Master Thesis
Effectiveness of startup product development processes Research from a lean startup perspective
Date of submission: January 31, 2016 Version: final
Total pages: 73 Total words: 16.757
Executive Program in Management Studies – Strategy Track University of Amsterdam
Author: Sebastiaan Paul Maria Menheere Student number: 10687831
Table of contents
0. Abstract p. 6
1. Introduction p. 7
1.1 Gap in the literature p. 8
1.2 Research question p. 9
1.3 Sub questions p.10
2. Literature review (research context) p. 12
2.1 Entrepreneurial development process (linear versus cyclical) p.12 2.1.1 Linear entrepreneurship (planning school) p. 12 2.1.2 Cyclical entrepreneurship (learning school) p. 15
2.1.3 Linear versus cyclical entrepreneurship p. 19
2.2 The lean startup p. 20
2.2.1 Lean p. 20
2.2.2 Cyclical product development process p. 23
2.2.3 Customer Development p. 24
2.2.4 Evaluation of the lean startup methodology p. 25
2.3 Variables cyclical product development of the lean startup p. 28
2.3.1 Process p. 29
2.3.2 Build p. 32
2.3.3 Measure p. 34
2.3.4 Learn p. 36
2.3.5 Overview of the variables of the cyclical entrepreneurial process p. 41
3. Research design p. 44 3.1 Research type p. 44 3.2 Research strategy p. 45 3.3 Data gathering p. 46 3.4 Data analysis p. 46 3.5 Research trustworthiness p. 47
4. Results, Analysis and Discussion p. 49
4.1 Results p. 49
4.2 Analysis and Discussion p. 49
4.2.1 Process: entrepreneurial approach p. 50
4.2.2 Process: structure of execution p. 52
4.2.3 Build: “Minimum Viable Product” p. 54
4.2.4 Measure: experiments p. 56
4.2.5 Measure: data collection p. 57
4.2.6 Learn: adjustments p. 59
4.2.7 Learn: deepness of learning p. 60
4.3Summary of analysis and discussion p. 62
5. Conclusion p. 64
5.1 Theoretical implications p. 64
5.2 Managerial implications p. 66
6. References p. 69
7. Appendices p. 73
7.1 Appendix 1. Template questions semi-structured interviews 7.2 Appendix 2. Overview of interviews
7.3 Appendix 3. Transcripts
7.4 Appendix 4. Summaries of interviews 7.5 Appendix 5. Code tree after selective coding
Tables
1: Essentials for Successful Startups (Thiel, 2014; Schevernels, 2015) 2: 10x What is Wrong with Linear Product Development (Blank, 2003) 3: Linear versus Cyclical Entrepreneurship
4: Mass Production versus Lean Production (Melton, 2005) 5: Forms of Cyclical Entrepreneurship Compared to Lean Startup 6: Pitfalls of Lean Startups
7: Overview of Variables related to Lean Startup 8: Comparison Waterfall, Agile and Lean Startup 9: Iterative versus Incremental Adjustments 10: Overview of Concepts related to Variables
11: Overview of Entrepreneurial Approach Interviewees 12: Overview of Effectiveness of Variables
Figures
1: Linear Entrepreneurship 2: Cyclical Entrepreneurship
3: Build -‐ Measure -‐ Learn Feedback Loop 4: Customer Development
5: Minimum Viable Product
6: Effectiveness of Experimenting (Maurya, 2012) 7: Incremental versus Iterative
0. Abstract
This study provides a critical examination of how different theoretical perspectives in entrepreneurship research, linear versus cyclical entrepreneurship, translate into current startup behavior, and in which form such behavior is effective within the product development process. Using characteristics of the cyclical lean startup methodology, resulting in 7 separate elements (variables), the effective application of product development is evaluated on the basis of the experience of 13 successful startup founders in the Netherlands. This research gives insights on best and worst practices and highlights similarities and differences with the various theoretical perspectives. The more uncertainty about the product and the market, the higher the reward for choosing a cyclical product development approach. Insights on the effectiveness of cyclical product development within startups, supported by this research, include the importance of: well-considered and conscious handling through-out the whole product development process, structured and focused building of small featured products and out-of-the-box interpretation of collected data. Do not storm the castle with just an idea. Use the energy you save by not storming the castle to deepen out your assumptions until you found the assumption’s core, after which you walk the castle exactly aware of what you need to know. The provided insights on theoretical perspectives could be integrated in future research on the effectiveness of product development within startups.
1. Introduction
Entrepreneurship is considered a driver for economic growth (Wennekers, 1999), contributing to technological innovation, job creation and societal goals served (OECD, 2004). Entrepreneurship is stimulated by venture capitalists that back startup ventures with early stage money to further develop promising propositions. Interest in startups as a potential source of profit for investors is growing year-to-year (PWC Annual Venture Capital Report, 2012, 2013 and 2014). In the year 2015, startup ventures in the U.S. only have raised $58.8 billion (Veghte, 2016) from venture capitalist to invest in their economic activities. With the growing interest in startups the need for insight on why some startups become successful and others do not is more relevant than ever before.
Many authors have tried to answer the question what the factors are that make a startup become successful. Among those who tried are most recently for example Peter Thiel (former PayPal) in his bestseller Zero to one (2014) and Quinten Schevernels (former Layar) in his book Suits & hoodies (2015). Both come up with 7 essentials for successful startups.
Zero to one – Peter Thiel Suits & hoodies - Quinten Schevernels
1 Engineering (technology) Big dreams
2 Timing Innovation
3 Monopoly Product
4 People Culture
5 Distribution Talent
6 Durability Focus
7 The secret Funding
“The single most important thing for a startup is your product” (Peter Thiel)
TABLE 1: ESSENTIALS FOT SUCCESSFUL STARTUPS (THIEL, 2014; SCHEVERNELS, 2015)
The comparison of both lists of 7 essentials shows more differences than similarities, something that is very common in the relatively young literature about startups. About one aspect though, there is no doubt about its importance: the product/technology of the
startup. The product is the foundation and as a consequence undoubtedly an essential element to become successful as a startup.
Lately, a cyclical type of product development within startups has been evangelized: the lean startup (Ries, 2011). In this research a closer look at the effectiveness of product development processes within startups has been taken from the perspective of the lean startup methodology. More specific, the research focuses on the most effective balance between aspects of linear and cyclical entrepreneurial processes, to be explained below, and in which form variables within the product development process of lean startups – such as: the process, build, measure and learn – are effective.
1.1 Gap in the literature
Traditionally the entrepreneurial process has been seen as a predominantly linear process (Fisher, 2012). An entrepreneur identifies an opportunity worth exploiting, writes a business plan that he executes and, if successful, the startup creates returns for the venture. It is all about planning and progressive insights along the way. So, is it because of a lack of planning that the majority of startups, according to Forbes about 90%, fail (Patel, 2012)? Not according to entrepreneurs themselves; nearly half of the founders of a failed startup give “No Market Need” (42%) or “Ran Out of Cash” (29%) as the main argument for mortality (Griffith, 2014). Will there be a market and will I hit the market sweet spot in time, before I will run out of cash?
In the desire to cope with this uncertainty, an alternative way of running a startup has come into play: the lean startup (Ries, 2011). The lean startup is a customer-centered approach (Mueller, 2012; Blank, 2003) that applies cyclical and lean principles (Womack, 1990) within the startup environment. The cyclical customer-centered approach concerns the first uncertainty of “No Market Need”, the lean principles the other of “Ran Out of Cash”.
According to Ries (2011), an entrepreneur has nothing more than assumptions when starting a new business; a startup is an organization in search for a viable business model. Assumptions of the entrepreneur need to be tested through a repeated learning cycle – a build-measure-learn feedback loop – against the lowest possible costs. Through the learning cycle, the entrepreneur fine-tunes the business model by making adjustments,
called pivots or iterations, to his initial assumptions. This cyclical process goes on for as long as the startup has not found a viable business model.
There is, despite the hype (Lacy, 2013), more than four years after publication of Ries’ (2011) bestseller, to the best of our knowledge “no scientific evidence that the methodology of being lean in startups is being advantageous” (Heitmann, 2014, p. 1). At the same time there has been critique on the lean startup methodology. The essence of the critique has been neatly described as follows: “[The lean startup] is based on the concept of throwing shit at your users very early on and then iterating. The problem is that users have less and less patience now” (Newman, 2012). The critique is a consequence of bringing an inferior product to market in an early stage. One could say that customers have to “pay for the learning of the startup, while they probably are more interested in getting a reliable [product] that delivers value consistently” (Finneran, 2013). What is left, potentially, are unsatisfied customers when the startup is failing, while it may not be a surprise that satisfied customers are more likely to stay than unsatisfied ones (Mittal, 1998). Additionally, dissatisfied customers are likely to share their bad experience with others, which could ultimately result in the loss of even more potential customers (LeBoeuf, 2000). How does a lean startup makes sure customers stay satisfied during the trip of adjusting the assumptions and product? And there are more pitfalls, such as: retaining the balance between failure versus waste and the focus on experimenting, which should not become a goal on itself. No scientific research has been found addressing the specific topic of coping with these potential reverse sides of lean startups. Research in this field is complex because of the interaction between elements that influence effectiveness. Research asks for an in-depth look within, and focus only on, the product development process of young startup organizations in practice.
This thesis researches the effectiveness of the lean startup methodology in practice. The characteristic elements (variables) of the lean startup product development methodology will be reviewed, which variables will be tested in-depth on effectiveness in practice.
1.2 Research question
and the characterizing elements of the startups product development process. The core of the research discusses and investigates the effective application of characteristic product development elements at successful startups.
The research question is:
To which extent and in which form is cyclical entrepreneurship effective?
Effective entrepreneurship will be defined as the process that leads to the desired result of a viable business model with the least effort within the least time.
1.3 Sub questions
Within the literature on entrepreneurial product development there are two schools: the linear planning school and the cyclical learning school. Sub question 1:
• What is the difference between traditional linear entrepreneurship and cyclical learning entrepreneurship and what are the (dis)advantages of both?
The answer to this sub question describes the characteristics and (dis)advantages of both schools and links those characteristics to each other. This sub question addresses the to which extent part of the research question as it introduces the cyclical learning school as an alternative for the traditional linear planning school.
The lean startup methodology, an example of the cyclical learning school, is, as the cause for this research, chosen as the ground perspective of evaluation. Sub question 2:
• What are the characteristics of the lean startup methodology?
The answer to this sub question describes the main characteristics of the lean startup methodology, with special attention to the lean, specific cyclical and customer development aspects, and serves as an introduction to the review of the variables within the cyclical product development process.
As mentioned in this introduction, the effective execution of the product development process within lean startups has been ambiguous so far. Sub question 3:
• In case of cyclical entrepreneurship, how does a startup most effectively develop a product?
The answer to this sub question evaluates the characterizing elements of choice and effective execution of product development regarding the cyclical entrepreneurial process of the lean startup. This sub question addresses the in which form part of the research question.
2. Literature review
2.1 Entrepreneurial development process (linear versus cyclical)
As been stated in the introduction, uncertainty is a big factor entrepreneurs have to cope with. Regarding this uncertainty there are two options when starting a business: “engage in extensive business planning or just storm the castle by rallying resources together, orchestrate an immediate offering, and hustle for a first customer” (Brinckmann, 2010, p. 24). “Two prescriptions dominate the topic of what firms should do next in uncertain situations: planning approaches and adaptive [learning] approaches” (Witbank, 2006, p. 1). The first option is closely related to the traditional linear entrepreneurial process advocating planning (Ansoff, 1979; Porter, 1980), while the later is more related to the cyclical learning process of “getting out of the building” (Blank, 2003; Ries, 2011), and focuses on the learning aspect of developing a proposition or product (Mosakowski, 1997).
2.1.1 Linear entrepreneurship (planning school)
Traditionally the entrepreneurial process has been seen as a predominantly linear process (Fisher, 2012). In the linear view on entrepreneurship markets are rarely created, they already exist (Casson, 1982). The most widely recognized sources of inspiration for entrepreneurship are market inefficiencies (Kirzner, 1979) and technological progress (Schumpeter, 1934). The entrepreneur searches for an opportunity and evaluates whether it is worth exploiting. After deciding to exploit an opportunity, an entrepreneur takes action by seeking resources to establish an entity (the startup), which will develop and deliver a product or service (proposition) to make use of the identified opportunity. Adequate predictions are necessary for planning. Planning is key to become successful. If successful the startup create returns for the venture. This conventional product development process is a linear multi-phase process, which is analysis and prediction driven (Wiltbank, 2006). Key concepts of the planning school discussed in this literature review are: competitive analysis (Porter, 1980), planning (Ansoff, 1979) and market orientation (Hurley, 1998).
FIGURE 1: LINEAR ENTREPRENEURSHIP
Competitive analysis
From the planning school perspective the entrepreneurial process starts with the search for trends, seeking opportunities (Shane, 2000). According to Porter (1980) there are five forces that determine the competitive intensity and therefor the attractiveness of an industry. The five forces are: threat of new entry, threat of substitution, bargaining power of customers, bargaining power of suppliers and intensity of competitive rivalry (ibid.). The systematical, quite formal, analysis of industries leads an entrepreneur to the choice for an industry and opportunity that is worth exploiting (ibid.).
Planning of the process: predetermined path
After the choice for an industry the planning school focuses on integrative planning. Planning implies the specification of goals and fosters the identification of effective steps to achieve these goals (Brinckmann, 2010). The planning approach relies substantially on predictions (Witbank, 2006). Planning generally improves effectiveness of human action and facilitates goal achievement (Ansoff, 1979). In uncertain situations the benefits of planning increase, planning scholars argue, as it reduces the uncertainty of the process, facilitates faster decision-making and introduces controls for personal bias or subjectivity (Goll, 1997; Dean, 1996; Priem, 1995; Delmar, 2003). This finding suggests that rational planning can guide organizations to successful positions for the future, even in uncertain situations.
The process of product development follows more or less the same subsequent steps every time again. Key phases (specific names may vary among authors) of business planning are: definition of strategic goals, generation of alternatives to reach these goals, evaluation,
Competitive analysis Opportunity Planning Predictive rationality Predetermined Subsequent steps Goals
decision among alternatives and decision among implementation control (Brinckmann, 2010).
Market orientation
The product development process derived from planning is mainly but not only, organization internal oriented. Kohli and Jaworski (1990) define market orientation within the planning school as “the organization-wide generation of market intelligence that pertains to current and future customer needs, dissemination of intelligence across departments and organization-wide responsiveness” (Hurley, 1998, p. 43). Besides making plans, organizations stay in touch with the market and incrementally adapt the organizations plan to the changing market circumstances.
Evaluation of (dis)advantages
A big advantage of the linear entrepreneurial process is the certainty of the process through subsequent phases. Beforehand, entrepreneurs know, or at least think to know, that comprehensive analysis, secure planning and excellent execution will make their company outplay others. The way of execution of this linear process is the distinctive factor for success or failure. Besides the certainty of the process another advantage of the linear process is that in general planning improves productivity.
At the same time there are some pitfalls to the linear entrepreneurial process, like: limitations of predictive rationality, limited flexibility and limited customer involvement. While the described process of planning is predetermined, the future is not (at least for as far we humans know). In-depth analysis of the past and current situation, resulting in a plan for the future, always contains a certain asymmetry with the future as it rolls out. This predictive rationality comes with a limited flexibility as the process is linear planned through subsequent steps. The set process comes at a risk that in the end predictions turn out to be not matching customer needs, especially because customer involvement along the process is limited.
In The Four Steps To The Epiphany (Blank, 2003), the foundation of what would later result in The lean startup (Ries, 2011), 10 major flaws to using the traditional linear product development process within startups are summed up, see table.
The Four Steps To The Epiphany – Steve Blank 1 Where are the customers?
2 The focus on the first customer ship date
3 An emphasis on execution instead of learning and discovery
4 The lack of meaningful sales, marketing and business development milestones 5 The use of a product development methodology to measure sales
6 The use of a product development methodology to measure marketing 7 Premature scaling
8 Death spiral: the cost of getting product launch wrong 9 Not all startups are alike
10 Unrealistic expectations
TABLE 2: 10 WHAT IS WRONG WITH LINEAR PRODUCT DEVELOPMENT? (BLANK, 2003)
Wrap-up: linear entrepreneurial process
From the linear planning perspective, in-depth analysis of opportunities followed by the proper execution of predetermined planning based on predictions is the foundation for an effective and successful product development process. The traditional product oriented linear entrepreneurial approach recommends being systematic, focused on subsequent phases and formal (Brinckmann, 2010).
2.1.2 Cyclical entrepreneurship (learning school)
As opposed to the linear entrepreneurial process of planning, the cyclical entrepreneurial process of learning suggests organizations learn what to do next by minimizing the use of predictive rationality and, instead, experiment and move quickly to capture new opportunities (Mosakowski, 1997). According to the learning school the entrepreneurial process is not benefited by a linear process from “left to right” through subsequent steps, but profits from a more flexible and iterative development process containing repeated learning circles of development. The accompanying adaptive approaches avoid defining future event spaces, and instead position the firm for quick responses to uncertain and unpredictable events as they emerge (Quinn, 1980; Mintzberg, 1994). Adaptation research argues that in dynamic and uncertain situations planning slows adaptation and that comprehensive planning can actually blind the organization to important changes in its
environment (Mintzberg, 1990 and Schoemaker, 1993), while adaptive organizations are able to adapt fast to changing circumstances.
FIGURE 2: CYCLICAL ENTREPRENEURSHIP
Cyclical learning is relatively new to the entrepreneurial literature. Over time there have been several authors that have touched upon this topic using definitions like: probe and learn (Lynn, 1996), disciplined entrepreneurship (Sull, 2004), hypothesis-driven entrepreneurship (Eisenmann, 2013) and effectuation (Sarasvathy, 2005). In the next paragraphs these four mentioned forms of cyclical entrepreneurship are discussed in more detail.
Probe and learn
Lynn (1996) identifies, in line with the planning school, a focus of businesses on the importance of continuous improvements and innovation to maintain certain market positions. Discontinuous innovation is often neglected, while exactly it is this discontinuous innovation that is responsible for establishing certain competitive market positions. Besides the importance of discontinuous innovation, conventional market research proved to be of limited utility, because in hindsight, predictions were sometimes strikingly inaccurate.
As an alternative to the linear process, Lynn (1996) advocates a more cyclical process of probe and learn. The repeating cycle starts with probing by introducing an early, immature, version of the product to a plausible market. Key element of the cyclical process is to learn after introduction about the technology, about the market, which market segments are most receptive to the various features and whether and how the product can be scaled-up. The learning part of the cycle is an iterative process. “The firm enters an initial market with an
Experimental Assumptions Learning
Known uncertainty Adaptive (iterations)Repeated phases Business Model
early version of the product, learn from the experience, modify the product and marketing approach based on what they learned, and then try again” (Lynn, 1996, p. 19). Entrepreneurs should strive to maximize learning. Not blind trial-and-error learning, though a structured process of experimental design and exploration.
Disciplined entrepreneurship
Sull (2004) introduces a model, which can help entrepreneurs, structure their thinking about managing the uncertainty inherent in trying something new. The model follows the cyclical principles as advocated in the probe and learn theory, and tries to give more pointers for practical application of the learning school.
According to the disciplined entrepreneurship model, entrepreneurs should start by formulating a hypothesis. “A working hypothesis will contain implicit and explicit assumptions about multiple variables – including technology, customer demand, competitive response and the availability of resources – each of which is uncertain” (Sull, 2004, p. 72). With the hypothesis in hand, entrepreneurs should assemble the resources needed to conduct experiments. Think of cash, equipment, but also intangible assets such as a network and distribution capability. The goals of assembling resources is to run designed experiments, at some point entrepreneurs must take the plunge and test their plans in the world. Entrepreneurs should raise sufficient capital to finance the experiment they next envision, with a cushion built in for contingencies.
Hypothesis-driven entrepreneurship: the lean startup
Eisenmann (2013, p. 1) introduced the term hypothesis-driven entrepreneurship as an approach to “maximize, per unit of resource expended, the amount of information gained for resolving uncertainty”. An entrepreneur starts off with a vision translated into falsifiable business model hypotheses, and then test those hypotheses using a series of so called minimum viable products. Based on the feedback, an entrepreneur must decide whether to persevere with the proposed business model or pivot to a revised model. This process is repeated until all of the key business model hypotheses have been validated through testing.
The most well known form of hypothesis-driven entrepreneurship is the lean startup (Ries, 2011). The lean startup methodology will be discussed separately and in more detail in chapter 2.2.
Effectuation
Sarasvathy (2001) introduces the effectual cycle, which provides a way for entrepreneurs to control a future that is inherently unpredictable. Instead of focusing on goals, like linear entrepreneurship does, the entrepreneur exerts control over the available set of means, things over which the entrepreneur has control (ibid.). Effectuation “dictates that in highly uncertain and dynamic environments, target customers can only be defined ex post through whoever buys a product or service” (Fischer, 2012, p. 1024). The cycle of effectuation is build on five core principles: start with your means, focus on the downside risk, leverage contingencies, form partnerships and control above prediction.
The effectuation circle starts with the means, which is followed by interaction with other people based on the means with gathering stakeholder commitment as a result, which leads to new means, after which the cycle starts all over again.
Evaluation of (dis)advantages
A big advantage of the cyclical entrepreneurial process is the flexibility in order to cope with the uncertainty of not yet precisely knowing what customers are waiting for. Firstly, there is flexibility in the product development through cycles of adaptation and iteration. Secondly, due to the increase of involvement of customers in de development process, there is more certainty to develop a product the customer really wants.
Disadvantages of the cyclical entrepreneurial process is the uncertainty about how the product or proposition will develop over time and whether there is more to experiment or not (when are we finished?). There is no formal frame of reference to reflect what your doing is right. At the same time, customers are potentially being bothered with an inferior product, what could give them the feeling of paying for the learning of the startup.
Wrap-up: cyclical entrepreneurial process
From the cyclical learning perspective, experiments with customers based on assumptions followed by the iterations towards the product based on the (dis)validated feedback is the
foundation for an effective and successful product development process, as entrepreneurs should be known with uncertainty. The customer oriented cyclical entrepreneurial approach advocates to be adaptive and incremental within the product development process (Brinckmann, 2010)
2.1.3 Linear versus cyclical entrepreneurship
While the linear and cyclical product development approach both have “coping with uncertainty” as a starting point, the translation into a structured product development process is quite different. The linear process tries to solve the uncertainty issue by structured analysis and planning: what are we exactly going to do when. The cyclical process on the other hand embraces the existence of uncertainty by trying to structure the collection of data in order to shrink the uncertainty. In the following table the characteristics of both types of product development are compared.
Linear product development Cyclical product development
Starting point Opportunity Assumptions
Source for knowledge Analysis Experiments
Basic assumption Predictive rationality Known uncertainty Desired outcome Achieve goals, based on analysis Finding a business model Method of execution Planning of the process Learning through experiments Product development Predetermined Adaptive (iterations) and
incremental
Phases Subsequent phases Repeated phases
Focus Traditionally product oriented Customer oriented Advantages 1. Certain, predetermined, process
through subsequent phases; 2. In general, planning improves
productivity.
1. Flexibility in order to cope with uncertainty;
2. Customer involvement lowers chance of “No Market Need”. Disadvantages 1. Limitation of predictions;
2. Limited process flexibility, planning slows adaptation; 3. Limited customer involvement.
1. Uncertainty about product development
2. Uncertainty about progress
2.2 The lean startup methodology
As mentioned in the introduction of this research, lately the lean startup methodology (Ries, 2011) has become a hot topic in startup land. Every month, about 50,000 searches are undertaken in Google for the exact words “lean startup”. The lean startup focuses primarily on the product development aspect of startups. In this chapter of the literature review the characteristics of the lean startup methodology, a cyclical development process, will be discussed in more detail.
The lean startup has been evangelized by Eric Ries’ bestseller The lean startup (2011) and is a form of hypothesis-driven entrepreneurship, as discussed in the previous chapter. Ries defines a startup as an organization in search for a business model that is viable. Generally speaking, each startup starts off with only a set of assumptions. Assumptions about the business model need to be validated in goal-oriented experiments in practice (Blank, 2003). According to Ries, a lean startup should be in a constant loop of learning for as long as it hasn’t found a viable business model (“build-measure-learn feedback loop”). Due to early stage testing with and learning from (potential) customers the premise of the lean startup approach is to increase the probability of success by decreasing the chance of mortality due to the “No Market Need” argument, see the introduction of this thesis.
The lean startup is an example of cyclical product development. Besides the cyclical aspect, the lean startup is characterized by the incorporation of customer development and lean principles. Customers have a prominent role in the product development process, the “customer development” part, and the journey to a viable business model needs to be as efficient as possible, the “lean” part.
In this chapter the characteristics of the lean startup are discussed in more detail; the lean startup specific cyclical, customer development –and lean aspects will be discussed separately, to start with the lean characteristics.
2.2.1 Lean
Lean origins are from manufacturing. Manufacturing is traditionally characterized by mass production. Mass production asks for a stringent production process focused on high volumes and standardized products. Within the quite linear process of manufacturing, the
claim of lean manufacturing is to increase efficiency through minimizing waste, which results in higher value for the customer.
Manufacturing
The lean principles were born in Japan within the Toyota company in the 1940s (Melton, 2005). The term “lean” was coined to describe Toyota’s business of manufacturing by a research team headed by Jim Womack at MIT’s International Motor Vehicle Program during the late 1980s (Womack, 1990).
Mass production Lean production
Basis Henry Ford Toyota
People-design Narrowly skilled professionals Teams of multi-skilled workers at all levels in the organization People-production Unskilled or semi-skilled
workers
Teams of multi-skilled workers at all levels in the organization Equipment Expensive, single-purpose
machines
Manual automated systems which can produce large volumes with large product variety
Production methods
Make high volumes of standardized products
Make products which the customer had ordered Organizational
philosophy
Hierarchical – management take responsibility
Value streams using appropriate levels of empowerment – pushing responsibility further down the organization
Phylosophy Aim for ‘good enough’ Aim for perfection
TABLE 4: MASS PRODUCTION VERSUS LEAN PRODUCTION (MELTON, 2005)
Lean thinking does not offer a process that could be applied to solve a specific organizational problem. It rather offers a set of tools that can be applied case-by-case (Wessel, 2013). In order to remove waste from any process, lean thinking uses five core principles (Womack, 1990):
- Value: essentially is the capability provided by the customer at the right time at an appropriate price, as defined in each case by each customer (Bhasin, 2006)
- Value stream: is all the action needed to bring a product to the customer. It is necessary to eliminate steps that do not create value for the customers.
- Flow: is about what happens to the items that are processed and how the process works compared to what would be seen as a perfect process.
- Pull: is about providing what the customer wants only when he or she wants it. - Perfection: is never ending in the process of reducing time, space, cost and
mistakes.
Startup
The reduction of waste in the product development process is one of the key characteristics of the lean startup methodology. In order to avoid waste in the form of building unnecessary features, lean startups develop a Minimum Viable Product (MVP) tot start with. An MVP can be defined as a version of a new product which allows a team to collect the maximum amount of validated learning about customers with the least effort. The reduction of waste could be quantified in time, money and effort.
Startup versus manufacturing
Both the lean startup approach as lean manufacturing offer more of a framework, not a blueprint of steps to follow (Lacy, 2013). Both concepts are built on the same principle of reducing waste by giving the customer what he wants. In critical examination though, one could argue that the lean startup approach and lean manufacturing have opposite desired outcomes. As making mistakes, pivots and iterations, are common, even crucial, parts of the cyclical lean startup, these are the exact things lean manufacturing wants to eliminate – variation and discontinuity – in established, linear product chains (Heitmann, 2014).
Developing a startup proposition is also in another way different than developing a production line. The type of uncertainty differs. Where a startup is unsure about “what” the customer wants, for efficient manufacturing it is important “whether and when” the customer wants the product.
2.2.2 Cyclical product development process
Lean startups make use of a typical cyclical product development process as discussed in the previous chapter. The repeated phases are structured in a build-measure-learn feedback loop.
Build
A startup should start to produce a Minimum Viable Product based on their ideas and assumptions (build-phase). A Minimum Viable Product (MVP) means a version of the product that has just the characteristics to fully pass the feedback loop to learn, though is created with a minimum amount of effort and the least possible amount of development time, see the earlier mentioned ‘lean’ principles.
Measure
With the MVP, startups should “get out of the building” (Blank, 2003, p. 27) for customer development (measure-phase) and gather data on the hypotheses by the feedback of customers on the created product. Most probably the MVP fails to the expectations of both the customers and the startup.
Learn
After testing, a startup revises the assumptions based on the collected data (learn-phase) and makes pivots or iterations to the MVP based on the new idea(s) (build-phase). A pivot is a strategic change or adjustment designed to test a new fundamental hypothesis about the product. Other adjustments to the strategy, often smaller from nature, are called iterations. After pivoting or iterating the feedback loop starts all over again, until a product with a viable business model has been found.
FIGURE 3: BUILD -‐ MEASURE -‐ LEARN FEEDBACK LOOP
The build-measure-learn feedback loop comes to an end when a viable business model has been found. After finding a business model, the startup becomes a scale up and should focus on company creation and company building, see also the next paragraph about customer development.
2.2.3 Customer Development
The lean startup combines product development with customer development. Contrary to the more traditional approach, the focus within the process of developing is not only on the product but on the customer as well, see also the 10 flaws as described in paragraph 2.1.1.
Steve Blank described the process of customer understanding in a four-step framework about customer development, presented in his bestseller Four Steps to Epiphany (2003).
The first step is about customer discovery and involves discovering whether the problem, product and customer hypotheses from the business plan are correct. After this first round of testing hypotheses, in the second phase, about validation, a startup finds customers that are willing to pay for the product. This is the necessary checkpoint that a startup has found a viable business model, after which a startup won’t be a startup anymore (at least in the terms of Eric Ries). After this checkpoint a company scales execution in the third step of
Measure Product Build Ideas Learn Data
company creation and scales the organization and operations in the last step of company building.
Finding the right customers and markets is unpredictable. While people in the field might say they would love your product if you produce it (discovery phase), it could turn out they are not willing to pay (validation phase) when it is for real. Assumptions about product market fit could just be wrong due to misjudgement beforehand or unexpected complexity. That’s why between validation and discovery there is a way back via pivoting (see build-measure-learn feedback loop). This happens when it turns out that assumptions in the discovery phase do not lead to the necessary validation.
FIGURE 4: CUSTOMER DEVELOPMENT
2.2.4 Evaluation of the lean startup methodology
Key components of the lean startup are: lean principles, cyclical product development and customer development. At the beginning an entrepreneur starts with not validated assumptions. Via experimenting with a MVP through a repeating cycle of build-measure-learn, the startup makes iterations to the product until a viable business model has been found for the product.
The following table contains a comparison with other forms of cyclical entrepreneurship. Customer
Discovery ValidationCustomer CustomerCreation
Company Building
Search Execution
Lean startup Probe and learn Disciplined entrepreneurship Effectuation Starting point Not validated hypotheses Probing A working hypothesis Means Source for knowledge
Experiments Experiments Experiments Interaction
Basic assumption Known with uncertainty Known with uncertainty Known with uncertainty Known with uncertainty Desired outcome / goal orientation Finding a business model, with least effort (MVP) Finding a plausible market for the early, immature, version of the product Knowledge about product-market fit Maximum out of resources (control over available set of means Method of execution Lean experiments Probe: structured process of experimental design and exploration Designed experiments Stakeholder interaction: creating commitments and co-creation Product development Adaptive (iterations) Adaptive (iterations) Adaptive (iterations) Adaptive (iterations) Phases Buil-measure-learn Probe-learn Hypothesize-assemble resources-experiment Means-goals-stakeholder commitments
Focus Product and customer oriented Product and customer oriented Product and customer oriented Product and stakeholder oriented
TABLE 5: FORMS OF CYCLICAL ENTREPRENEURSHIP COMPARED TO LEAN STARTUP
Advantages of the lean startup are its reduction of waste, flexibility and connection with the customers. Disadvantage is the uncertainty about the product development itself and progress of the development process in general. Also, customers are confronted with inferior products that might not meet expectations of customers.
Resume: potential pitfalls of the lean startup methodology
Lean For manufacturing waste is reduced by minimizing failure, while for startups failing is essential to learn. How to balance between failure and reducing waste?
Lean For manufacturing the uncertainty is about “where and when” a customer wants the product, while for startups it is about “what” product the customer wants. Does this difference in goal leads to a conflicting situation in practice?
Cyclical Danger of keeping on experimenting, as the feedback loop could turn out to be a goal on itself in practice, and as a consequence not progressing. Customer
Development
The combination of validation and discovery: when is a business model validated for the build-measure-learn feedback loop?
Customer Development
Are early adopters paying for the learning of the startup?
2.3 Variables of cyclical product development of the lean startup
This research focuses on the extent and form in which the cyclical entrepreneurial process of the lean startup is most effective. There are several variables within the product development process of a lean startup and alternatives that should be borne in mind. Staying close to the core of the lean startup methodology and in line with the research question four categories of variables could be distinguished: process, build, measure, learn. This paragraph introduces and discusses those four categories, resulting in 7 variables for effectiveness.
Introduction of variables
The first category regarding the process contains two variables: type of entrepreneurial approach and structure of execution. These variables contribute to the “to which extent” part of the research, regarding the effectiveness of the cyclical entrepreneurial approach in comparison to a more linear approach, as discussed in paragraph 2.1. The second variable, structure of execution, sorts as a bridge towards the “in which form” part of the research, as basic presumptions on the elements within the process are discussed.
The other three categories assume cyclical elements of product development, as discussed in paragraph 2.2. The three steps of the product development process of the lean startup, build-measure-learn, are embraced as categories for variables. Build sees on the effective practices with the Minimum Viable Product. Measure splits up in two variables, effective methods for experiments and data collection. Learning splits also up in two variables, effectiveness of adjustments and the deepness of learning. These latest three categories are aligned with the “in which form” part of the research.
All mentioned categories and variables of the cyclical product development process of the lean startup are discussed in more detail in this paragraph.
Paragraph Category Variable
2.3.1 Process 1. Entrepreneurial approach
2. Structure of execution
2.3.2 Build 3. “Minimum Viable Product”
2.3.3 Measure 4. Experiments
5. Data collection
2.3.4 Learn 6. Adjustments
7. Deepness of learning
TABLE 7: OVERVIEW OF VARIABLES RELATED TO LEAN STARTUPS
2.3.1 Process
The answer to the question “to which extent” the cyclical entrepreneurial development process of the lean startup is effective, depends on a comparison with other types of entrepreneurial approaches. Does an entrepreneur enhance a linear, cyclical or lean approach and how effective is this? And is there a favorable structure of process elements with repeating phases and a set duration of cycles that is chosen while executing a cyclical approach? How does this compare to a more linear structured process? Questions that could be answered through the following two variables.
Entrepreneurial approach (variable 1)
In paragraph 2.1 the characteristics of and differences between a linear and cyclical product development process have been discussed. In paragraph 2.2 the same has been done with the lean startup approach in general. The advantages and disadvantages have been discussed. Amongst other characteristics, characterizing for the linear entrepreneurial process is analyzing and planning. Characterizing for the cyclical entrepreneurial process is experimenting and learning. Characterizing for the lean entrepreneurial process is reduction of waste along the process.
The variable derived from the comparison between linear, cyclical and lean development processes is the effectiveness of application of different types of entrepreneurship within the process of product development.
Structure of execution (2)
Structure is about the way the product development process works, about how the separate elements work together and about how they relate to each other.
The favored structure of phases from the lean startup methodology is the cyclical build-measure-learn feedback loop with repeating phases of building, measuring and learning, as used in this research to evaluate effectiveness. These cycles needs to be proceeded as fast as possible, over and over again. There might be alternatives though. Examples of alternative structures of the process could be derived from software development processes, like: waterfall development, agile extreme programming and agile scrum.
• Waterfall software development
The progress of waterfall software development traditionally flows from one phase to another in a downward fashion (Royce, 1987). The waterfall follows distinct steps, though short feedback loops to the previous step are allowed. This development model could be aligned with the linear planning school and is characterized with more or less the same (dis)advantages.
• Agile: extreme programming
Agile methodologies aim to avoid detailing and defining the entire project at the beginning, like it is the case with the waterfall development model, but instead to plan out and deliver small parts of the project at a time, incremental and iterative (see also 2.3.4). The extreme programming model consists of six sequential phases (linear) combined with feedback loops (cyclical) (Beck, 1999).
• Agile: scrum
The main idea of scrum is that software development involves several environmental and technical variables (e.g. requirements, time frames, resources, and technology) that are likely to change during the process (Schwaber, 1997; Takeuchi, 1986). The enhanced model for structuring the process should as a consequence be flexible and holistic. The scrum model consists of three linear repeating phases: pre-sprint planning, sprint, post sprint meeting. Scrum concentrates on how team members should function in order to produce the system flexibility in a constant changing environment.
Waterfall Agile: Extreme Programming
Agile: Scrum Lean Startup
Key points Linear development process through predetermined steps Customer driven development, small teams, daily builds Independent, small, self-organizing development teams, 30-day release cycles Customer driven development, against the lowest cost as possible Conceptual Incremental Incremental and
iterative Iterative and incremental Iterative Goal Effective software development process Effective software development process Effective software development process Effective entrepreneursh ip Development process
Predictable Predictable, but complex Unpredictable and complex Uncertain Duration of development Long, but certain in advance Fast development for uncertain period of time
Fast and certain (30-days) Fast, but uncertain Process Linear: sequential phases Linear and cyclical: sequential phases with feedback loops (basic set of deliverables- planning- implementing-testing-iterating) Cyclical and linear: returning sequential phases, set beforehand (pre-sprint planning-sprint-post-sprint meeting) Cyclical: returning sequential phases (build-measure-learn) Focus Progressive steps Small progressive steps Progress via sprints Lean learning Progress measure
Linear Velocity, number of implemented features Product backlog, list of features known to e built Validated learning
Advantages 1. Clear direction 2. Transparent progress Refactoring – the ongoing redesign of the system to improve its performance and responsiveness to change Enforce a paradigm shift from the “defined and repeatable” to the “new product development view of scrum” 1. Adaptive 2. Low cost Dis-advantages 1. Limited flexibility (no way back) While individual practices re suitable for many situations, overall view and management practices are given less attention While scrum details in specific how to manage the 30-day release cycle, the
integration and acceptance tests are not detailed
1. Uncertain progress 2. Learning at the cost of customers
TABLE 8: COMPARISON WATERFALL, AGILE AND LEAN STARTUP
Compared to the lean startup methodology the discussed software methods make use of a combination of cyclical and linear elements. This might meet with the objection (disadvantage) towards the cyclical product development of uncertainty about the progress. The duration of a cycle is the time which a full cycle of development takes to process, after which a new cycle starts. Is there a set duration for the completion of a full cycle? For the lean startup the cycle is characterized by being passed fast; the duration of a cycle is not necessarily set beforehand. On contrary, for the software development model agile scrum the duration of a cycle is set up front at 30 days.
The variable of process structure of cycles zooms in on effective elements of the structuring of the process and duration of cycles.
2.3.2 Build
Where does the entrepreneurial process start? Taking into consideration the more abstract choice has been made for a more cyclical approach like the lean startup methodology, the process starts with assumptions that are reflected in a Minimum Viable Product (MVP).
“Minimum Viable Product” (3)
The assumptions at the first start of the product development process could be a result of various sources, e.g.: experience, gut-feeling or knowledge in advance from potential customers. These first assumptions are remodeled during the following cycles of development, which is part of the product development cycle, see furthermore 2.3.3 “measure” and 2.3.4 “learn”.
Assumptions are translated into a first version of the product with the goal to collect data from customers in the measuring phase. Within the lean startup methodology this first version is called Minimum Viable Product (MVP), though it could appear in diverse typologies, like: early product, immature product or inferior product. What are the characteristics of the “MVP” or equivalent of this term?
The variables of the MVP are: assumptions, minimum, viable and the extend to which the product makes learning about the validity of the assumptions possible. How “minimum” should the viable product be? The entrepreneur should “remove any features, processes, or efforts that does not contribute directly to the learning you seek” (Ries, 2011). When is a minimum viable product “viable” (Kortman, 2012)? A minimum product is viable enough as it makes it possible to test your assumptions with customer feedback (Ries, 2011). Strictly taken, that is not the definition of viable, which is according tot the dictionary “capable of living” (Kortman, 2012). As a starting point of the build-measure-learn feedback loop the variables “viable” and “minimum” should be balanced in order to learn as much as possible against the lowest waste ratio.
FIGURE 5: MINIMUM VIABLE PRODUCT
Minimum Viable
Minimum + Viable
Good product for startups to build
Products built by companies that more mature Crappy products nobody wants to use
The variable of MVP consists of the balance between minimum, viable and learning opportunity.
2.3.3 Measure
With the chosen “abstract” structure of the process (2.3.1) and the more practical assumptions and “MVP” in hand (2.3.2) the question of execution comes into play. Most of the discussed types of cyclical entrepreneurship enhance experiments with customers as the appropriate method of gathering data to learn from. To measure is to know. The question that comes to mind is how to organize those experiments: evolving, planned, structured and/or lean? And are customers the only source for feedback? Effectuation, see paragraph 2.1.2, suggests another source for data gathering: interaction with stakeholders. Is this effective?
Experiments (4)
The running of experiments provides the fuel for learning. How could the process of experimenting be structured in a way that enables maximum learning? Should it be lean and quick (lean startup) in order to reduce waste and maybe gain time to run more essential experience without running out of cash, or carefully designed (disciplined entrepreneurship) in order to be certain there will be learned something?
Validated learning is extensively discussed in Running Lean (Maurya, 2012). Maurya (2012) defined a couple of ground rules for running effective experiments within startups systematically:
1. Maximize for Speed, Learning and Focus, see Figure 6; 2. Identify a Single Key Metric or Goal [to stay focused]; 3. Do the Smallest Thing Possible to Learn;
4. Formulate a Falsifiable Hypothesis;
5. Validate Qualitatively, Verify Quantitatively;
6. Make Sure You Can Correlate Results Back to Specific Actions; 7. Create Accessible Dashboards;
FIGURE 6: EFFECTIVENESS OF EXPERIMENTING (MAURYA, 2012)
The variable of experiments is characterized by the structure of the experiments.
Data collection (5)
Experiments as suggested by the lean startup focus merely on customer feedback as a source for data, as the goal is to not miss out on customer development. The lean startup focuses on quantitative data as the preferred type of data. The amount of data collected should not be more than necessary to learn from.
Source
Input from parties broader than customers could be an addition to the data collected by experimenting with customers. For example interaction with other stakeholders. Feedback from customers and interaction differ from each other in the way that experimenting is more or less one-way testing with (potential) customers and interaction is more or less a two-way conversation with stakeholders broader than customers only. Which stakeholders should be accredited with the most influence during the development process, e.g.: investors, employees or suppliers? And should commitments of the various stakeholders be valued differently?
Learning Focus
Running out of resources
Chasing your tail Premature optimization Speed The optimal learning loop
Type
Data can be collected in different ways resulting in different type of data-points and insights. Dependent on the assumptions and that derived MVP the type of feedback data could vary, among others, from: the number of sign-ups, conversion rates, willingness to pay and customer lifetime value. These examples are examples of quantitative data. The lean startup enhances a quantitative financial model of measuring, transparent via measurable outcomes and metrics (Ries, 2011). Besides quantitative feedback there is the option of collecting qualitative feedback. Qualitative feedback is based on verbal or comprehensive descriptions rather than numbers. Examples of qualitative feedback, amongst others, are: incoming e-mails at customer support, talks to (potential) customers and observations of the market.
Amount
Also the amount of data and involved parties necessary to learn, could vary. Think of two products with completely different customers groups like business-to-business and business-to-consumer. If the target-group varies, does the amount of needed feedback stays the same? At the same hand, is it desirable to open the startup’s MVP for everyone or limit the ones involved to a certain level? How do you know you gathered enough data to make the decision to persevere or adjust? Is there a set number of data-points or other rule to determine whether or not a startup measured enough to learn from?
The variable of data has three sub-levels: source, type and amount of data.
2.3.4 Learn
The foundation for learning is the data collected in the measuring part of the build-measure-learn feedback loop. Because of the collected data the basic assumptions most probably will change and with that the product should be adjusted. How do you do that in the most effective way? Variables within this category of variables are adjustments and the deepness of learning.
Adjustments (6)
The motivation and interpretation (source) for adjustments could vary. Is the main reason to help customers or the organization itself? When changing the product, there are various types of adjustments that could be made. The types of adjustments that could be
differentiated are: pivots or iterations versus incremental steps. Both differ in the approach and perspective on developing a product. Is there a limitation to the number of adjustments that could be made? How do customers deal with a changing product?
Source
An adjustment has an origin, a cause. From the customer development perspective of the lean startup the most obvious and effective source for adjustments are customer needs. The collected data leads to a change of basic assumption which should result in an adjustment towards the product. But is this always the most effective, or are there other sources of reasons for effective adjustments towards the product, e.g. when organizational costs are to high to maintain the operation of the current product.
Type
The lean startup (Ries, 2011) mentions several types of pivots: zoom-in-, zoom-out-, customer segment-, customer need-, platform-, business architecture-, value capture-, engine of growth-, channel- and technology pivot. Al these pivots, smaller pivots called iterations, have in common a change to the assumptions resulting in a changed, more mature, product. Iterative development steps are about planned rework. You create something, review it and than change it based on the feedback. Iterative development is closely related to cyclical entrepreneurship. Another way for making adjustments is via incremental steps. Incremental development steps are about planned stages of work. You create and schedule by small, subsequent and progressive steps. Incremental development is closely related to linear entrepreneurship.
FIGURE 7: INCREMENTAL VERSUS ITERATIVE (SOURCE: GOOGLE IMAGES)
Iterative (/pivots) Incremental
Conceptual Planned rework Scheduled work
Goal Correcting errors (change) Building blocks (improve) Process Cyclical: repeating stages Linear: stage after stage
Duration Short cycles Short increments
Focus On product On process
Strategic changes
Continuous changes, so flexible All changes are carried on during the process
Advantages 1. Adaptive
2. Improves product
1. Progressive 2. Improves process
Disadvantages 1. No certainty of process progress 1. Focus on process and progress, not on product-market fit
TABLE 9: ITERATIVE VERSUS INCREMENTAL ADJUSTMENTS
Number
Minimum viable products are often not deep, which kind of products are hard to love for customers (Sharkey, 2013). Inherent to a cyclical development process is the development through learning cycles from an inferior product to an increasingly mature product. This is most of the times in favor of the customers.
Incremental
Customers could also be dissatisfied though because of the process of iterating along the way: changes made to the first inferior product are made on an irregular, unexpected timeline and might not be in the desired way of every single existing customer, while as an existing customer you have invested in the relationship as well. The possible loss of customer satisfaction due to a number of iterations comes at a cost and as a result harms effectiveness. How to balance the scale and number of iterations and the interest of the existing customers? Is there a limitation towards the number of adjustments that could be made? And, is this necessary, or not a problem in practice?
The variable of adjustments has three sub-levels: source, type and number of adjustments.
Deepness of learning: single versus double loop (7)
This behavioral science model describes two ways in which we can learn from experience: single loop and double loop. Effective problem solving occurs to the extent individuals are aware of the major variables relevant to their problem and solve the problem in such a way that it remains solved at least until the considered variables change (Argyris, 1976).
Single loop learning is the most common used learning style for solving problems. Single loop learning involves connecting a strategy for action with a result. So, combining results and consequences (“what we obtain”) with action strategies and techniques (“what we do”). This results in answering the question how can we make our work more efficient.
Double loop learning looks at consequences from a wider perspective. Double loop learning involves re-evaluating and reframing our goals, values and beliefs. So, combining results and consequences (“what we obtain”) with governing variables (“why we do what we do”). This results in answering the question whether the way of working is the most effective way to reach our goals.
FIGURE 8: SINGLE -‐VERSUS DOUBLE LOOP LEARNING
The lean startup methodology is build on the principle that an entrepreneur starts with nothing more that assumptions. The build-measure-learn cycle should theoretically be applied as double loop learning.
The variable of single versus double loop learning is the deepness of the learning. Is there reflection-taking place of the underlying assumptions of the feedback of the customers? For example, does a startup listen to the feedback and act upon that specific feedback or does a startup try to dive deeper into the motives of the feedback of the customers in order to solve the real problem.
Techniques, tactics, tools (action strategy) Results (product) Why? How? Single loop: doing things right Double loop: doing the right things Ef�iciency Re�lection: underlying assumptions Effectivity
2.3.5 Overview of the variables of the cyclical entrepreneurial process
Variables Concepts
1 Entrepreneurial approach - Linear - Cyclical - Lean 2 Structure of execution - Elements 3 “Minimum Viable Product” - Assumptions
- Minimum - Viable
- Possibility to learn
4 Experiments - Process structure
5 Data collection - Source
- Type - Amount
6 Adjustments - Source
- Type - Number
7 Deepness of learning - Single vs. double loop learning
TABLE 10: OVERVIEW OF CONCEPTS RELATED TO VARIABLES
In total there are 7 variables with 16 concepts that will be tested in practice. To which extent and in which form is a combination of these variables effective?
