A case study of the dynamics of the market share
of a Taiwanese smartphone company
by Yin-Yi Rachelle Yang
June 2019
Student: Yin-Yi Rachelle Yang Student number: S1030022
Master specialization: Business administration
European Master Programme in System Dynamics (EMSD)
Supervisor: Prof. Dr. Andreas Größler
Institute of Business Administration, Operations Management Department University of Stuttgart, Germany
Co-Supervisor: Prof. Dr. E.A.J.A. (Etiënne) Rouwette
Chair of Research & Intervention Methodology, Nijmegen School of Management Radboud University Nijmegen, the Netherlands
2nd Supervisor: Prof. Dr. Pål Davidsen
ACKNOWLEDGMENTS
It is really a wonderful experience to work on my master's thesis at Radboud University Nijmegen with the great supervisory of Dr.Andreas Größler from the University of Stuttgart, Germany, Dr. Etiënne Rouwette from Radboud University Nijmegen, the Netherlands and Dr. Pål Davidsen from the University of Bergen, Norway. In Particular, I express my sincere
gratitude towards Prof.Größler for his great supervisory, enthusiasm and attention to details that continually guides me with a rigorous research perspective and discipline, and a deeper reflection to improve my master thesis to the higher level. It is indeed my greatest honor to learn from him and work with him. I would also like to thank Dr. Etiënne Rouwette for his support when I suffered from a terribly horrible bike accident in early November in 2018. Without his
understanding and encouragement, it would be even tough and challenging for me to survive the difficult time. I enjoy the course of GEO-SD303" Model-based Analysis and Policy Design" by Dr. Pål Davidsen a lot at the University of Bergen, Norway. It motivates me to dive deeper into the real-world applications in system dynamics. I am truly grateful for all of your efforts and supervisory to make my journey of joining European Master Programme in System Dynamics meaningful and worthwhile. Lastly, I would like to show my deep gratitude and love towards my beloved parents in Taiwan and my dear brother's family in the US who are always by my side with constantly selfless dedication and encouragement. I wouldn't have a chance to join and complete the master program without their unconditional love and support. I miss you all a lot. And I am looking forward to the gathering soon! Thank you!
TABLE OF CONTENTS
1. INTRODUCTION _____________________________________________________ 6 1.1. RESEARCH OBJECTIVE ____________________________________________ 8 1.2. CONTRIBUTION OF THE WORK _____________________________________ 8 1.3. RESEARCH STRATEGY _____________________________________________ 9 2. THEORETICAL BACKGROUND _______________________________________ 10
2.1. SMARTPHONE INDUSTRY__________________________________________ 10
2.1.1. How smartphone companies operate ___________________________________ 11
2.1.2. Case study company, HTC ___________________________________________ 13
2.2. SPEED TO MARKET _______________________________________________ 14 2.3. PRODUCT QUALITY _______________________________________________15 2.4. RELATIONSHIP OF SPEED TO MARKET AND PRODUCT QUALITY______ 16
3. METHODOLOGY _____________________________________________________ 18 3.1. SYSTEM DYNAMICS _______________________________________________ 18
3.1.1. System dynamics model- project management ____________________________ 18 3.1.2. System dynamics model- product adoption and diffusion ___________________ 20
3.2. MIXED RESEARCH STRATEGY _____________________________________ 21
3.2.1. Qualitative research ________________________________________________22 3.2.2. Quantitative research _______________________________________________23
3.3. RESEARCH ETHICS ________________________________________________ 23
4.1. PROBLEM DEFINITION __________________________________________ 24 4.2. DYNAMIC HYPOTHESIS _________________________________________ 25
4.3. FORMULATION OF A SIMULATION MODEL _______________________ 28
4.3.1. Formulation of project management model ___________________________ 29 4.3.2. Formulation of product diffusion and adoption model ___________________ 31
4.3.3. Formulation of a simulation model- Graph functions ____________________ 34
4.3.4. Formulation of a simulation model- Numerical data _____________________40
4.4. BASE RUN RESULTS AND MODEL ANALYSIS ______________________ 41
5. POLICY INTERVENTION ANALYSIS _________________________________ 45 5.1. POLICY LEVERAGE POINT: PARAMETER NORMAL TIME TO DISCOVER
REWORK ___________________________________________________________ 45 5.2. POLICY LEVERAGE POINT: SIMULTANEOUS CONSIDERATION OF BOTH
NORMAL TIME TO DISCOVER REWORK AND MINIMUM TIME TO PERFORM
A TASK _____________________________________________________________ 50 5.3. OTHER PARAMETERS ____________________________________________ 55
5.3.1. Time to change perception for discard ________________________________ 55
5.3.2. NORMAL FRACTION CORRECT AND COMPLETE ____________________ 56 5.3.3. Normal product lifetime ___________________________________________ 56
5.3.4. Effect of product quality on product lifetime ___________________________ 57
5.4. POLICY EFFECTIVENESS AND COMBINATION ANALYSIS ____________58 5.5. SUMMARY OF POLICY INTERVENTION ANALYSIS ___________________62
6. IMPLICATIONS AND DISCUSSIONS ______________________________________ 64
CONCLUSIONS __________________________________________________________ 68 LIMITATIONS AND FUTURE WORK ________________________________________70
7. REFERENCES ___________________________________________________________ 70
APPENDIX.1 MODEL DOCUMENTATION ___________________________________ 82
APPENDIX.2 MODEL TESTING AND VALIDATION ___________________________ 95 APPENDIX.3 INTERVIEW RESULTS ________________________________________ 108
1. INTRODUCTION
In the high-tech industry under a rapidly changing environment, speed to market is always considered to be a crucial factor that greatly impacts product attractiveness and firm competitiveness, such as market share.
Time-to-market decisions clearly play an important role in determining the ultimate success or failure of a new product (Bayus, 1997). While industry top managers regard speeding up new products as a corporate priority (Hyatt, 1994), other researches take the view of speed to market as the driving forces of change for the ’90s to manage the future (Tucker, 1998), “the next
source of competitive advantage” (Stalk, 1988). However, Crawford also reveals several
“hidden costs” of accelerated development (1992). A particular concern regarding speed to market is that extreme speed may jeopardize product quality (Clark and Fujimoto, 1991). However, some researchers (Kessler and Chakrabarti, 1996) suggest that speed to market even improves product quality. Product quality is customer perception of the extent to which a product or service meets or exceeds their requirements relative to competing alternatives (Sethi, 2000). Kessler and Chakrabarti (1996), Kessler and Bierly (2002) argue speed to market improves product profitability by increasing market acceptance in relation to product quality to meet customers’ expectations. The sooner a firm can launch a new product, the more certain a firm can forecast customer preferences and develop a product concept that customers find attractive (Kessler and Chakrabarti, 1996; Kessler and Bierly, 2002), while others (Clark and
Fujimoto,1991) suggest firms must balance both speed to market and product quality to
and product quality as “balanced excellence,” proposing that managers must make trade-offs between both requirements to strike a balance between speed to market and product quality to achieve success. The mixed and conflicting views regarding the relation of speed to market and product quality arise because the relationship of speed to market and product quality takes an inverted-U shape, where speed to market improves product quality to a certain point after which quality levels begin to degrade (Lukas and Menon, 2004). The finding suggests that both too high and too low product development speed influences product quality in a negative way. Hence it follows that product development speed has to be at a moderate level to gain the best quality results. The reason for the negative effect of low product development speed on quality is that if the development process slows down under a certain level, the slackness within the
company will increase (Gulati and Nohria,1996). On the other hand, if the development speed is too high time pressure will increase immensely (Sethi, 2000) which limits the employees to find better solutions to problems. In this research, a case study of a Taiwanese smartphone company, HTC[1] and its declining market share is investigated using the system dynamics approach. Speed to market and product quality are major concerns and significant predictors of firm performance (Ittner and Larcker, 1997; Clark and Fujimoto, 1991; Jayaram and Narasimhan, 2007). The relationship between speed to market and product quality that determines firm competitiveness is of interest. As a typical firm from the smartphone industry, HTC exhibits extremely high dynamic complexity because of inherent technology complexity, the
interconnectedness between different functional departments, and the competitiveness of its environment. Because of these characteristics, smartphone companies face challenges of rapidly
changing technological complexity, customer requirements, and market environment. In the case study, the impact of time to market on market share, the impact of product quality on market share, and the interaction of time to market and product quality that forms the dynamics of market share are studied. The problematic behavior observed by HTC is a continuously decreasing market share. Once the market share has started eroding, it continues to drop
dramatically. Presumably, it will take considerable effort and time to return to its original market share. [1] https://www.htc.com/ 1.1. Research Objective This research aims to explore the impact of time to market on market share, the impact of product quality on market share, and the interaction of time to market and product quality that forms the dynamics of the market share at a Taiwanese smartphone company in the high-tech industry. The objective of the research is to reveal and analyze the structure underlying the problematic behavior and to provide potential leverage points using the system dynamics approach. Thus, the thesis work aims at answering the following research questions:
(1) What is the interplay dynamics of “time to market” and “product quality” underlying the structure that drives the problematic symptom of declining market share? (2) What is the leverage point to mitigate HTC's dwindling market share?
1.2. Contribution of the Work
to market and product quality are related, simultaneous consideration of both factors enhances insight into their joint effect that creates dynamics of market share. A simulation model
combining project management and product adoption and diffusion is presented to address that speed to market and product quality both enhance product attractiveness. However, the impact of speed to market and product quality is observed after a certain delay that might mislead
managers to make incorrect inferences attributing to the wrong causes. Second, it proposes a policy intervention lever to alleviate the problem. Insights for sustainable business growth are suggested.
1.3. Research Strategy
The research work starts with a background study of the smartphone industry to understand how managers in this industry make decisions in product planning and project management. With the case study, understanding the processes of product development in the smartphone industry is required. Reviewing the relevant system dynamics work and literature on product development and project management provides a fundamental basis for the construction of a simulation model, representing the essential characteristics of project development in the smartphone industry. Quantitative and qualitative data collection are required for constructing the system dynamics model. The quantitative research collects historical numerical data of the company, such as market share. The qualitative research collects and analyses interviews with employees. Figure 1 below illustrates the planning of data collection and analysis in the research strategy.
Figure 1. Research strategy-the planning of data collection and analysis
The thesis work is structured as follows. In Chapter 2, the case study of the smartphone industry is introduced. Relevant literature of new product management relating to speed to market and product quality is addressed. In Chapter 3, the methodology to analyze market share dynamics is presented combining qualitative interview data with quantitative data from historical company records in a simulation model. In Chapter 4, the theoretical discussion and model simulation results and analysis are described. Model sensitivity tests are also shown. Complete model documentation and model testing and validation results are included in Appendix 1 and Appendix 2, respectively. In Chapter 5, policy recommendations based on the findings of the structure that drives problematic behaviors are discussed. The work concludes with the insights and implications learned from the case, and the future work and limitations.
2. THEORETICAL BACKGROUND
2.1. Smartphone Industry
The invention of the smartphone and mobile technology changed the way we live.
Advances in mobile technology have shaped our lives in fundamental ways. Most people read
Literature Review Research Proposal Thesis writing and organizing Qualitative Method Qualitative data collection ( Primary data interview Questionnaire) Coding Data Analysis CLD Quantitative Method Quantitative data collection (Secondary data documentation) SFD Simulation Modelling Problem Replication Model Analysis & Testing Insights & Conclusions
nowadays. Because of the popularity of smartphones and the advances in mobile technology, the world is more connected than ever. People could contact each other instantly and even “meet” their friends at different places with a video call. The smartphone industry is undoubtedly one of the most challenging and rapidly- changing industries. Because of its increasingly irreplaceable importance in our daily lives in modern society, the mobile industry attracts lots of interest from investors and shows an extremely competitive market environment. In the rapidly changing and competitive mobile industry, success never lasts too long. Nokia has dominated the mobile market with a remarkable 49.4% global market share in 2007 (source from CNET, March 2009) [2] and led the mobile technology standards and trends in the mobile empire. Within just a few years, Nokia was no longer a major player in the mobile industry. Reports and researches try to explain the rise and fall of the mobile giant attributing to the main reason are due to the weak position of Nokia in the technological ecosystem failing to meet customers rapidly-changing expectations and to recognize the continuously-changing demand and requirements of the mobile era (Bouwman et al., 2014). In 2014, Microsoft officially acquired Nokia’s mobile device
business that includes Nokia handset division and a license to Nokia’s portfolio of patents for 10 years. It is the end of Nokia handset manufacturing story.
[2] https://www.cnet.com/news/nokia-fights-to-hold-on-to-smartphone-dominance/ 2.1.1. How smartphone companies operate
When it comes to producing smartphones, every smartphone company has its own secret recipe and know-how to integrate hardware and software components, and how to guarantee high manufacturing quality and efficiency to deliver a final smartphone product into the
rapidly-changing market quickly but also with good quality to meet international telecommunication certification. Usually, the procedure of how smartphone companies operate is initiated when the market research helps identify the market focus and customers’ segment. In order to meet the customers’ expectations, the mobile appearance design teamwork with professional hardware engineers to make an initial model of the circuit board. Cost evaluation, technical feasibility, and performance evaluation are the critical objectives at this stage. Once the design is done, all these blueprints of the circuit design are sent to their own or contracted outsourcing semiconductor manufacturing company. A lot of the job at this stage is done in China or India because of the low cost. Once the phone is done, global field testing and laboratory testing are extensively executed and verified. Software teams including the mobile operating system and wireless protocol communication take the main responsibility to debug and help fix the issue at this stage. Acquiring international certification of a mobile phone, such as GCF (Global
Certification Forum) and PTCRB (PCS Type Certification Review Board) is a mandatory step to
guarantee good telecommunication quality and interoperability conforming to global
telecommunication standards when launching smartphones worldwide. The certification process involves a considerable amount of resources including laboratory cost and manpower. After the certification is received, the smartphone phone is considered ready for the market with an advertising campaign. The case study of smartphone company HTC is primarily a smartphone manufacturer. As shown in Figure 2 below to illustrate the smartphone industry relationship chart from components to customers, upstream suppliers provide smartphone components and
parts, and smartphone operating systems. Downstream channels include telecom service providers and distributors and retailers.
Figure 2. Smartphone industry relationship chart.
2.1.2. Case study company, HTC In this research work, the smartphone company HTC is studied during 2009-2015 when the company experienced a dramatic market share loss since 2011. Initially, HTC is a Taiwanese-based manufacturing company, known for its OEM/ ODM PDA (personal digital assistant) products. The company made an outstanding and remarkable success when it brought to the market a combination of a cell phone and a personal device assistant. The company initially grew at a fast pace to win global market share of mobile phone and announced its own brand afterward in 2006. In 2008, HTC also became one of the five biggest mobile phone producers to serve Nokia, Samsung, Motorola, LG, and Sony Ericsson in 2008. The company hit its highest historical record of global smartphone market share of 9.1% in 2011(Source from HTC 2011Annual Report) [3]and also ranked amongst the world's top 5 manufacturers of smartphones and the top ten manufacturers of mobile phones. In 2011, the Mobile World
Congress (MWC) named HTC its "Device Manufacturer of the Year"; that same year, HTC was also listed by Interbrand as one of the world's 100 most valuable brands. HTC’s successful story inspired Taiwanese’s confidence and showed recognition towards the country’s identity,
nationals’ discipline and creativity. The employees working for HTC showed high cohesion and commitment towards the company even under demanding and stressed working environment. They felt proud to be part of the company, to work on the product, to achieve what has never been done before. However, the market share kept falling after 2011. The sudden fall of the company changed the whole mobile ecosystem in Taiwan, and the dramatic talent flow impacted greatly in other relevant companies and industries. Not only resigning employees of HTC but also most Taiwanese people feel connected to the company brand. This research work aims to reveal and analyze the structure underlying the problematic symptom of market share loss. The lessons learned from the case study should be remembered and educated to the firm managers for sustained business development and growth. Figure 3 describes critical and major events and milestones in HTC’s history from 1997 to 2018.
Figure 3. Major events happening in HTC's history from 1997 to 2018. The source is from HTC official website http://topic.cw.com.tw/2017htc/index.html
[3] http://investors.htc.com/phoenix.zhtml?c=251354&p=irol-reportsAnnual 2.2. Speed to Market
financial (i.e., profit, sales, payback period, costs); 2) market-based (i.e., market share); and 3) technical (Montoya-Weiss and Calantone, 1994). The case study in this research is market-based research with a focus on its market share. In particular, the relationship among speed to market and product quality is emphasized. Speed to market is also referred to as development cycle time and is the time elapsed between initial development of the product idea and ultimate
commercialization (Clark and Fujimoto, 1991; Griffin, 1993; Kessler and Chakrabarti, 1996), which is also named time to market. The sooner a firm is able to launch a new product, the easier it is to forecast customer preferences and to attract more customers (Niedrich and Swain, 2004), to obtain first-mover advantages (Dröge et al., 2000), to enjoy the advantage of pricing freedom (Smith and Reinertsen, 1991), and to standardize technological settings (Golder and Tellis,1993). These advantages result from the firm's competitiveness over rivals and are expected to result in dominant market positions (Smith and Reinertsen, 1991). Some researchers (Ali et al.,1995) suggest that the importance of speed to market is related to the degree of product innovativeness which addresses the question of whether increasing development speed is uniformly successful in improving profitability across new products that differ in their degree of innovativeness (Ali, 2000). Generally speaking, development speed and new product profitability are believed to be causally related. While a number of studies show positive results (Lynn et al., 1999), some demonstrate mixed results (Ittner and Larcker, 1997), and others present no evidence of a
relationship between development speed and new product profitability (Griffin, 2002). 2.3. Product Quality
success and profitability (Sethi, 2000) where product quality refers to customer perceptions of superiority relative to competing alternatives in dimensions such as appearance, performance, workmanship, and life/durability (Aaker and Jacobson 1994; Buzzell and Gale 1987; Clark and Fujimoto 1991; Garvin 1988; Jacobson and Aaker 1987; Phillips et al., 1983). Product quality is also known as product superiority or uniqueness (Cooper, 1979), product advantage (Henard and Szymanski, 2001; Montoya-Weiss and Calantone, 1994), product performance (Bayus, 1997; Swink et al., 2006), product attractiveness to the customers. (Clark and Fujimoto, 1991). Abundant firm-level studies on the importance of product quality (Clark and Fujimoto 1991; Menon, Jaworski, and Kohli 1997) reinforce the view of the crucial role of product quality in
influencing product attractiveness and firm’s competitiveness. 2.4. Relationship of Speed to Market and Product Quality
Time-based competition of supporting speed determines the success of a new product has held its long and abundant research tradition in the field of new product development
(Blackburn,199; Stalk, 1988). Espousing that speeding new product to market is necessary for today's dynamic environment, Bayus (1997) further identifies explicitly the tradeoff between time to market, development costs, and product performance. He discusses the relationship between product development time and costs and formulates a mathematical model that
simultaneously considers the decisions regarding time-to-market and product performance levels. The analytical findings in his work show that the fast development of low-performance products is optimal for markets with a relatively short product lifetime, a weak competitor, and relatively high development costs. For example, if the competitor is weak, high-performance levels are not
necessary and the firm can safely reduce time-to-market. While the fast development of products with high-performance levels is optimal under conditions of a relatively long product lifecycle, relatively high sales, and relatively flat development costs. With a long product lifecycle, stable margins, and high sales, the firm can generate sufficient revenue to compensate for the increased cost incurred in speeding a high-performance product to market. Proponents of the traditional view suggest trade-offs to be made among speed to market, quality, and costs in order to
maximize product profitability. However, other researchers (Stanko et al., 2012) argue that faster speed to market is related to better quality and lower costs indicating speed to market and
product quality could be achieved simultaneously without sacrificing each. The work of McNally et al. (2011) further provides evidence that speed to market and product quality are related, and thus simultaneous consideration of both factors affecting product profitability is required. They argue speed to market and product quality both enhance product profitability, but the impact of speed to market is larger than that of product quality based on their model analysis using partial least squares (PLS). The structural model indicates speed to market and product quality both are significantly associated with product profitability while speed to market (with path coefficients and significance levels = 0.44) exhibits a stronger effect than product quality (with path
coefficients and significance levels = 0.21) from the results of their hypothesized model. In addition, speed to market is associated with increased product quality to a certain point after which quality levels begin to degrade, the speed to market–product quality relationship exhibits an inverted-U shape (Lukas and Menon, 2004). This means that superior new product quality is less likely to be achieved when product development is conducted too quickly, or too slowly.
The best product quality is achieved at a moderate development speed. 3. METHODOLOGY 3.1. System Dynamics Because of the bounded rationality and limited information processing capabilities of human beings (Morecroft, 1983; Sterman, 2000), it is usually difficult for business managers to identify major factors and their interconnectedness that impact a firm’s performance and competitiveness in a complex and dynamic environment. System dynamics is a computer-aided approach that offers a holistic representation of complex systems arising in social, economic or ecological systems that are characterized by feedback loops, delays, and nonlinearities (Richardson, 1991; Sterman, 2000). In addition to understanding the whole system, system dynamics helps identify the high leverage points in systems that can produce sustainable benefits, and avoid policy resistance. (Sterman, 2001). Successful policy intervention in complex dynamic systems requires more than technical tools and mathematical models. (Sterman, 2001). In consequence, system dynamics is well-positioned to explain the causal relationships of a dynamic problem between different elements of the system. 3.1.1. System dynamics model- project management Project management is one of the most successful fields for the application of system dynamics. (Lyneis and Ford, 2007). Because project conditions and performance change over time as a result of feedback, nonlinear relationships,and delay responses, the dynamic complexity of project dynamics increases the challenge to control the project status and performance with certain project specifications (requirements) within a limited time, funding, and resources.
(Project Management Institute, 2000). With a system dynamics approach, project dynamics is understood holistically in terms of the underlying structures that create unintended adverse problematic behavior. In the work of Lyneis and Ford (2007) to assess and evaluate system dynamics work on projects, four model structure groups integrated into project models are categorized: (Lyneis and Ford, 2007) 1. Project features: Taking real project features into consideration, such as development processes, resources, managerial mental models, and decision making to simulate real-world problems, and have direct insights relevant to firm managers.
2. Project control: Project Managers aim to achieve the desired completion time within budget to control the project.
3. Ripple and knock-on effects: “Ripple effects” describes the primary consequences of well-intentioned project control efforts. Modeling ripple effects in projects captures and
leverages the concept of policy resistance. “Knock-on effects” refers to the secondary impacts of project control efforts, i.e., the impacts of ripple effects.
4. A rework cycle: Rework represents a large amount of the time spent in a project to fix the undiscovered errors. Rework discovery can be expected in every project. When more time is spent on rework, this inevitably leads to lower project progress and performance. Rework cycle is the canonical structure of modeling project dynamics.
The rework cycle is considered by Lyneis and Ford (2007) as the most important feature of system dynamics project models that capture the essence of typical project management processes both in software and hardware development. In the rework cycle, initially
undiscovered errors or defects create more work in the future generating a feedback loop that worsens the project status. The concept is illustrated in Figure 4 of the rework cycle model adapted from Cooper (1993).
Figure 4. The rework cycle adapted from Cooper 1993.
In the present research, I will rely on the work of project dynamics by Lyneis and Ford (2007) to address the influencing factors of time to market and product quality on firm’s performance and competitiveness because of the resemblance of the fundamental structures underlying the case study which includes the characteristics of project features, project control, and rework cycle mentioned above.
3.1.2. System dynamics model- product adoption and diffusion In the present work, the essence of smartphone penetration and market share analysis is captured by product adoption and diffusion model based on the framework developed by Bass (1969). Diffusion is s a fundamental process in physical, biological, social and economic settings (Rahmandad and Sterman, 2008). Consumer products often go viral with sales driven by the
+ -+ + + Original work to do Work Done Undiscovered Rework programming error rate rework discovery rate completion rate time to discover rework Error Fraction Productivity Effort Applied
epidemic. The classic Bass diffusion model divides a population into two groups of people: potential adopters who are likely to adopt a product, and adopters who have purchased the product and influence others to adopt. At each point in time, new adopters join the market as a result of two types of influences: external influences, such as advertising, and internal market influences that result from interactions among adopters and potential adopters in the social system. The Bass model states that the probability that an individual will adopt the innovation is linear with respect to the number of previous adopters given that the individual has not yet adopted it. The Bass diffusion model is translated in a system dynamics diagram in Figure 5 (Sterman, 2000) which is easier to understand and to be extended compared to the classical expression of mathematical equations.
Figure 5. The Bass diffusion model (1969) in a system dynamics diagram.
3.2. Mixed Research Strategy The research will apply system dynamics approach with a case study to explore the key factors and underlying structure that create dynamics of market share. A mixed research strategy
+ Potential Market Customers of HTC smartphone Attractiveness of Relative Product Quality on adoption Fraction adoption rate of HTC smartphone Total Market Potential adoption from innovation adoption from imitation coefficient of innovation Adoption Fraction normal adoption Fraction Attractiveness of Relative TTM on adoption Fraction coefficient of imitation
combining qualitative data and quantitative data is adopted. Qualitative data in the form of a questionnaire is applied to build the dynamic hypothesis of the conceptual model. Quantitative data includes sources from credible online smartphone market data and the website data of the smartphone company. The wise balance between quantitative and qualitative models
(Richardson, 1996) adopts both quantitative and qualitative data sources and methods. Because it’s believed that retrieving qualitative data sources of rich and detailed information embedded in the mental data is irreplaceable and couldn’t be obtained from quantitative data (Forrester, 1992). In addition, Homer and Oliva (2001) suggest that quantitative data and simulation nearly always adds value beyond mapping alone in most cases even with significant uncertainties about data and the formulation of soft variables. Only with quantified simulation, we could infer dynamics
from the structure. (Homer and Oliva, 2001). 3.2.1. Qualitative research
Qualitative research starts with primary data collection in the form of interviews based on a questionnaire targeting the employees working for the smartphone company during the
observation period 2009-2015. Selection of the employees is based on a purposive sampling of representative employees from four major and distinct functional departments inside the company, that is, software department, hardware department, product champion planning, and customer account management. These four functional departments are considered the most crucial and fundamental departments to determine the product development process. The first-hand qualitative data is valuable because it is intended to elicit the mental models and insiders’ expert perspectives embedded in employees from different departments in a structured and
interactive way. Based on the interviews coded by the same data collector, an initial stock and flow simulation model based on the causality revealed from qualitative interview data is
constructed. A causal loop diagram to explain the underlying structure based on interview results is also illustrated and will be used as a communication tool with interviewees because a causal loop diagram is considered easier and more friendly for visualization and understanding the feedback and loop structure compared to a simulation model. The causal loop diagram
incorporating the perspectives from four departmental managers will be further examined and validated by each interviewee again which serves as the foundation of the dynamic hypothesis of the simulation model. 3.2.2. Quantitative research Quantitative research is started from secondary data collecting, such as online historical data of the smartphone company. The quantitative data is used to provide problematic behavior of
reference modes, and the parameters to build the simulation model. 3.3. Research Ethics
The research is intended only for academic research and master thesis. The only ethical concern is during the interview for qualitative data collection, the employees from the smartphone company are forced to provide deep reflection and recall their past memories of four to eight years ago. This might cause some unintended emotional burden, and pressure to recall some unpleasant moments and feelings. In addition, it might also run the risk of revealing uncovered hidden internal culture and routines of the smartphone company. However, care is taken before, during and after the interview to reduce this ethical concern by providing the interviewees with
safe and comfortable support with confidentiality and anonymity. Before the interview, the purpose and the goal of the interview, and the way to conduct it are communicated sufficiently early and clearly with each interviewee in the form of a questionnaire. Interviewees review the questionnaire and decide if they are willing to join. After receiving the confirmation from interviewees, the researcher provides sufficient time and freedom for interviewees to provide feedback. The interview results are reviewed and assessed by the interviewees in a formal report before it becomes public. The implications and the insights learned from the research will also be delivered in a formal report to interviewees.
4. SYSTEM DYNAMICS MODEL 4.1. Problem Definition In the work, the case study of the smartphone company is investigated with a focus on the interplay dynamics of “time to market” and “product quality” underlying the structure that creates the problematic symptom of declining market share. In addition, the leverage point to mitigate HTC's dwindling market share is proposed. The company experienced dramatic market share loss during the observation period of 2011-2015. The global market share of the
smartphone company, HTC reached its highest record in 2011 at 9.1% (Source from HTC 2011Annual Report) [3], but then kept falling. The reference mode is illustrated in Figure 6. The data is collected from the official annual report of HTC from 2009 to 2015 at the website of HTC. [3]
Figure 6. Reference mode of HTC global smartphone market share in 2009-2015.
4.2. Dynamic Hypothesis Focusing on the critical factors that impact market share, the dynamic hypothesis to build the
simulation model is based on the causality revealed in the interview. For detailed interview results, refer to Appendix 3 of the interview results. The interview targeting the employees working for the smartphone company during the observation period 2009-2015. Representatives from these four functional departments, that is, software department, hardware department, product champion planning, and customer account management are interviewed because these four departments are considered the most crucial and fundamental departments to determine the product development process. When it comes to the question of “ what's the core value and the most crucial factors that are considered most important and put a high emphasis on in your daily routines job at your department?", all interviewees reply immediately “ product quality“ and “ on schedule “undoubtedly. They describe better product quality is the product functions stably and smoothly with lower device crash rate, fewer defects, and less undiscovered bugs. They argue
7 8 9.1 4 2 1.6 1.4 0 1 2 3 4 5 6 7 8 9 10 2009 2010 2011 2012 2013 2014 2015
“product quality“ and “time to market” determine product attractiveness that further impacts market share. The director of product champion planning who is responsible for product position planning, product architecture design and innovation at the conceptualization stage, further explains the product quality also influences the product lifetime. Even in the rapidly-changing mobile industry, customers have a higher tendency to continue to use their old phones implicitly extending the product lifetime instead of immediately switching to the latest models if their smartphones are with high quality. This customers’ behavior extends the product lifetime, and further impacts the discard rate of smartphones. In addition, the director of product champion planning addresses the concern of market entry time and time to market in the product planning and marketing campaign. They point out that time to market is shorter, faster, and better, especially to be relatively shorter than the launch schedule of competitor’s products. They indicate a shorter time to market not only has the advantage of the leading position of advanced technology but also builds a strong positive image and reputation of the company's cutting edge technology that attracts more loyal customers. When it comes to questions regarding market share loss, interviewees start to blame the failure on constantly –changing customer’s product scope, increasing workload, work overtime and the wrong cooperate strategy. All interviewees point out product diversification is the main cooperate strategy when faced with declining market share. More new projects and product lines are simultaneously planned and ongoing in order to regain market share. The hardware and software engineer managers state that there are really too many projects simultaneously ongoing that team members feel overwhelmed and confused with various product lines and product models because of the product diversification strategy. Product
diversification leads to more work, and inevitably also results in more rework. In addition, engineer managers complain constantly-changing product scope that leads to more unpleasant and unanticipated work and rework. Once the vicious circle starts, it gets worse because of accumulating work and rework. Increasing workload deteriorate product quality gradually because lasting fatigue results in a lower concentration of employees, less efficiency, lower accuracy, and higher error rates. Based on the interview results, the dynamic hypothesis is illustrated in Figure 7.
Figure 7. Dynamic hypothesis.
In Figure 7, three major reinforcing loops R1, R2, R3, and one balancing loop B1 are identified. Longer TTM results in lower attractiveness of the product, lower adoption rate, and lower market share which is confirmed by all interviewees. Researches (Bayus, 1997; Stalk, 1988; Lynn et al.,
-+ + -+ + -+ + + + + -+ + -work to do
Product life time Relative TTM Market Share of HTC Actual TTM "... 1" Pressure from Market Share on work to do Customers of HTC Smartphones Attractiveness of Product Quality on adoption Fraction Adoption Rate of HTC Smartphones "..." Adoption Fraction Product quality Attractiveness of Relative TTM on adoption Fraction '' ''' Discard Rate R2 B1 Time to market matters. Product Quality Matters. Customer Discard R3 R1 Product Quality Matters again!
1999) also support the view of the advantage of speed to market. Lower market share implies more research efforts and more work to do on to modify the original work, to improve product design, or diversify product category to attract more customers of different market segments. This explains the reinforcing loop R1 implying the importance of time to market. Abundant researches (Chatterjee and Wernerfelt, 1991; Hoskisson and Hitt, 1990; Montgomery and Hariharan, 1991; Wernerfelt, 1984; Buhner, 1987;Grant et al.,1988; Han et al.,1998) provide evidence for a positive relationship to support the corporate strategy of product diversification as a solution for competitive advantage, growth, and the survival of firms (Montgomery, 1994; Penrose, 1959; Rumelt, 1974). However, more work to do deteriorates the product quality gradually when working overtime becomes a routine instead of an exception. Lower product quality leads to lower product attractiveness to customers resulting in lower adoption rate, fewer customers, and decreasing market share. Under the high pressure of declining market share, more research work of product diversification or better design is required which further deteriorates the product quality again. This explains the reinforcing loop R2 implying the importance of product quality. The defect products with lower product quality shorten product lifetime, higher discard rate, lower customers and decreasing market share that further leads to more work to do. This explains the reinforcing loop R3. The balancing loops B1 leads to even fewer customers because of higher discard rate.
4.3. Formulation of a Simulation Model
According to the dynamic hypothesis in 4.2, the simulation model is constructed based on the work of project dynamics by Lyneis and Ford (2007) and the product diffusion model by Bass
(1969). The simulation model mainly consists of two subsectors. One is the project management model that depicts the smartphone development based on the work of project dynamics by Lyneis and Ford (2007); the other is product diffusion and adoption model based on Bass (1969). Their interconnections are represented in Figure 8.
Figure 8. Subsectors of the simulation model.
4.3.1. Formulation of project management model The left part of the simulation model is based on the project management model by Lyneis and
Ford (2007) to illustrate the canonical structure of the rework cycle revealed in the interview. As elaborated by the software and hardware engineering managers, they explain the product
development inevitably contains a certain fraction of errors (bugs) undiscovered initially and will be discovered by testing. When the bugs are identified, this type of work named as “ rework” accumulates to original work to do. The rework cycle is illustrated with three stocks meaning the tasks planned for the product development in Figure 9 including work to do, work done, and
Project Management Model
Product Diffusion and Adoption Model
Time to market
undiscovered rework. These stocks are defined as "amount of work" in the unit of a task. No further distinction is made between different activities of project development on the type of task, such as, “design” task, “build” task,” “diversify” task, “software” task, and“ hardware” task. All these tasks are treated as identical tasks with uniform size, and with the same degree of complexity and difficulty in the model. Only the number of tasks is used to describe the amount of work to do (workload). The stock of “work to do “ decreases when tasks are progressing either completing the task correctly that moves tasks from “work to do “ to " work done" or with
undiscovered errors that move tasks from “work to do “ to “ undiscovered rework ”.The stock of “work to do” increases when rework is identified. The completion rate that moves tasks from “work to do“ to " work done "is determined by potential completion rate, and fraction correct and complete. Potential completion rate is bounded by the maximum completion rate. The value of fraction correct and complete is determined by normal fraction Correct and Complete and
effectiveness of staff that is also affected by the accumulating workload. Normal Fraction correct and complete is used to represent the fraction of correctly complete work under the normal situation which is set to 0.85 as suggested by engineering managers. Other sources of lower fraction correct and complete may fold but will not be included in the model, such as technical complexity and uncertainty, incorrect prior work, and incorrect assumptions about technology or customer requirements, etc. Product quality is also indicated in the model and is defined as the ratio of work done correctly over total work that includes both work done correctly and
Figure 9. Project development rework cycle. In the present model, Actual Time to Market (TTM), desired TTM (i.e, the desired project
completion time), relative TTM defined as Actual Time to Market divided by desired Time to Market are distinguished. Moreover, every 25 month as suggested by interviewees is assumed for product development cycle in the subsequent product release. The cycle of product
development is 25 months (Desired TTM), which is also a benchmark of the competitor’s launch time. In the base run of the current model, the number of staff is not considered because less attention to the number of employees is addressed by any interviewee. 4.3.2. Formulation of product diffusion and adoption model The right part of the Product Diffusion and Adoption Model is based on the Bass Diffusion
+ + -+ -+ -+ Work to do Work Done Undiscovered Rework rework generation rate( error rate ) rework discovery rate completion rate total work time to discover rework fraction correct and complete MINIMUM TIME TO PERFORM A TASK maximum completion rate potential completion rate product Quality NORMAL FRACTION CORRECT AND COMPLETE effectiveness of staff
Model (1969). The diffusion of technological innovation is a process among firms that transform abstract ideas, concepts, substantial technological information and the needs of society into practical use (Rogers, 1995; Dao and Zmud, 2013). A number of researchers have investigated the diffusion of new products and services (Bass, 1969; Meade, 1984; Gatignon and Robertson, 1985; Mahajan and Muller, 1996; Meade and Islam, 2003&2006.). Among the numerous
diffusion models, the Bass model (Bass, 1969) has been extensively utilized because it can easily depict the effects of external and internal influences. It has been tested in many industries with many new products and technologies. Numerous variations of the original Bass Diffusion Model have been introduced and applied in various fields such as retail service, industrial technology, agricultural, educational, pharmaceutical, and consumer durable goods markets (Mahajan et al., 2000; Mahajan and Muller, 1996; Jerome and Seungmook 1985; Islam and Meade, 2000; Dekimpe et al., 1998; Gruber and Verboven, 2001; Koski and Kretschmer 2005). Bass (1969) modeled new product growth as a function of either mass media or word-of-mouth
communication. According to the Bass Diffusion Model (BDM), diffusion processes are uneven and evolve as innovations are communicated through certain channels over time. The first group adopting the product (or technology) are called innovators (Rogers, 1995) and they are followed by imitators in different levels of the product life cycle. The innovation adoption by innovators is not affected by adopters, while imitative adoption is generated by word of mouth, that is, the communication between adopters and potential adopters. The main dynamic hypothesis is based upon three feedback loops of adoption by innovation, adoption by imitation and adoption by market potential (Bass, 1969; Mahajan et al., 1990). The present diffusion and adoption model is
built upon the Bass Diffusion Model (1969) with the substitution structure by Maier (1998). According to Maier (1998), he argues that competition between several competitors' products and the processes of substitution between one company's subsequent substitutive products are to some extent very similar. Therefore, the present diffusion and adoption model with substitution is adequate to model the case study of the smartphone company, HTC. The simulation model incorporates the repeat purchase or product substitution process as a result of product
obsolescence without considering other competitors' products. Product obsolescence is due to rapidly-changing mobile technology or improved new smartphone models. In the present model, the adoption rate of HTC smartphones is the sum of adoption from innovation and adoption from imitation. Adoption from innovation is determined by the coefficient of innovation and potential market, the total remaining market potential of all the company’s products. Adoption from imitation is determined by existing customers of HTC smartphones, Potential Market, Total Market Potential, and coefficient of imitation while the coefficient of imitation is determined by adoption fraction driven by product attractiveness. With higher product attractiveness, adoption fraction is higher that leads to higher adoption rate, and more customers. In the present research scope, the influencing elements of product attractiveness are time to market and product quality. In addition repeat purchases or product substitution processes are also considered when adopters discard the products and return to potential adopters again because of product obsolescence. Discard rate is determined by the number of adopters and the product lifetime which is affected by product quality perceived by customers. The product diffusion and adoption model is
Figure 10. Product diffusion and adoption model
4.3.3. Formulation of a simulation model- Graph functions The simulation model aims to replicate the behavior of the market share of the smartphone
company HTC during 2009-2015 with a focus on the interplay dynamics of “time to market” and “product quality” underlying the structure that drives the problematic symptom of dwindling market share since 2011. Statistics of global market share are obtained by reviewing the annual report from the company’s official website [2]. Based on this data, the reference mode of the market share of the smartphone company HTC during the observation period of 2009-2015 is illustrated in Figure 6 in 4.1. The simulation model is constructed with the packaged software STELLA Architect Version 1.8.2. The Stock and flow diagram of the model is presented in Figure 11. + Customers of HTC smartphone Potential Market adoption rate of HTC smartphone Attractiveness of Relative Product Quality on adoption Fraction Total Market Potential adoption from innovation adoption from imitation coefficient of innovation Discard rate Normal Product lifetime Product lifetime Adoption Fraction normal adoption Fraction Attractiveness of Relative TTM on adoption Fraction Effect of Relative Product Quality on Product lifetime coefficient of imitation
Figure 11. Stock and flow diagram of the smartphone company, HTC.
For detailed model equations and parameters, refer to Appendix 1 of Model documentation. Regarding the relationship between market share on more work to do, interviewees confirm product diversification is the main corporate strategy when facing with the dwindling market share. More product lines and product models are planned simultaneously that inevitably increases workload and more work to do within limited time constraints and engineering resources. Abundant research (Chatterjee and Wernerfelt, 1991; Hoskisson and Hitt, 1990; Montgomery and Hariharan, 1991; Wernerfelt, 1984; Buhner, 1987; Grant et al.,1988; Han et al.,1998) support the strategy of product diversification to win market share. Evidence of a strong positive relationship between diversification strategy and market share is provided in the recent work of Oyefesobi et al. (2018). Fluck and Lynch (1999) also argue failing to diversify
+ -+ -+ + + -+ + + + + + -+ -+ Work to do Work Done Undiscovered Rework Work Done Work to do Potential Market Customers of HTC smartphones Potential Market NORMAL TIME TO DISCOVER REWORK rework generation rate( error rate ) rework discovery rate total work INITIAL WORK TO DO Desired TTM fraction believed to be complete Desired TTM Desired TTM effect of work progress on time to discover rework Desired TTM time to discover rework Actual TTM effectiveness of staff product Quality effect of Product Quality on product life time Pressure of market share on work to do fraction correct and complete effect of product quality on fraction complete coefficient of imitation Total Staff Technical staff staff change in work amount Target Work To Do Work residence time Actual TTM Product Lifetime Normal Product Lifetime MINIMUM TIME TO PERFORM A TASK maximum completion rate effect of Relative workload on Effectiveness potential completion rate Relative workload industry growth Rate work obsolete industry growth Perceived product Quality for Adoption Time to change perception for Adoption AT for new work to do initial potential Market HTC market share Attractiveness of Relative TTM on adoption Fraction normal adoption Fraction Attractiveness of Relative Product Quality on adoption Fraction Perceived product Quality for Discard Time to change perception for discrad Desired TTM Relative TTM NORMAL PRODUCTIVITY Adoption Fraction productivity Discard rate EFFECTIVENESS NEW STAFF EFFECTIVENESS EXPERIENCED STAFF effectiveness of staff coefficient of innovation rookie fraction completion rate adoption from Imitation adoption from Innovation Total Market Potential NORMAL FRACTION CORRECT AND COMPLETE adoption rate of HTC smartphone INITIAL WORK TO DO
products greatly reduces market share. In addition to confirmation of the strategy of product diversification by interviewees, the relationship between market share and workload is also illustrated in Figure 12 which explains lower market share results in approximately twice amount of the original work to do, while more customers with a higher market share also bring more business and more work to do. The impact of increased market share on workload is more intense and has a higher and rapid influence when market share is relatively low compared to that at extremely high market share. The impact will be gradually level off when reaching an extremely high market share. Figure 12 illustrates that when market share is extremely low close to 0%, the workload is approximately twice the amount of the original work which is estimated and confirmed by interviewees. On the other hand, while market share is extremely high close to 100%, the workload is approximately five times of the original work with the reference point of the initial market share at 7% with workload impact equal to 1 and market share ranging from 0-100 %.
The effect of Product Quality on product lifetime and then on discard rate is determined by normal discard rate and product lifetime affected by the effect of perceived product quality. It’s revealed product lifetime plays a significant role that determines HTC's market share with the observation of rapid market share drop since 2011. Discard rate defined as
Customers_of_HTC_smartphones over Product_Lifetime addresses the product obsolescence because of advances in mobile and telecommunication technology or new smartphone models. In addition to normal discard, customers would discard the smartphones earlier because of the poor quality before the phone is still usable that consequently shortens product lifetime and increases discard rate. On the contrary, if customers feel satisfied with the quality and functionality of the smartphone. They tend to extend the usage period of their smartphones even the phones are not the latest and updated model. According to the Taiwan news report [5], the product life cycle varies between 23 to 33 months during 2009-2018. The average value of 28 month is taken as the value of the normal product lifetime in the model. Time to change perception for discard is set to 28 month as well. The parameter” Time to change perception for discard” is applied in the smooth function of real product quality. The relationship of product quality on product lifetime is illustrated in Figure 13 that suggests when product quality is lower than a certain point, product lifetime is reduced significantly and rapidly while product quality is higher than that certain point, customers have a higher tendency to extend the usage of smartphones. When interviewing with employees, the tipping point of product quality is around 0.75 as suggested by interviewees which indicates when product quality is higher than 0.75, it has a higher chance that customers would extend their smartphone usage, while lower than 0.75,
product lifetime drops quickly. Besides, interviewees argue the possibility of the product
lifetime grows higher than 1.5 times is quite low because Normal Product Lifetime in the model is 28 months. 1.5 times of Normal Product Lifetime will be 42 month (three and half years) which is assumed quite long in the smartphone industry. Thus the relationship of product quality and product lifetime is bounded by 1.5 times of the Normal Product Lifetime. The relationship between product quality and product lifetime reflects the tipping point of product quality. The tipping point of product quality and the value of normal product lifetime, in turn, determines the value of the spike in market share, and also the specific time the spike appears at. If the impact of product quality on product lifetime is more intense with dramatic change abruptly, the spike in market share appears more significantly. With shorter normal product lifetime, the spike in market share reaches higher in terms of absolute value and appears earlier.
[5] https://3c.ltn.com.tw/news/33451
Moreover, a higher workload lowers staff’s effectiveness which further impacts the ratio of correctly completed tasks over total workload, that is, product quality. Research (Deming and Edwards, 1982; Ivarsson and Eek, 2016; kolus et al.,2018;) shows the evidence of linkage of workload, human factors, and quality. Undesirable and unacceptable physical or mental
workload deteriorates product quality because of some human factors, such as fatigue and injury-related risk factors. The effect of relative workload on staff’s effectiveness is illustrated in Figure 14 where the relative workload is defined as Work to do over INITIAL_WORK_TO_DO. When the workload is twice as much as initial work to do, the effectiveness is 80% of the original effectiveness, and only half effectiveness when the workload is three times as much as initial work to do.
Figure 14. The effect of relative workload on effectiveness
The initial value of the potential market and the initial value of customers of HTC smartphones are sensitive model variables and treated as exogenous elements. Other variables including the time to change perception for discard, rookie fraction, effectiveness of staff, NORMAL
exogenous but not sensitive variables. Although the model is sensitive to changes in some
exogenous inputs, the assumption of lower market share leads to more work to do, lower product quality results in higher discard rate, relatively higher workload lowers staff’s effectiveness still leads to robust and plausible model behavior, but cannot duplicate empirical data completely. Therefore, it can be stated that the model has demonstrated its usefulness as a simulator for policymaking in the case of substitution among successive product development.
4.3.4. Formulation of a simulation model- Numerical data According to HTC annual report 2009, it is reported that domestic sales of 394,000 units and
export sales of 36,915,000 units are shipped in 2009 including smartphones and other items such as accessories. In the model, the sales number of HTC’s smartphones is used to represent the customers of HTC, that is, customers of HTC smartphones in the model in the unit of customers. The initial value of HTC customers is set to 394,000+36,915,000= 37,309,000, that is, 37.309 million. In order to meet the starting point of the initial HTC's market share of 7% (HTC annual report 2009), the initial value of the potential market is intentionally set to 532,985,714. The total potential market is set to 1473.4 million which is the total units of global smartphone shipment in 2016 (Source: Gartner, Feb 2017). The value of the potential market, 1473.4 million, is also the maximum yearly shipments during 2009-2018. (Source: Statista 2019, Jan 2019)[4].Potential Market growth is formulated with the reference data (Source: Statista 2019, Jan 2019)[4]. The data is summarized in Table 1.
year
smartphone shipments
(in million units) Yearly Growth Rate (%)
2009 173.5 75.62 2010 304.7 75.62 2011 494.5 62.29 2012 725.3 46.67 2013 1018.7 40.45 2014 1310.7 28.66 2015 1437.2 9.65 2016 1473.4 2.52 2017 1465.5 -0.54 2018 1404.9 -4.14
Table 1. Global smartphone shipments from 2009 to 2018.
Other parameters used in the model, such as rookie fraction, effectiveness of new staff, effectiveness of experienced staff, normal fraction correct and complete are estimated by the interviewees of the smartphone company.
4.4. Base Run Results and Model Analysis Based on the current model structure, the simulation results of the projected market share are
shown together with reference mode in Figure 15. Figure 15 replicates the market share trend indicated in the reference mode in Figure 6 that describes if no policy intervention and no changes of the procedure of current project management, HTC market share starts decreasing
from the peak value at the year 2011 (month 24) with the simulated value of 8.46% and continues dropping.
Figure 15. Simulated result of HTC market share.
The Correlation Coefficient, r, and coefficient of determination, R2 of the reference data and the
simulated data are 0.82, and 0.77, respectively.
Examining the behavioral graph of Desired TTM, Actual TTM and work residence time, spikes at every product shipping cycle (say 25 month in the model) are observed in Figure 16. This happens because every time when a new product cycle starts, work to do increases abruptly with 200 new tasks added, the completion rate is still quite low, work residence time equal to
Work_To_Do over completion_rate creates a spike. As the project progresses, the work residence time gradually decreases reaching the Desired TTM. However, a gap between work residence time and Desired TTM always exists when approaching Desired TTM (say, 25month)
7 8 9.1 4 2 1.6 1.4 7 7.92 8.46 7.76 3.94 1.96 1.12 0 1 2 3 4 5 6 7 8 9 10 2009 2010 2011 2012 2013 2014 2015 Ma rk et S ha re (% ) Year HTC Market Share Reference mode Market share Simulated Market share
residence time is always a little bit larger than Desired TTM. The gap is 8.8 month, 16.9 month, 19.9 month along the time horizon of 25month, 50month, 75month, respectively which indicates the project delay accumulates more and more along the time horizon that leads to longer time to market.
Figure 16. Desired TTM, Actual TTM and work residence time.
Examining the behavioral graphs of the stock of “work done” and “undiscovered rework “, and “work to do “, “work done “is not reached initial work to do (=200) at every product shipping cycle (25 months) which indicates the failed on-time product shipping. The difference between “work done” and “initial work to do” indicated in Figure 17 results in failed on-time product delivery because of accumulating work to do and undiscovered rework when the project progresses. Undiscovered rework is 21.7 task, 43.5 task, 54.8 task along the time horizon of 25month, 50month, 75month, respectively which indicates undiscovered rework accumulates more and more along the time horizon that leads to more work to do, delayed project completion,
Time m on th 0 80 160 240 320 400 2009 2010 2011 2012 2013 2014 2015 1 2 3 1 2 3 1 2 3 1 2 3 Actual TTM 1 2 Desired TTM Work residence time 3
and longer time to market. Undiscovered Rework is one of the most important factors that leads to schedule slippery, and most managers fail to recognize the rework efforts that overestimate real product progress and discourage the reporting of rework.
Figure 17. Comparison between work done, work to do, undiscovered Rework and Initial work to do.
In Figure 18, the product quality is defined as work done correctly over total work, that is, the ratio of work done over total work which is the sum of undiscovered Rework and work done. Perceived product quality is formulated with a smooth function of the product quality with a time delay of “Time_to_change_perception”, that is, SMTH3(Product_Quality,
Time_to_change_perception). In the model, two perceived product quality is distinguished. One is perceived product quality for adoption, while the other is perceived product quality for discard. Perceived product quality for adoption is used to describe product quality perceived by potential customers as product attractiveness when considering to purchase a smartphone, while perceived product quality for discard is used to describe the discard tendency of the existing customers
Time ta sk 0 50 100 150 200 250 300 2009 2010 2011 2012 2013 2014 2015 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 Undiscovered Rework 1 2 Work Done INITIAL WORK TO DO 3 4 Work to do
lifetime. Time_to_change_perception applied for discard (28 month) is shorter than that for adoption (36 month) because it is easier and quicker to destroy the reputation of the product with poor quality than to build an image of a high-quality product. This explains why perceived product quality for discard drops faster than perceived product quality for adoption because of the smaller value of Time_to_change_perception. In Figure 18, the trend of decreasing fraction correct and complete, that is the correctness of the completed work, shows a similar trend to product quality.
Figure 18. Product quality, perceived product quality and fraction correct and complete. 5. POLICY INTERVENTION ANALYSIS
5.1. Policy Leverage Point: Parameter NORMAL TIME TO DISCOVER REWORK Because of the rework cycle, increasing undiscovered rework accumulates that creates more work to do. More work to do further lengthens the time to market and also deteriorates product quality that leads to lower market share. The structure creates a problematic symptom of declining market share. The first policy leverage point on parameter NORMAL TIME TO DISCOVER REWORK is tested because of its direct impact on rework discovery rate. TIME
Time 1 0 0.2 0.4 0.6 0.8 1 2009 2010 2011 2012 2013 2014 2015 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 product Quality 1 2 Perceived product Quality for Discard Perceived product Quality for Adoption 3 4 fraction correct and complete
TO DISCOVER REWORK is the time for the team to identify undiscovered errors during product development which is defined as the amount of undiscovered_Rework over
time_to_discover_rework. Time to discover rework is further determined by NORMA TIME TO DISCOVER REWORK modulated by the effect of work progress on it. Sensitivity test on parameter NORMAL TIME TO DISCOVER REWORK in Figure 19 provides insights for potential policy intervention point. Different scenarios with different values of NORMAL TIME TO DISCOVER REWORK over the plausible range of uncertainty from 0 to 10 (original value of NORMAL TIME TO DISCOVER REWORK is 10 month) are tested in Figure 19
Figure 19. Sensitivity test of “NORMAL TIME TO DISCOVER REWORK" over the plausible range of uncertainty from 0 to 10 on HTC market share during the observation period.
Sensitivity test of NORMAL TIME TO DISCOVER REWORK on HTC market share Time 0 2 4 6 8 10 2009 2010 2011 2012 2013 2014 2015 1234567891012345678 9 10 12 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 NORMAL TIME TO DISCOVER REWORK=1 1 2 NORMAL TIME TO DISCOVER REWORK=2 NORMAL TIME TO DISCOVER REWORK=3 3 4 NORMAL TIME TO DISCOVER REWORK=4 NORMAL TIME TO DISCOVER REWORK=5 5 6 NORMAL TIME TO DISCOVER REWORK=6 NORMAL TIME TO DISCOVER REWORK=7 7 8 NORMAL TIME TO DISCOVER REWORK=8 NORMAL TIME TO DISCOVER REWORK=9 9 10NORMAL TIME TO DISCOVER REWORK=10 Sensitivity test of NORMAL TIME TO DISCOVER REWORK on HTC market share Time 0 4 8 12 16 20 2009 2011 2013 2015 2017 2019 2021 1 23 4 567 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 789 10 1 2 3 4 5 6 78 9 NORMAL TIME TO DISCOVER REWORK=1 1 2 NORMAL TIME TO DISCOVER REWORK=2 NORMAL TIME TO DISCOVER REWORK=3 3 4 NORMAL TIME TO DISCOVER REWORK=4 NORMAL TIME TO DISCOVER REWORK=5 5 6 NORMAL TIME TO DISCOVER REWORK=6 NORMAL TIME TO DISCOVER REWORK=7 7 8 NORMAL TIME TO DISCOVER REWORK=8 NORMAL TIME TO DISCOVER REWORK=9 9 10 NORMAL TIME TO DISCOVER REWORK=10
