How do literature and start-up companies think about the decision making process regarding product architecture and product variety in
early design phase? - A case study
Master thesis in Innovation and Industrial Management Alexandra Zou
Supervisor Johan Brink 2018-03-18
Acknowledgement to
Professor Johan Brink
Thank you for all your support and constructive feedback. You kept me going when it was really hard. This was a lot of fun!
Dr Qiuhong Hu
Thanks you for all the great wisdom and support
Mr Gregory Carson
Thank you for this internship opportunity at Luxbright
Luxbright team
Thanks for all your support
Abstract
The thesis looks at applicability of decision making models introduced by literature in the area about product architecture and product variety, on start-up companies that are high-tech and production oriented. Decision making in the product design phase will impact on product architecture and product variety. These two areas are where the combined information and technology that comes externally from customers and internally from research are merged to create a competitive product. Both in terms of customer satisfaction and cost optimizations.
Existing literature have focused on larger companies in various industries that apply algorithms and decision making frameworks to see whether the result is significantly better. It would therefore be interesting in seeing what start-ups think about these methods, whether they are interesting and applicable in the early stages of product development. The result is very interesting, because the start-up companies do care about these areas and does use a lot of decision making methods. However sometimes literature methods are not fully applicable to them because of lack of quantitative data or simply not relevant. That said, methods that offer cost control and innovative information leveraging are interesting to start-ups and they do use all the resources available to them in order to realize their vision. For Luxbright that is to make a safer world.
Keywords: product architecture, product variety, decision, start-up
Table of content
1. Introduction ... 1
2. Background and Fundamental Concepts ... 3
2.1 Strategic decision and resource based view ... 5
2.2 New product development: ... 7
2.3 Product design ... 9
2.4 Product family/platform ... 11
2.5 Product architecture ... 14
2.6 Product variety ... 16
2.7 Decision making literature ... 19
2.7.1 Decision challengers in start-up companies ... 20
2.8 Articles of key interest ... 20
3. X-ray tube ... 24
3.1 Parts and function ... 24
3.2 X-ray Industry Facts ... 26
3.2.1 User industry ... 26
3.2.2 Producer industry ... 27
3.3 Luxbright- Creating X-ray Brilliance ... 28
4. Research Method ... 29
4.1 Research paradigm ... 29
4.2 Case study ... 31
4.3 Semi structure interviews ... 33
4.4 Qualitative reasoning ... 34
4.5 Part II (Practical steps) ... 36
5. Result ... 37
5.1 Matrix over components ... 37
5.2 Research and development interview 1 ... 39
5.3 Research and development interview 2 ... 40
5.4 Marketing interview ... 41
5.5 Production interview ... 41
5.6 Decision interview 1 ... 42
5.7 Decision interview 2 ... 43
5.8 My Observations ... 45
6. Analysis ... 45
6.1 Comparison of Wan Dresden article with interviews ... 45
6.2 Comparison of Gao et al article with interviews ... 47
6.3 Comparison of Tai with interviews ... 49
6.4 Overall analysis ... 51
7. Conclusion ... 52
8. References ... 53
9. Appendix A ... 58
Figure table
Figure 1 6
Figure 2 8
Figure 3 21
Figure 4 24
Figure 5 26
Figure 6 50
1 1. Introduction
The manufacturing of products is at the heart of any producing company. This is especially true for high-tech companies whose competitive advantage is embedded in the production process. Therefore, the production process will affect and be effected by all aspects of the company from top managers’ decisions to everyday tasks performed in a factory. The relations between production process, design, variety, decision making, strategy and other aspects of the company are complicated but l shall still try to briefly lay them out here. Production process is at its heart, the making of a product that will generate value to the company and customers. Production process is determined by what a company makes and how top managers have decided to make it. There are many management theories in connections to production process, such as product development decisions, product architecture, product platform design, product family, strategic management, total quality management, constraint management, and organizational theories to mention some. Production process is a widely infiltrating aspect of any company that will help shape the organization of a company. The factors that affect production process are decision on product design, decision of competence within a company, collaborations with suppliers or other partners, customer demand, material innovation and corporate strategies. The tasks that are effected by production process are location of facilities, key customers, skills of employees, marketing, competitiveness of the company, profit and value capturing to mention the broader aspects (Jacobs, Chase, 2014).
This thesis will look at decision impact on areas of production process with focus on start-up companies. And compare the decision process there to literature’s findings and reviews, in order to discuss similarities and differences. In particular, it would be interesting to look at motivation behind them. Literature have paired up decision approaches to conditions and set a guide for which is most appropriate production method to use during particular circumstances, as the reader will see in Chapter 2. Decision making is what ultimately shapes the company. Every choice requires a decision, whether it is voluntary or not. Production design decisions will impact the type of components that exist inside the product. The design will determine a components interface, material, strength and weaknesses among other characteristics. It is important to have a decision of how the design will look like when bringing in collaborators and partners to help in production development. Outside influence from customers can push a company to take certain decisions between architecture and function.
That is what I am interested in.
Krishnan and Ulrich (2001) look inside what they call “the black box of product development”
and the decisions firms make in this process. Krishnan and Ulrich (2001) call this the decision perspective of product development and it has the advantage that how firm’s make decisions
2
remain fairly consistent. Because firms operate in different industries, make different products and have different characteristics, it is not possible to find something mutual in this aspect.
The decision perspective cuts across all idiosyncratic features and bypasses all the details in each company. The focus is more on the structure itself and framework of decision making.
Decision making is further divided into four consecutive steps by them, 1) concept development, 2) supply-chain design, 3) product design and 4) production ramp-up and launch.
For my thesis the focus will be on decision making in product design. Some main sub questions that are answered here are: What are the key design parameters? How do you identify these parameters? In the production design phase researchers at the firms are concerned about design parameters’ relations between functionality and performance and to clarify the link to profit (Krishnan Ulrich, 2001). The goal is to be able to measure production process efficiency, to ensure optimal production and for that firm’s need geometrical models and detailed design understanding. For example, Papalambros and Wilde (1988) describe a monotonicity analysis that uses qualitative reasoning to get insights about parametric design.
This models can also help companies make decisions regarding the life-cycle of a product.
To narrow down the research area within production process, I am going to focus on product architecture and product variety because there are a lot of research in this area. Furthermore I am interested in getting a start-up company’s perspective on this. I will look at decision impact on product architecture and product varieties. In order to make it easier to explain l will use the example of an X-ray tube as a product for explaining product architecture and product variety. I am very fortunate to be an intern at Luxbright, which I think will be a great opportunity for me to learn about X-ray tube and understand this decision area better. I hope this research will be interesting to them as well. So my research question is:
How do literature and start-up companies think about the decision making process regarding product architecture and product variety in early design phase? - A case study
3 2. Background and Fundamental Concepts
In this chapter, I will first introduce strategic decision making and resource based view theory.
Resource based view is a strategic theory that connect resources in a company to unique capabilities in a firm. This theory is relevant because all companies have limited resources, in particular start-ups. I will introduce new product development under the resource based view using its assumptions. Key areas of interest in new product development are product design, family, architecture and variety. In this way I will narrow down my research width. Before I explain decision making and subcategories of production, I will explain briefly the different production methods that exist. The selection depends on the needs of the production and scale of production. After that l will discuss strategic decision making and in the end discuss characteristics of a start-up company in general.
Manufacturing of products are related to quality of final product, cost effectiveness and economy of scale. Methods used during research and development needs to be redesigned to fit larger scale production. Production process is the result of the trial and error that characterise the transition from research and development methods to large scale production methods. The manufacturing process can be studied from many different aspects and thus focus on different challengers that arise in the process. There are 5 general categories of manufacturing methods, repetitive, discrete, job shop, process (batch) and process (continuous) (Goldensen, 2015). A company can use more than one method during the production process.
Repetitive manufacturing refers to a production that makes the same item all day long with little setup or change over activity. Discrete manufacturing describes a production with many setups, changeover activities and that result in products that are different from each other. Job shop manufacturing means that a factory is divided into production areas. The manufacturing is both continuous and intermittent, which means designers working in this kind of manufacturing environment would also need to design equipment occasionally. If the production process is repetitive, then it is very likely to be automated. Process (batch) works in a similar fashion to discrete or job shop production. Batch process can be continuous as in one batch after another. Process (continuous) is analogous to repetitive process. The primary difference is the use of material such as gas, liquids, powder in the process etc. (Goldensen, 2015).
After having introduced types of manufacturing process, let’s look at the preparations leading up to a clear manufacturing process. Simply explained, when research and development department have completed the search for an acceptable product/prototype that can bring value to customers, it is time to start producing it. First comes the production design phase during which the prototype or the effective/active part of the product will be given an appealing shape. The technology is capsulated in a shell or other packaging in order to be user-
4
friendly and to protect the technology. The design phase is very creative and there will be many solutions to the “packaging problem” or the design of the final product. During the design phase the company could also collaborate with potential customers in order to fully identify the value achieving aspect of the product and compare the prototype to existing products on the market. Sometimes the prototype is considered ready to be sold fairly quickly, other times the product design phase takes a long time (Ulrich, Pearson, 1998). This is especially true if the prototype has the potential to be used in many different markets with similar but still distinctive needs. Other times the product can later be updated, as in the case with Apple’s iOS products.
The duration of the production design depends overall on the needs of the customer and how long it takes to shape a satisfying final product. The product architecture is what decides the number of components in a product and how flexible they are. By flexible, it means here, how easily they can be switched to other similar components or if the same component can be used in many products. Here the company will try to identify constraints in their product, such as limited amount of space, strict regulation about material used given by the government, etc.
In the case that the product can be differentiated in various markets or updated in later generations, the company will also look at how components in the product can vary in order to achieve results closest to what the customer want and also future competitive advantages that can be found in the components. From a strategic point of view product architecture will enable product variety and product variety is beneficial to the company in the following ways.
It can increase sales, help companies retain market share and help companies gain competitive advantage in technical skills. But companies should not have too much variety because it will make the supply chain unnecessarily complicated. It can be both a good strategy but also a treacherous one.
Increased variety can bring up sales by making the product more fitting to customer needs, but too much variety can instead lead to customers not buying anything at all, a phenomenon called ‘overchoice’. Increased variety will also diminish economies of scale, since more differentiation means more components that are different and this means smaller production batches (Jiao et al, 2007). The production variety decision can also affect a company’s strategic positioning on the market. In an engineering perspective the production variety, earlier known as product portfolio management, can be used as a market segmentation grid (Tarasewich Nair, 2001).
Another way of segmenting the product is using platform-based product family design. Here the product approach can be divided into scalable and configurational product family design.
The scalable approach first coined by Simpson et al (2001) was focused on minimizing variation in scaling factors and increasing commonality in products. This approach identifies
5
design parameters that share common values and design parameters that have distinctive values that satisfies performance and economic requirements. And the value of the product is in the physical components themselves.
Thus the design phase is linked with level of product variety through product architecture.
Product architecture is about the study of design dependencies in components. Architecture embodiment focus on constraints of a product from two perspectives, individual product variants and product family constraints. From a design perspective, product architecture is about understanding product flexibility, which is part interfaces, types of interfaces, functions and modules in product. (Asikoglu 2012). An example of research focus in this area is the balance of loss of modularity against cost of reconfiguration of parts. In summary, product architecture design focus on mapping from function to component and focus on trade-off between distinctive ability of a product and commonality in a product family.
2.1 Strategic decision and resource based view
A firm’s strategy is concerned with how the resources and capabilities in a firm can be used to capture opportunities in the external environment, the market. Capturing opportunities will allow the firm to be profitable. Strategy interacts with the external market and internal organization of a firm. In this thesis the focus will be on the internal aspects of a firm such as resources, capabilities and competitive advantages. The internal resources in a firm can help ensure stability for the long term performance of a company because the changes that happens there can be controlled. A theory about how to find those competitive advantages is the resource based view (RBV). The main argument in the resource based view is that companies become profitable by being unique and exploiting differences in capabilities and resources (Grant, 2010). Exploiting in this sense, really means coordinating labour in a company to make a product that the market wants. Efficiency in production emerge when each labour division becomes specialized in its own task and the coordination between different divisions happens smoothly. However, the more a production process is divided, the more complex it is to coordinate. Moreover, the more volatile the external market is, the greater the cost of coordination. Therefore, the success of a firm depends on balancing specialization cost with coordination cost. A decision that is complex to make (Grant, 2010).
Strategic decisions are made at the highest level in a company and often with great uncertainty and little structure. These decisions will affect the company in the long run and directly affects the effective use of company resources such as capital resources, technology, plant and equipment etc (Nemati et al, 2010). Much in the way research in strategic decision is made, it is influenced by behavioural decision theory and transaction cost economics. Another big area
6
is decision cognition research. Nemati et al (2010) suggest there might be important connections between managers own style and the type of organization they are in, to the kind of decisions that are made and how they are made. The labelling of information, as a crisis or opportunity will impact on managers’ attitude and influence the company’s response. The level of rationality in the decision process, will also influence how much information is processed before decision has
been made (Nemati et al, 2010).
Taking decision making one step further, research has looked at the cognitive process behind it.
Behavioural decision theory (BDT) is about the underlying cognitive process behind a person’s choice. The behaviour according to some researchers
can be classified as rational or irrational and whether or not the decision is connected to a specific goal. The rational behaviour would be the best actions to take in order to reach the goal, (Einhorn and Hogarth, 1981). However, intuition also plays a part in decision making, which underscore the validity of this model. Instead a better way of explaining this is that, decision is made rationally and in context to some existing conditions. Thus decision making depends on task structure, information processing abilities and representation of task (Einhorn and Hogarth, 1981). Many researchers can’t deal with the opposing views between rational and intuitive processing, but Calabretta et al (2016) try to lessen the tension between these two concepts by using paradoxical thinking method. The paradoxical thinking method considers rationality and intuition to be complementary perspectives. Calabretta et al (2016) argue that by adopting this mentality, it will further the company’s strategic decision making.
A kind of trade-off between intuition and rational cognitive process is similar to the kind of trade off often seen in transaction cost economics. Transaction cost economics is about carrying out transactions while minimizing the cost for it. Transaction cost is defined as an alternative way to use a certain asset without lessening its value. The 3 ways to control transaction costs are through contract law, bilateral dependencies between parties and internal dispute solving skills. Just like BDT deals with irrationality and intuition, transaction cost deals with uncertainty and how to lessen this by using asset specificity and checking transaction frequency (David and Han 2004). Both theories contribute to strategic decision making, in that they study the cognitive aspects of it and rationalizes it by using objective measures. Strategic decision making is at its core about intuition, rationality and political
Figure 1 relation between resource based view and decision making
7
behaviour. It is also affected by external environment and internal corporate environment.
Strategic decisions can also be ranked depending on importance, uncertainty and motives (Elbanna and Child, 2007). This ranking is closely connected to how resource based view approach the classification of corporate assets. The resource based view identifies the company’s relative strengths and weaknesses and use that knowledge to guide the development of competitive capabilities for profit making (Grant 2010).
Often strategic decision making begins by simplifying a complex problem and categorizing information into decision areas. Those are defined by RBV as valuable, rare, inimitable and non-substitutable. Hence RBV and strategic decision making complement each other. A company should thus make decisions that leverage architecture and variety that allows for use of competitive skills. The two are closely linked since strategic decision making will affect long term performance of a firm and thus its capabilities development. The long term planning of resource use will also affect a company’s ability to bring new products to the market.
2.2 New product development:
New product development can be considered a long term strategy for survival of a company, by creating new customers, market share and keeping existing customers. Developing a new product takes a lot of investment and it involves a lot of risks. It is therefore important for the company to manage new product development in a way that lessens the risk of failure. New product development theories gained importance during the 1990s when Japanese car producers started to outcompete US car producers. During these time researchers focused on developing formal procedures that could help companies discover initial product development issues. For this period, focus was on developing a product strategy and model that included all essential elements of new product development process (Shipley and Armacost, 1993). In particular, the model should combine managerial model, design technique and management strategy together. Shipley and Armacost (1993) call this the systematic approach to product development and it has two parallel running strategies, management strategy and new product strategy. The successful Japanese total quality management philosophy should be included as well as corporate goals and objectives. A prominent change to NPD process was the introduction of a stage process model. Stage 1 would be a general outline of what the product should achieve and blueprint for production and marketing. Stage 2 will detail all the processes that will bring the product to market and even post launch evaluation. The most prominent difference is the larger use of objective measurements, whilst before managers relied on experience (Shipley and Armacost, 1993).
8
New product development is about using capabilities gained whilst making the current products inside the company and then finding a similar product within a distinct submarket so that the company can benefit from product variety through increased sales and market share.
The resources at the company therefore need to be increased, as well as, be redistributed in order for the company to achieve this. Thus companies should take advantage of their core capabilities during new product development. Core capabilities are distinctive resource that a company has that differentiate them from competitors (Barton, 1992). Core capabilities is a well discussed topic and has been given other names in other research studies, such as core competencies, company specific competencies, resource deployment, invisible assets. Hayes (1985) and Quinn (1980) argue that successful competition is more due to the correct utilization of capabilities rather than strategies and innovation (Barton, 1992). Barton talk about the 4 dimensions of core capabilities that are institutionalized within a company, linked to the values of company founder and early leaders, can be traced back to the first products and they are interrelated and interdependent to each other. Core competencies drive a company to collect certain resources and vice versa. When developing a new product, the capabilities connected to the company are applied to it. At the same time the dimensions of the core capabilities shift due to the addition of a new product. Therefore, how closely the technical, managerial needs of the new product fit to the core capabilities is determined in the process (Barton 1992).
Figure 2 parallel running strategies
9
Paradoxically core capabilities can become core rigidities due to less emphasis is placed on other skills that are considered outside of the core capabilities. People are more reluctant to apply themselves in areas that are undervalued in case it effects their personal abilities. Core rigidities hamper innovation because the values connected to these capabilities can also restrict its development (Barton, 1992). This is especially tough when a new product development process requires a skill that is considered non-core. It could limit cross functional integration and by extension critical learning process. Furthermore, it’s a negative impact of decision making because its irrational.
A more fully comprehensive new product development model was studied by Mullins and Sutherland in 1998. They two studied the more “regulated” NPD in relation to marketing with focus on how NPD practices can mitigate risks and manage uncertainty by leveraging marketing capabilities. The latest techniques within NPD utilizes quality function deployment, stage-gate, cross functional teams and production champions. These techniques should help companies bring the product to the market faster and increase the chance of success. The risk mitigation happens when the definition of uncertainty is made more concrete. In the Mullins and Sutherlands (1998) case, they defined uncertainty on 3 levels, customers are unable to make clear what they want, company cannot decide how much capital to invest into one venue and company is unable to decide which market is the best to enter into. When faced with an uncertain market, is it good to focus on bringing prototypes to the market early, gain customer feedback and focus on quality of the product. Therefore, the company wold gain knowledge along the process, long term investment commitments will be less risky and there is more certainty about the attractiveness of the product on the market. Uncertainty can be dealt with by prototyping, continually redefine the product and doing qualitative screening. During the prototyping process the design of the product will take shape as well. It will be interesting to see if these decision making methods about NPD are different for start-ups and established companies. Do bigger companies think they have a good knowledge about the market and are thus overconfident? Do start-up companies underestimate their knowledge about the market?
2.3 Product design
The demands and risks on the market will translate into the product design. Research and development will push product to do better the task that is in demand. Product design per definition given by Zeng and Gu (1999) is an evolving process that starts with design requirements and continuous on until the product have a design description. Research regarding product design touch upon what should be included in the product design phase and how this process is linked to and further influence later proceedings. Fundamental questions in this area are what is design? What are the nature of design? Thus in order to explain design it is important to understand the same law and speak the same language. For a while now
10
researchers have tried to establish a formal framework and give an objective representation of the design process that can be translated into mathematical equations. Zeng and Gu (1999) propose a 3 step problem solving model using a mapping process for design specifications.
Design can be a creative and unstructured process and it usually starts due to complaints on existing products or improvement on existing products. Zeng and Gu begins by describing the basic dimensions and constraints on the product as sets of product descriptions (S) and product performances (P). X is denoted the union between S and P. Given additional constraints that are bunched into R, the equation looks as follows:
𝑂 = 𝜆(𝑋, 𝑅)
This is a predicate, which states that the outcome O will be based on X dimensions and R constraints simply put. Just putting it mathematically like this is not enough. Because every X and R, are unique and what really goes into the equation is an index or number which describe the dimension and constraint. So what does the union actually mean?
The two main phases in a design process are conceptual configuration and detailed design phase. The configuration process focus on concrete product architecture and components. Key design parameters are also determined at this stage. To view the design process as linear would be wrong because adjustments are made all the time and it is more common to jump between these two stages fluidly (Rupinder, 2016). However apart from how the design process itself looks like, there is also what the design process entails. Product design combines engineering skills with aesthetics since the product needs to be able to perform according to the customers wishes and look good. The three product level dimensions in product design are form, function and ergonomics. Form design strive to fulfil the promises that the marketing department have given. Here the company can obtain customer loyalty through “intrinsic experiential value”
(Rupinder, 2016). Functional design focus on performance, such as time to reach top speed, how finely grained the flour become etc. This is why the customer buys the machine.
Ergonomic design is about how comfortable the customer can be while using the product.
Here the objective is to maximise usability (Rupinder, 2016).
It is immediately apparent that there is an information gap between the model designed by Zeng and Gu and the product design phase described by Rupinder et al (2016). With the mathematical model the focus is very internal towards the company’s own opinion regarding what is product performance and how should it be measured, whilst in the product dimension model there isn’t any measuring at all and there is an interaction between the external and internal opinions of the product. But instead of rejecting one or the other, these models can be considered in parallel with each other. The mathematical model does focus on technical capabilities and feasibility of the product. The product dimension model would help the company incorporate customer opinions into their final product. But these small discrepancies
11
in language, law and definition can translate into very different results in the mathematics.
The mathematical way of describing a non-linear design process is to map the process on 3 levels. Already here is a discrepancy. Rupinder et al (2016) states 3 levels of product design, abstract knowledge, divergent process and property knowledge. These 3 focus on past experience and information, optimization of resource use and functionality and ergonomics of the product. Zeng and Gu (1999) sees design requirement and specifications feed into the design process and out comes product description. Specification are representations for basic components and connections in between them. Then an algorithm will solve primitive recursive functions and unification operations. (Zeng and Gu, 1999, Rupinder et al (2016)).
Dekker et al (2013) attempt to close this gap by looking at the interface between product design and engineering and manufacturing. There is a lack of research bridging this gap. Studies focusing on multiple case studies usually do not look at causality and lack accuracy, which makes the validity of the study limited. When the research only look at a single case the conclusion aren’t enough to draw generalisation and abstraction for grounded theory. In this research Dekker found 6 major themes in this research. Those that are connected to new product development are integral productivity, product life cycle management and integrated process and coordination. Integrated process is mostly focused on managerial types and internal organization. It neglects operational relationships. Product life cycle looks at use of resource, maintenance of products and disposal of products. The third integrated processes and coordination focus on product innovativeness and uses theories such as concurrent engineering and cross-functional collaboration.
The design process influences the continuous process by setting a framework for the level of diversification of later products or similar products. When designing a product, a company can strategically add in functions or modularity into the product to make it easier to integrate with a product family or product platform.
2.4 Product family/platform
The description given in product design phase will set the limits for the product family or product platform that the product will belong too. Product family or product platform indicate a group of products made by a company that share similar features and components. The two terms are used interchangeably here and I will make no special distinction. The product platform approach during product design is connected to a company’s long term strategic planning. Muffatto (1999) uses Wheelwright and Clark’s (1992) argument for why companies use this method. A product platform fulfil two needs, it allows for product differentiation and simplifies the assembly line for the products. When several product share components, the company will be able to develop them faster and the production can happen more smoothly
12
because it uses more of the existing machines. When the volume of both product increases the company will also enjoy economies of scale for the components that are shared. All in all the company can enjoy greater flexibility in production and reduce cost (Wilhelm 1997, Muffatto, 1999).
The platform strategy has considerable impact on the product development process since it will put constraints on product architecture and product variety. The product architecture indicate what will be the components and how they will be made. Here the decision is about level of modularity and integral components. Which means that platform strategy in product development must also contend to the different components and their requirements. Products in the same platform will share basic components and interfaces between those. It would be beneficial for companies to keep to those modular forms when increasing the product family.
However too much focus on modularity can put unnecessary rigidity and lower the flexibility of new product development. When it comes to product platform strategy and product variety, the decisive factor for its success is about the design transfer between platforms. The main decision here is about model/platform ratio, expansion or reduction of number of platform and whether to develop new platforms. In practice companies strive to increase model/platform ratio and often the number of platforms are five according to Muffatto (1997), because along with more product variety there is increased production complexity. However no actual study have found an optimal number of variety. Variety brings value to company but at a diminishing rate, because more variety make it harder to make distinctive the individual benefit of a model (Muffatto, 1999).
The development of product platforms is to both adhere to specific customer demands, shorter product life cycle and increase in mass customization techniques. It is logical that the higher the product variety, the greater the complexity and cost of operation in the production (Magnusson Pasche, 2014). Magnusson and Pasche uses Muffatto’s (1999) definition of platform as “a relatively large set of product components that are physically connected as a stable sub-assembly and are common to different final models”. The use of product family allows the company to better shape the core products, thus reducing system complexity in regards to design transfer between products and lead time and engineering hours. The approach to product platform development is to either focus on modularity or integration of components. Modular design allow companies to decouple components and extract more information and coordination during the production process. A module “is a unit whose structural elements are powerfully connected among themselves and relatively weakly connected to elements on other units” (Magnusson Pasche 2014).
When combining product platform with modularity it is clear that product architecture is the integrating component. Baldwin and Clark (1997) describe the interface in product
13
architecture as information glue that help coordinate decentralized processes that occur in modular production processes.
Having mentioned the benefits that comes from using product platform/family strategies, Magnusson and Pasche also highlight an important aspect, the negative effects of product platform have not received as much attention. One thing that they found is that external environment will also effect on the decision company’s make as to whether or not to use platform strategies. Speed of change in regards to technical or fashionable change in the market will determine the successfulness of platform strategy. Platform strategy is a long term change, so if change happens slowly in the market, this strategy will fit the company better, ceteris paribus. Another discovery is that optimal product structure cannot be fully determined by a modularization or platform strategy. Having too much modularity or platforms can lead to product cannibalization and customer dissatisfaction. These effects are most prominent in industries where the customers are also a business, since they have a more extensive knowledge about the product than private customers. Modularity approach and platform approach may also impose conflicting demands on the organizational structure and decision making process of the company (Magnusson and Pasche, 2014).
However empirical research have been developing in order to help companies make decision to optimize production process on several levels or by taking into consideration more dimensions. Du et al (2014) proposes the Stackelberg game theoretic model, which is a bi-level decision structure. When solving an optimization problem, one starts by looking at the conditions in the problem. In Du et al’s model, they optimize the product family design by looking at module configuration and parameter scaling. The Stackelberg model is suitable for this since it allows for doing a joint optimization problem solving. The module configuration and parameter scaling belong on 2 different levels of decision making. The parameter scaling is subset to the module configuration. The mathematical design of this optimization is a leader-follower model, which is the basic for of bi-level optimization. “The follower’s role can be seen as solving a parametric optimization problem, whose parameters are determined by the leader” (Du et al 2014). The leader and the followers have different decision powers. In the product family design situation, game players can be designers or design teams. The leader has the privilege to design vector X according to its objectives and constraints. The followers then optimize vectors Z inside vector X. this will result in a joint decision module configuration and parameter scaling. Product family/platform are relevant because many articles in literature discuss how architecture can help achieve business goals through product family. Literature are designing mathematical models to help companies make decisions.
14 2.5 Product architecture
Product architecture is about product constraints and components in a product that relates to production process and customer perception. These are the two biggest focus areas in this line of research. Ko (2013) define product architecture as “an abstract conceptual structure that underlies the engineering artifacts”. This means how well the product architecture is designed, will have an influence on product quality, design quality, performance, production cost and customer satisfaction. “It determines how external complexity is translated into physical products” (Marti, 2007). Ulrich (1995) define product architecture as “the scheme by which the function of a product is allocated to physical component” (Marti 2007). For example, the product architecture determines complexity of cost and number of variables in the production process. Product architectures are notoriously difficult to capture because they are qualitative aspects and abstracted from reality. They are shown in the actual modules of a product and are the results of trial and error. Another concept that is often used alongside product architecture is modularity. Balwin and Clark (2001) explained it this way: product design is about finding measurable and clear parameters that can indicate the performance of the production. These parameters come from design tasks. A design structure is a network of these parameters and it shows their interdependencies. The more interdependencies there are between modules, the more complex the design becomes. Modules thus makes the design rules for the products, which establishes architecture, interfaces, integration, protocol and testing standards (Baldwin and Clark, 2001).
All four authors agree that product architecture focus on 3 aspects of the production, functionality, components and mapping relationships between the two.
Many researchers have done quantitative research on product architecture using different design structure matrices, depending on the complexity of the production system. There are four types of designs structure matrices. The DSM describe the relationships and constraints between design tasks. It measures the design parameters in the product and determine in which order they should be put together, (Sharman Yassine,2007). Asikoglu (2012) argue that design structure matrices are beneficial because system design dependencies found in product architecture can be divided into two big categories, direct and indirect connections. As a product becomes more complex, that is increased interdependencies between components, it will lead to what is termed the “ripple effect” (Asikoglu, 2012). One change in a component will result in direct changes in other components but there are also indirect effects that will occur throughout the system, such as changed assembling order. Some architectures are more susceptible to design changes within a product, which can also be understood as more sensitive to ripple effect.
15
It is quiet logical that changes in product architecture will change the assembly or product design of a product. A system that measure component interactions should have parameters that can indicate consistency, simplicity, practicality, objectivity and benchmarking. The DSM can visualize the mapping between components and subsystems within a structure (Asikoglu, 2012). A functional approach in design structure matrix is very useful during new product development.
Another way of describing product architecture is by looking at the production order hierarchy.
As an organization can be horizontally and vertically hierarchical, the same levels can occur in production process. Some steps can be done in parallel to each other and some must happen in a specific order. If a process is complex, depends on what the assembly system looks like.
Marti (2007) breaks down product architectural factors into two big concepts modularity and integral systems. He uses Simon’s model where there is a vertical hierarchical system and a horizontal near-decomposable system (Marti, Simons 1962). Complexity horizontally increases when the system goes from non-assemble to assemble system, which is further divided into closed and open system. The vertical complexity estimates the impact of external complexity onto the internal complexity. A modular system architecture would then be the second least complex one. External complexity will influence the internal but the effect is small. The relationship is weak between external complexity and product architecture. A most complex hierarchical system would be subsystems merge into one single unit. Here external complexity would have the same impact on product architecture as internal complexity.
Hierarchical division of a product is an order of assembly that means adding components to a unit. Whilst a vertical division mean several subcomponents are assembled separately then put together. This obviously depends on design on product architecture.
Product architecture mapping start with functional elements that together form a functional structure. Level of details in a functional structure depends on what is needed by the departments using it. Next it is important to specify contact surfaces between components.
The contact surfaces also called interfaces, defined the protocol that products in the same line follows. Here a product is said to have a typology of either modular or integral. Modular architectures are slot (interface between components are all different and cannot be interchanged), bus (all physical components connect to the same interface) and sectional (all interfaces are the same, but components are not connected to all other components in the product). Most real products have a combination of these typologies. Ulrich focuses less on integral architecture, which he defines as complex mapping where one component can have several functions, there is no one-to-one mapping. Any change in function would mean changes in several product components. Products with more integral components have better global performance compared to more modular architectural products.
16
Optimization problem is no longer so straightforward, as it has been, because research has shown that business standpoints, technical efficiency, modularity and integration all effect the value of the final product differently and the connections are not so easily expressed (Otto Weck, 2007). More effort has been made to bridge the gap between the qualitative information that exist about product variety and the quantitative data needed to optimize a product.
Modularity is a good strategy when it comes to optimizing local performance, but increased use of modularity will make the product heavier, slower and less energy efficient. Integration on the other hand makes it more difficult to change a product or create variety in a product family, because a simple change to one function will lead to changes in other areas due to the integrated way of the design.
Early decision making about product architecture during design phase could affect larger production process of an item. This will also affect how flexible it can be to add on variety in the future. It would be interesting to see how small high-tech companies look at this.
2.6 Product variety
Product variety in its simplest form could be different colour on a pencil or prints on T-shirts.
The fundamental product is the same but their differences is important enough that buyers separate these items and see them as unique. Product variety is the result of when external demand differences is big enough and internal capabilities within a company is efficient enough to meet that demand. The phenomenon of product variety depends on the type of product a company is making, whether it is a tangible or intangible good. It depends on how production of this product is shaped, whether the supply chain is international or in-house. It depends on the complexity of the technique used in producing it. Research has discussed product variety in relation to marketing, engineering and scale of production. It is difficult to gain a grounded theory about this due to the idiosyncratic nature of production. Because every product is unique and have product specific characteristics. Ramdas (2003) instead look at product variety decisions. Ramdas has found that similarities in variety decision making process and thus attempt to find most recurring steps. Companies approach variety decisions in two steps, variety creation and variety implementation (Ramdas, 2003). Variety creation decisions is about the amount of variety to offer, the type of variety and the timing of end- product variety. Whereas variety implementation decisions focus on design and operations of internal process and supply chain in order to make variety creation possible.
A company’s whole organization is built to meet the manufacturing demands of a specific product. Variety creation and variety implementation decisions thus effects a company’s long term performance through moulding customer perception, create degree of synergies among
17
products, responsiveness to demand uncertainty and increasing manufacturing flexibility (Ramdas, 2003).
The paper by Fisher, Ramdas and Ulrich (1999) discussed general product variety issues such as how many different products can use the same product design, how components developed by suppliers will impact production process, how much to invest in new product development.
They conclude that companies should focus on developing well those components that have a strong impact on customers’ perceived product quality. Their quantitative study on automotive manufacturers, looked at sales data using Monte Carlo gamma distribution. While that is relevant for product variety, a better method has been developed by Arciniegas and Kim (2011) that focus on optimal component sharing in a product family. Their method is more realistic l argue, because they attempt to optimize multiple products using a design structure matrix. Their paper take inspiration from Ulrich (1995) about function allocation to physical components. What they discovered is that architectural decisions effect the current products and also products developed in the future. When developing multiple products, the DSM for each one will be different when it comes to size and content. Simpson (2004) say that an important factor to consider is which components should be shared across the product family. Metrics that has been used for this endeavor are graphs/networks, matrices and commonality indices (Wang Antonsson 2005, Sosa et al 2007, Stewart 1981, Browing 2001).
Arciniegas and Kim (2011) goes on discussing how these matrices can be used for finding minimal description length, coupling index for interfaces.
The framework that they introduced is called sharing decision and optimal clustering framework. This framework identifies candidate for component sharing based on functional description and IM score. The matrix will show 1 for component sharing and 0 for unique component. There are many different ways that a company can work with these concepts and one aspect is interesting to do a comparison on is the methodology used to discover the rows and columns of the matrix. The design structure matrix is what linked these two concepts together. There are many different types of structure matrices, such as dependency, design etc.
Product variety focus on how easily a component can be applied to several products.
Dependency structure matrix can be drawn at different levels of details, such as task level, parameter level, team level, component level, based on the type of problem. This type of dependency structure matrix are useful in visualizing the connections in architecture. Other
“pictorial models” are CAD, function diagram, interaction graphs. Sharman and Yassine (2003) prefer the DSM because the other models are hierarchical and complex architecture can easily violate the rules for such graphs. Although the matrices used for variety and architecture are similar, the analysis done to them are different. With product variety the analysis is about optimizing. Product architecture is discussed in terms of buses, chunks and
18
modules. Chunks are the smallest singular component. Chunks can be connected into buses.
Buses can be organized into modules. The design rule for a module is that parts inside a module have the same relationship with other parts inside the module and the same relationship with parts outside of the module. The reason for these rules is to lessen visual complexity and make it easier to use a type of algorithm called clustering. Clustering allows for smaller matrices to appear in bigger ones. It visualizes connections inside and between different clusters. It easily shows which components are used to achieve a function etc.
(Sharman Yassine 2003).
To summarize, a company uses strategy to maximize the profit they can make from existing and available resources. The resource based view provides a framework that allows company to plan for short term and long term steps in order to achieve this. A big part of strategy and planning is decision making. Research about decision have split the reason behind it as either rational or irrational. New studies have added intuition as a way to bridge between the two.
A company that has decided to make a new product will have to track both the management strategy and new product strategy. Because management will track the response from submarkets and production of new product will tie to the existing capabilities in a company.
Here are also space for decision making because all understanding about the past and current situation will lead to a decision. The new product must be within the company’s capabilities and what the market are asking for. The product design will then realize the markets demand and deal with the practical part of production process. One way to process this information is to use mathematical models to understand product dimensions and constraints for changes that needs to be made for the new product to be produced. The dimensions and constraints comes from existing products and supply chain. Many larger companies have an existing product portfolio and a common strategy is to leverage this and include as many of existing components into the new product without taking away from the innovation. Product platform as a strategy helps company keep their supply chain from becoming unnecessarily complex.
Using existing components would also mean adding flexibility and modularity to the new product.
And this takes us into the product architecture. Existing components need to be able to perform the function that the new product have. In this research area there are several optimization methods that deal with cost, space, weight. An optimization problem could be to find level of optimal modularity and integration of function into components. That is one component only does a part of function or instigate several functions. The parallel running decision is about product variety. Product variety is relevant when a company brings a new product into their portfolio, partly to see the profit and costs that is added to the company and partly because the product needs to make sense to the strategy of a company. That is why it is important to look at product architecture and product variety together with decision making.
19 2.7 Decision making literature
Decision making is a process that a human performs in his/her mind before choosing a course of action. Decision making is often the result of a combination of rational information processing, intuition, risk preference and learning. These are the biggest areas of interest in the research field about decision making (Banai 2005, Garcia Penalvo Conde 2014, Zeigenfusse et al 2014, Young et al 2012). An underlying tool for many of these studies is calculations about probability and weighted calculations. Decision making are done in context to a situation which entails certain conditions that can be judge relevant beforehand and the development of conditions can be guessed beforehand. In the field of economics there is the assumption that a person is utility maximizing and rational. That means given all information, that person will make the choice that benefit him/her the most. It sets a specific assumption of how information is processed and how the choice is made. In the real world there is rarely only one way to make a decision. However most of these decision making evaluations are based on this assumption set in economics.
Zeigenfuse et al (2014) look at how subjective values influence a person’s arrangement of alternatives and thus influence choice. They begin with an example of people selecting pieces of noisy information and make choices that lead to uncertain outcomes. The game that the participants in this research played is to select the most amount of dots. This study looked at behavioral development, choice and response time. It showed that “preferential choice under uncertainty” give significantly different outcomes compared to “when participants made perceptual judgement under uncertainty” (Zeigenfuse et al, 2014).
Another study about informal learning in employees done by Penalvo and Conde (2013) show that informal learning is the basis for lifelong learning, it is connected to observation and experience, it can enhance employability, and positive benefits to mention a few. Naturally many models have been developed to try to capture that effect. The one that Penalvo and Conde (2013) uses starts by identification and storage, organization, analysis and it results in learning. Conclusively the article state that different companies have different needs and therefore the organizations had need for different information. The key towards greater informal learning is also in finding new indicators to solve specific organizational needs.
There are also psychological mechanisms underlying decision behavior. Common method to study this is using compensation as an approximation for utility maximizing. Some patterns that has been discovered are individual perform worse under time pressure, amount of deliberation on a decision and confidence in that decision are inversely correlated. Increased time pressure led to greater risk taking for positive expected values and greater risk aversion
20
for negative expected values (Young et al, 2012). There are many influencing factors behind decision.
2.7.1 Decision challengers in start-up companies
This thesis focus in part on decision making in unstructured new environment. The conditions in a start-up company is very different from that of an established company. For example, it may lack formal protocols for decision making, information processing and approval processes. Start-ups benefit in having flexibility and less complexity compared to bigger companies. But they also lack the long term expertise that big companies have. There are little experience to rely on when it comes to parameter decisions. After having been an intern at Luxbright for a year, I would use these words to describe the decision process. There are risk seeking behaviors in the company. Many decisions are made with incomplete information and with large uncertainties. A single person’s learning from previous decisions make a huge impact on the outcome of the next decision.
On the other hand, research about literature require decision making to be done in an orderly manner that are often found in larger companies where information is spread out between many people. It takes longer time to gather the necessary information and many more people are involved in the final decision. This will complicate the decision process compared to smaller companies where fewer people are involved. The way research split decision making process up into steps could lend itself better to describe this process in larger companies.
However smaller companies have clearer accountability for decisions made since it is more clear who made what decision. Without the decision structure of bigger companies, smaller companies could take advantage of informal learning in workplaces. This type of learning can help employees develop task-oriented skills and develop competences cost-effectively (Garcia- Penalvo & Conde, 2014). It is very possible to develop very different decision making style in a start-up compared to a larger company l think.
2.8 Articles of key interest
There are a lot of literature in the cross-section about production process and decision making.
In order to have more concrete comparison points between the literature and the interviews, I have selected 3 articles that touch upon, for me, the most interesting areas of this research.
These articles are about quantitative studies that combine product architecture and product variety with decision making, parameter determination and product life cycle management. I think these are the territories most influential early on during product development phase. The hypothesis are strongly connected to discussion in the articles, which will make comparison
21
much easier later in result. The results are also generalizable and therefore of interest to ask to start-up companies to see on what level they agree or disagree with it.
Wan and Dresden (2015) quantitatively test decisions impact on product variety. The mathematical model that they built, can be used to help a manager better control the outcome of product variety decisions. Product
variety bridges into marketing and supply chain management and as such it can increase a company’s market share and control cost in larger production supply chains. A product with more modularly design components would allow for an increased variety without increasing cost with the same ratio. Increased variety could attract variety seeking customers. Variety will also make supply chain more complex. But the
right type of product architecture would allow for more variety without it effecting the marginal cost or cause it to increase.
H1: Product variety has a positive curvilinear effect (at a diminishing rate) on customer demand.
H2: Product variety has a positive curvilinear effect (at a diminishing rate) on production and operational costs.
H3: Past customer demand has a positive effect on present product variety
H4: Past production and operational costs have a negative effect on present product variety.
This article is interesting because it touch upon external and internal aspects that a company alters when variety increases. Three reasons that are given for hypothesis 1, is that the more segmented the market is, the harder it will be to find an additional one to add to it. Increased variety will lead to larger inventories and higher production and operational costs. Adding product variety could be a short term decision due to the speed of demand change in some cases.
A more encompassing view of the whole production process, is the product life cycle management (PLM). Tai (2017) combines this with new product development. The environment is such that manufacturing firms have less time to develop a product compared to a decade ago and customers are demanding more preference specific items. As such companies need to leverage their network and utilize the external resources they have through
Figure 3 Tai (2017) Overview of PLM system in NPD
22
partners and collaborators. Tai (2017) study PLM in conjunction to process management capabilities, coordination capabilities and absorptive capabilities.
These are according to him critical for good management. Product life cycle management (PLM) are about information flow and how software can help companies make sure that complex and interdependent activities reach the right places/people inside the company. It’s about streamline business processes and “continually improve and optimize process execution and control” (Tai, 2017). The hypothesis that are in the article are:
H1. Process management capability is positively related to NPD performance H2. Coordination capability is positively related to NPD performance.
H3. Absorptive capability is positively related to NPD performance.
H4. PLM system capability is positively related to process management capability H5. PLM system capability is positively related to coordination capability
H6. PLM system capability is positively related to absorptive capability
These were all confirmed in the regression analysis and were significant. Hypothesis one looked at company’s ability to regulate and cost control a new product development process.
Hypothesis two looked at company’s ability to coordinate partners resources to achieve own strategic goals, based on information sharing. Hypothesis three looked at company’s ability to create value from information given by external sources. Hypothesis four looked at company’s ability to internally coordinate production process and make decisions based on the available information. Hypothesis five looked at company’s ability to standardize workflow and objectives inside the company and lastly hypothesis six looked at company’s ability to routinize product lifecycle management (PLM) (Tai, 2017).
The third article l have chosen discuss methodology to identify product platform parameters.
A good method to identify parameters would will help company with decision support and achieve maximum commonality among components in products whilst allowing for largest variety. The article mention product platform concept exploration method, variation-based platform design methodology. One way to mathematically determine relevant components is to look at uncertain relations between variable and performance of product. These calculations can be done using tandem evolutionary algorithm or other similar ones. However it does not answer how these variables are selected, remarked Gao et al (2014). Below are summary of key points that this article makes, written by me.
P1: Designers during the innovation phase will determine all variables that will enter into a product and thus knows the product intimately.
23
P2: Platform parameters are chosen in such a way that only those that influence product performance in the same way in all products are common.
P3: The membership degree is a sensitivity analysis that determines either number of interfaces or relative assembly complexity.
The discussion further specify that this thesis focus on scale-based platform, which is relevant because during new product development there is a transition into larger scale production.
The article gives list of factors to consider when choosing parameters such as design factor, die factor, assembly factor and performance factor. Assembly factors touch upon interfaces and modularity.
I am interested in this research because start-ups have a higher failure rate and are more susceptible to consequences from bad decision and product design failures, it seems. Little research has been done about start-ups in regards to this and a second challenge is also to genuinely describe the organizational structure of a start-up. So much of the product architecture and product variety are very product specific and many articles have tried to generalize it by using large sample sizes or computer generated models. Both of which rely on similarity in classification and similar structural processes. Start-ups have more “chaos” in their organizational structure and many “departments” are not clearly separated from each other. Being new also makes the comparison harder because the working environment there is more similar to a project or team environment rather than that of a company, meaning more customised to the present needs of a company. All this ads to the difficulty to explain why some start-ups are successful and some are not. The innovation and what makes a start-up competitive is in the components of the product. This is true to any company. Most articles about product architecture and variety look at companies that are adding to their portfolio.
Start-ups are only working on the first product. Everything from production design to supply chain is built from scratch. How useful are the models from literature to start-up companies?
In what sense are production process different when comparing start-ups to larger companies?
Do start-ups use computer-based models? The 3 key articles above are used as a guide to sharpen the interview questions posed to the start-up company Luxbright. They touch upon the 3 most relevant areas within this research field. And would give a better comparison point for the interview answers. I will also discuss the answers in general and in comparison to the impressions I have made about where research is concerning product architecture and product variety.
