section 3.2.2, the best type of decision-making for NPD projects with strict time constraints is described. Next, we formulate a proposition, which will be tested in a case study.

3.2.1 Decision-making in complex NPD projects

For complex NPD projects, literature agrees on the optimal decision-making process. According to Huber (1984), the greater the complexity of a project, the more complex the decision-making due to the consideration of more variables and more complex relationships among these variables. Moreover, more complex and innovative projects need more types of expertise and consideration of more criteria (Leenders, van Engelen & Kratzer, 2007). This leads to a need for more structural differentiation. Knowledge is important, not the formal authority (Bunderson, 2003). That is why decentralized decision-making will be more suitable. Moreover, as Jaworski and Kohli (1990, 1993) state, the less central the locus of authority is and the less formal rules, codes and instructions are, the greater the extent of information generation and dissemination. Referring back to figure 2, this information processing is a main characteristic of NPD. Because a complex project consists of many components and has to deal with a high level of uncertainty, information is even more important. However, through this participative decision-making approach, relations will be more complicated. Nevertheless, literature is unanimous in the conclusion that decentralized decision-making will be required for complex NPD projects.

3.2.2 Decision-making in NPD projects with strict time constraints

Literature is not unanimous about which type of decision-making is best to save time in projects. Some authors claim that decentralization is the best way to save time. Decentralized decision-making, where sub team leaders or even engineers themselves make the decisions, has less communications sent to and processed by higher managers (Jin & Levitt, 1996). The need for communication and information processing is in this way reduced, which saves time and also costs (Jin & Levitt, 1996). So the horizontal communication linkages are improved and will increase efficiency (Song, Thieme & Xie, 1998). In decentralized teams, decision involvement tends to be derived from expertise, and is specific to the task at hand (Ibarra, 1993). ‘A decentralized approach acknowledges that the most relevant knowledge, information, and competence may be only loosely coupled with formal position’ (Bunderson, 2003, p. 462). As a result, those involved in decision-making are the members who are most qualified to contribute to team decisions (Bunderson, 2003). This will reduce time, because the necessary information is already in place. Moreover, a decentralized decision-making structure improves the involvement of employees, which will increase the commitment to a NPD project and in turn results in shorter product completion times (Lukas, Menon & Bell, 2002). Kim and Burton (2002)

also agree that a decentralized structure is better for the performance of a NPD project in terms of both time and costs.

However, decentralized decision-making also has its drawbacks. Decentralization may decrease the quality of the process, because lower level workers make less conservative decisions, due to their limited perspective on the overall project (Jin & Levitt, 1996). Besides, by decentralization, relationships are getting more complex and it will be harder to control them. This could lead to delays (Ainamo, 2007). Olson, Walker and Ruekert (1995) agree with this. They contend that in a decentralized structure, the complexity of informal communication patterns increases, which could be more time consuming and less efficient than more centralized processes. Moreover, in a project with a decentralized decision-making structure, there is less control over time. A centralized decision-making structure may solve this. Because the project leader has a good overall picture of the activities and has to make the decisions, he/she is more in control. As Keegan and Turner (2002) point out, centralization will facilitate closer control over time in innovation projects. Additionally, project leaders do have more experience with decision-making, and also have the power to make decisions. Because of the competences and power of a project leader, he/she could make decisions more quickly. They have little need to consult and build consensus (Wally & Baum, 1994).

3.2.3 Decision-making in complex NPD projects with strict time constraints

It is now well-known that in complex projects a decentralized decision-making structure is optimal. Nevertheless, there is no consensus about which decision-making structure is best for ‘normal’ projects with strict time constraints. So, what kind of decision-making structure will be best for complex projects that are also restricted in time? Kim and Burton (2002) have investigated relationships among task uncertainty, level of centralization, and project team performance. Project team performance is measured in quality, duration and cost. A summary of the results of the study of Kim and Burton (2002) is pictured in table 1. Kim and Burton (2002) conclude in their research that a decentralized structure is best in a project with medium or high task uncertainty, when there are strict time restrictions. In this research the focus is on high task uncertainty and not on extreme task uncertainty. Extreme task uncertainty is in fact a situation which is rare and related to systems that are on the edge of chaos, like the army in combat situations. However, these results are based upon situations with a project team within one organization. Nevertheless, more and more new product development projects are executed within an

Centralized Structure Decentralized Structure

Low Task Uncertainty Better quality

performance

Duration and cost the same as in a

centralized structure

Medium Task Uncertainty Better quality

performance

Better duration performance

High Task Uncertainty Better quality

performance

Better duration performance

Extreme Task Uncertainty Better duration, cost

and quality performance

No option

Table 1: Task uncertainty related to (de)centralization and project performance (Kim & Burton, 2002).

alliance of more than one company. Yet, a decentralized decision making structure is indeed proposed to be more suitable for complex NPD projects with strict time constraints. This is due to the need of knowledge and creativity. Complexity refers to the many components and the relationships between these components in a project. The lack of knowledge about these components and their relationships, which can be elucidated by carrying out some research, leads to uncertainty. This uncertainty is thus caused by a lack of information. To get this information, expertise is very important, especially in unique NPD projects. So if decisions have to be made about specialized matters, expertise power should be more important than formal authority. In a decentralized structure, ‘what you know’ is more important than ‘who you know’ (Bunderson, 2003). Moreover, a decentralized decision-making structure will encourage creative thinking among team members, which is very important in NPD projects aimed at developing a radical innovation. ‘By facilitating the open exchange of creative ideas and analytical perspectives across multiple functions, the odds of producing innovative products that successfully address market desires as well as technical and operational requirements are increased’ (Ainamo, 2007, p. 846; Sutton & Hargadon, 1996). This results in our second proposition:

Proposition 2: Complex NPD projects with strict time constraints require

decentralized decision-making.

4. Case Study

To test proposition 1 and 2 we carried out a case study. In the case study a project has been observed named the passive radar project. A logbook of this project is shown in appendix 1. We used two research methods to test the propositions; 1) observations, and 2) interviews.

We made the observations throughout the first 6 months of the project. We paid attention to the decisions that were made during the project, observed

how these decisions were made, and how the project team structured the activities during the project. Appendix 2 shows the results of the observations.

The interviews are conducted with the project managers of the responsible company. Jan Reitsma is one of the project managers. He is the owner of KXA Project Professionals since 2009, which is a new company dealing with the management of (technical) complex projects. Jan Reitsma has been involved with project management in the information technology and service industry since the beginning of his career in 1987. Wilma Mulder, the other owner of KXA Project Professionals, has also much experience in managing projects. During her career the projects she managed became more and more complex. Moreover, Wilma Mulder not only managed technical complex projects but also projects with a more human face, like reorganisations. Other team members were not interviewed in a formal setting. However, during the observations a lot of information is gathered from the other team members in informal conversations. In the interview the focus is on decisions made about the process model used and about the decision-making structure within the project. The project managers will use their experience and the circumstances of this particular case to underpin their decisions for the kind of process model and decision-making structure used. The results of the interviews are shown in appendix 3. Appendix 3 is written in Dutch as the project managers explicitly wanted to be interviewed in Dutch and be able to check the results in their own language.

We are going to explain the case further in section 4.1 and will present the results in section 4.2.

4.1 Project Background

As mentioned above, KXA Project Professionals is a new company which deals with (technical) complex projects. Currently they execute a project named ‘passive radar’. The passive radar is a product with a revolutionary new technique that is able to enlarge safety and control of air traffic in an environmentally friendly way. The main reason for the development of the passive radar is to take advantage of a technical development. To do so, a network was established because KXA Project Professionals does not have all the expertise needed in house. The network consists of, evidently, KXA Project Professionals who will manage and structure the project, Astron and the LOFAR foundation who will provide the know-how of the passive radar, Dysi who is responsible for the modelling of the software, and Groningen Airport Eelde that is the knowledge provider specific for the aeronautical knowledge. These firms are all SMEs. It is important that the project will be carried out as fast as possible because of the intense competition in

this line of business. Different companies are trying to get access to this new technology, so competition is upon the catch. The project is complex because of 1) needed cooperation between different organizations, 2) the need for different specialties to work together, 3) the product consists of many different parts, and 4) there are many different stakeholders in many different locations. In addition, the cooperating network is new to all companies, and the market of air traffic control, airports and government is quite a new adventure. This leads, together with the newness of the technology, to a highly uncertain project.

Many gaps in the current knowledge have to be filled while completing the project. Besides time, also quality is very important in this industry. Lives are at risk if the product does not function properly. That is why the product has to be tested for a fixed time period. On this aspect, time cannot be reduced. This however, also applies to competitors, so no competitive advantage can be taken from the test period.

4.2 Results

Observations were carried out right from the start of the project. The first six months of the project were observed completely (appendix 2). In this period decisions were made about the structure of the decision-making process and the project process. Immediately after this period, we interviewed the project managers (appendix 3). Looking at the type of making structure, we focussed on three elements of the decision-making process. First of all it is important where a decision is taken; this is referred to as the locus of decision-making. Second, the focus is on the degree of autonomy of decision-making at a specific level. Lastly, it is important to know if the one making the decision is also concerned with that decision (INES OECD, 2000). To determine which process model is used during the project, we observed the process in which activities are executed. After the observations, we used the interviews to evaluate the choices that were made. The results in section 4.2.1 and 4.2.2 illustrate the way in which the passive radar project is executed. This, of course, does not mean that it is the best way. We evaluate this in section 5 in which we will draw conclusions with respect to our propositions 1 and 2.

4.2.1 Process model

The passive radar project is rather hectic because of the high uncertainty at the start of the project. The technology, which makes the passive radar unique, was discovered by accident during the development of a new type of radio telescope. It was not known what the technology could bring about and if it could be made suitable and marketable in another industry. One thing

is for sure; with a passive radar you can not only locate objects in outer space but also in the near airspace. Consequently it is to be expected that the passive radar can be very suitable for air traffic control purposes. However, there is no demand from the market which makes the project technology-push. A network was established to look for opportunities for this new technique in air traffic control. Because of the newly created network of organizations there was no formal process yet. Moreover, the network did not have enough financial resources to execute the project. Nevertheless, without having a real formal process a preliminary investigation was set up to inform the government about the possibilities of this innovative project to contribute to a sustainable environment. This in order to get subsidised. During this application for funding, the project team set up a process for the development of the passive radar and started their formal process. This process is based on work package management (see table 2). Work package management is a method in which the project is divided into subprojects or work packages. For the passive radar project it was important that these work packages were not too strictly defined, in order to be able to deal with changes.So the work packages were made flexible. As a lot of aspects were still unknown, it was also impossible to get clearly defined work packages. However, future activities were set. Unfortunately the government rejected the application for funding. Consequently, one partner left the project because he did not want to take any risk. Besides, the project team now had to find other ways to get the necessary financial resources. Moreover, it changed the sequence of activities. If the project had received funding, the project partners would have continued to develop the product without first considering the market extensively. This because it is very difficult for customers to imagine this new technology as it is a radical innovation. Besides, this also costs a lot of money. So, the next step was a more detailed investigation in which a market research was conducted. This market research is a cheap method to get a better idea of what requirements the passive radar has to fulfil, which also leads to a decrease of technical uncertainty. Moreover, in this way a prediction could be made about the size of the market and thus about the potential success. The project team hopes to attract investors with the promising market information. During the process of information gathering of the potential market, some elements were already brought to the development stage. Thanks to the market research, more information was collected about the potential market. It seemed that the passive radar could be marketed into the civil aviation sector and the military aviation sector. This also had consequences for the activities in the future. If the passive radar will be launched onto the military

Working packages Mission partners

1. System definition A. Definition placing a mini-station at Groningen Airport Eelde: Astron & Dysi

B. Definition product with real time signal processing: Astron

C. Feasibility study for a station at Groningen Airport Eelde: KXA, Groningen Airport Eelde & Astron D. Definition demonstrator: Dysi 2. Modelling A. Algorithms: Astron

B. Dimensioning/modelling: Astron & Dysi

3. Proof of Concept A. Rollout station at Groningen Airport Eelde: KXA, LOFAR Foundation & Groningen Airport Eelde

B. Experimenting: Astron & Dysi 4. Business Development A. KXA, Astron & Groningen Airport

Eelde (regulation) 5. Project Management A. KXA

The underlined partners are responsible for the relevant project component

Table 2: Work Package Management of the passive radar project

aviation market, the activities will be more time consuming because the Ministry of Defence depends on many other parties and uses more strict procedures and legislation. The project team decided to market the passive radar only into the civil aviation sector. So the process proved to be dependent upon choices made and information gathered.

In conclusion, there was not yet a formal process in the beginning. This meant that there was even more uncertainty in the beginning of the project, because nobody knew what to do. Unconsciously some steps were taken towards a more formal process. In the interview both project managers indicated that a formal process is really important to get a clearly defined project. This was just not possible in the beginning of the project of the passive radar because of the high level of uncertainty. At the moment a formal process is being developed for the project. This process has to be flexible. In the interview the project managers claimed that the partly-parallel process would be best suited for the project. This process was chosen because of the control you keep of the project, which is necessary to deal with all activities and the relations. On the other hand, this process will not negatively influence the creativity which is also necessary for the development of this radical innovation. During the observation we studied the process of the project. In figure 7 we present a schematic representation of the process that already has finished and the activities that still need to be executed in the future. The process model resembles the third generation model of Cooper (1996, p. 479) but does need some adjustments for this specific project. The first stage is disconnected

Figure 7: The process model of the passive radar project based on Coopers’ third generation stage-gate-model (Cooper, 1996, p. 479).

from the other overlapping stages. This is due to the complexity of the project and the high level of uncertainty. It is important to first gather important information to fill knowledge gaps, before further executing the project. In this way you can deal with the uncertainty derived from the knowledge gaps which are greatest in the beginning of a project.

4.2.2 Decision-making

As indicated in the interview with Wilma Mulder (appendix 3), you can see a project as the construction of a house. You first have to build the foundation. It seems like nothing happens because you cannot see anything yet. However, the foundation is the most important part of the house. A construction failure in the foundation is difficult to undo once the house is finished. This resembles the situation of a project. One of the things that has to be taken care of in the beginning of the project is the kind of decision-making structure. In the project of the passive radar this decision-making structure has been agreed upon right from the start. In such a project lots of decisions have to be made. Like indicated in section 4.2.1 the project managers make use of work package management. So the project is divided into subprojects. Every subproject gets a subproject leader who is responsible for the task at hand (table 2, the underlined partners). However, if a decision has to be made outside the scope of the sub leader’s subproject, the decision will move to a higher hierarchical

level. This also applies to subprojects. In subprojects there also is a breakdown into subtasks. The sub leader shifts the responsibility of a subtask to one of its employees. The employee is responsible for his/her subtask. However, if a decision has to be made outside the scope, the decision will scale up to the sub project leader. Nevertheless, KXA Project Professionals has the ultimate responsibility. Below an example is given of an observed decision-making process in the passive radar project:

In a complex project, the partners will not accept that decisions are only made by one player. Furthermore, nobody wants to make decisions on their own because of the great uncertainty that is involved. Decision makers look for co-perpetrators. This also applies to the project of the passive radar. The participation with the partners and stakeholders is very important in this project, because of the high level of uncertainty. To deal with that, knowledge and expertise are the most important competences. The formal position is less important.

The structure used in the project of the passive radar is not that quick and straightforward. Real time information is analysed and different alternatives are created. Information is important to make the right decisions. Nevertheless it is also important not to exaggerate. It is a

In the beginning of the project the government decided not to fund the project of the passive radar. The project team had to decide how to proceed. This decision is a strategic decision and was taken at the highest hierarchical level. KXA Project Professionals has the most expertise and experience about managing projects within a budget and in time, so they have the most information to make this decision. They decided, in consultation with the other partners, that a market research was the first priority. This consultation took place in one-on-one meetings. The market research is a subproject in which an employee of KXA Project Professionals got full responsibility. The employee decided which tools to use and how to get information. She presented the results to the higher level project team in a formal meeting. After the market research a strategic decision had to be made about which market is best to serve. The employee gave advice but the ultimate decision was made on the highest hierarchical level, because this decision was out of the scope of the employee that had responsibility about the market research.

Astron and the LOFAR Foundation have all knowledge about the technique. Without them, there is no project. KXA Project Professionals knows how to manage a technical complex project within a budget and in time. If decisions have to be made about the project organization, KXA Project Professionals is the expert. Like the decision made about how to proceed after the rejection of the funding. If decisions have to be made about the product features and thus also which market to serve, Astron and the LOFAR foundation have more power because of their expertise. They are dependent upon each other, only KXA Project Professionals is replaceable and ASTRON and the LOFAR foundation are not.

matter of the right balance. Making bad decisions will cost more time than taking more time to make the right decision.

In the project of the passive radar both centralization and decentralization are used in decision-making. A centralized decision-making structure is used for decisions made outside the essential preconditions, like the one made after the market research, about how to proceed and which market to serve. Decentralized decision-making is used within the essential preconditions. Here the employee gets the full responsibility about the execution of the market research. In this way a good overview of the entire project can be kept, which is necessary to deal with the complexity. In addition, the decisions are made by the employees with the most knowledge, which will lead to good decisions. The more strategic the decision, the higher the decision is put on the hierarchy ladder. So they decentralize if it is possible, and centralize if it is necessary. As a result both complexity and time are managed simultaneously.

5. Conclusion

The aim of our research was to explore how to manage complex NPD projects that also have strict time constraints. Our focus led to some contradictions found in the current literature about managing complex projects on the one hand, and managing projects with strict time constraints on the other hand. In this, we further investigated two subjects, 1) the decision-making structure and 2) the sequence of activities in a formal process model. We derived two propositions and found support for the first one. This means that in the project of the passive radar, which is a complex NPD project with strict time constraints, a formal partly-parallel process is most suited. However, it seems that in the beginning of such a complex project it is very difficult or even unthinkable to start parallel processes. This follows from the great uncertainty involved. The project managers first need some critical information before further executing the project, and thus entering a new stage. This process is still in line with the third generation model of Cooper (1996), but needs a small adjustment for complex NPD projects with strict time constraints (see figure 8).

Our second proposition is only partly supported by our findings in the passive radar project. A decentralized decision-making structure is definitely required in managing a complex project. This results from the need for creativity and the need for knowledge and expertise. However, it is also very important to keep sufficient control over the project to manage the complexity in a certain time frame, which asks for a more

Figure 8: Process model for complex NPD projects with strict time constraints, based on the model of Cooper (1996, p. 479).

centralized approach. Moreover, good decisions are more important than quick decisions induced by time constraints, especially in the beginning of the project. In the project of the passive radar a mixture of decentralized and centralized decision-making is seen as best suited. So decentralize where it is possible and centralize if it is necessary. In this way, decisions are made by the members with the most expertise about that specific subject. However, strategic decisions and go and kill decisions are made on the highest hierarchical level. Consequently, the project managers keep control to manage the complexity, but also keep the flexibility and give employees responsibility to get a creative and motivate project team which leads to an accelerated product development process.

6. Discussion

The results of our research indicate that complex NPD projects with strict time constraints have to be managed otherwise than complex NPD projects without strict time constraints or ‘normal’ NPD projects with strict time constraints. Current literature does not cover this yet, so a new research field shows up. It is important for project managers to see the difference between complex NPD projects with strict time constraints and other types of NPD projects to increase their chances to successfully launch an innovative product. This result is only supported by one case, so it is important to further investigate this outcome. The definition of complexity is also rather important. In future research the same definition of

complexity has to be used, to make the outcomes valid. However, the first step is made now in the research area of complex projects with strict time constraints.

Another topical subject that could be researched on the basis of this article is the interaction in a network of firms. Who has the power in complex NPD projects and how does this change when there are strict time constraints? Moreover, what about the involvement of users? Here there is also a contradiction in literature about managing complex NPD projects with strict time constraints. Some authors claim that the involvement of users in a development process of a radical innovation is impossible, because customers only think in terms of ideas that are possible today. However, according to Fang (2008) for example, the involvement of customers could also lead to more innovative ideas. But what does the user involvement do with respect to the element of time? Handfield and Bechtel (2002) think an early user or supplier involvement leads to reduced costs and lead times. However, you also have to put in time to give explanation and information. In fact, you are dependent on them. Fang (2008) argues that if process interdependencies are high, customer participation has a negative effect on speed-to-market. Future research could search for more contradictions in management implications among on the one hand complex NPD projects and NPD projects with strict time constraints to get a more comprehensive picture of how to manage complex NPD projects with strict time constraints.

Acknowledgements

The author thanks Wilma Mulder and Jan Reitsma for their cooperation and Cees Reezigt for his careful, constructive and stimulating comments.

References

Ainamo, A. (2007). Coordination mechanism in cross-functional teams: A product design perspective. Journal of Marketing Management, 23(9-10), 841-860.

Alhadeff-Jones, M. (2008). Three generations of complexity theories: Nuances and ambiguities. Educational Philosophy and Theory, 40(1), 66-82.

Bachelard, G. (1934/2003). Le nouvel esprit scientifique. Paris: Presses Universitaires de France.

Beer, S. (1959). What has cybernetics to do with operational research? Operational Research Quarterly, 10, 1-21.

Bozarth, C.C., Warsing, D.P., Flynn, B.B., & Flynn, E.J. (2009). The impact of supply chain complexity on manufacturing plant performance. Journal of Operations Management, 27(1), 78-93.