Business Model Robustness: A System Dynamics Approach

TITLE:

Business Model Robustness: A System Dynamics Approach

RUNNING TITLE:

Business Model Robustness

Nizar Abdelkafi Fraunhofer Center for Central and Eastern Europe Neumarkt 9-19, 04109 Leipzig, Germany Department of Innovation Management and Innovation Economics Grimmaische Straße 12, 04109 Leipzig, Germany [email protected]

Karl Täuscher* Fraunhofer Center for Central and Eastern Europe Neumarkt 9-19, 04109 Leipzig, Germany Department of Innovation Management and Innovation Economics Grimmaische Straße 12, 04109 Leipzig, Germany [email protected]

1 *Corresponding author

The paper has to be cited as follows: Täuscher, Karl / Abdelkafi, Nizar (2015): Business Model Robustness: A System Dynamics Approach. Proceedings of the 15th EURAM Conference, Warsaw, 17th 20th June 2015.

Business Model Robustness – A System Dynamics Approach Abstract: Effective business models enable firms to achieve their strategic objectives. Past experiences, however, have shown that many promising business models failed because they could not appropriately deal with unexpected conditions. Several business model scholars have therefore argued that effective business models should exhibit a certain level of robustness. Yet, there is little knowledge about the specific characteristics of robust business models. This work identifies the major characteristics of robust business models by drawing on a system dynamics perspective. A case example, the failed online grocer Webvan, illustrates our theoretical insights. Four factors that influence business model robustness are identified: (1) component dynamics, (2) tolerance to variation, (3) feedback about effectiveness, and (4) adaptability. Consequently, we propose four strategies for the design of robust business models and illustrate them with recent case examples of online grocery business models. Finally, we discuss how entrepreneurs can quickly check their business model’s robustness.

Key Words: Business models, Business Model Robustness, Business Model Assessment, Causal Loop Diagrams, System Dynamics, Online Grocery

2

1

Introduction

Over the last years, academics and practitioners have been increasingly interested in the characteristics of effective business model design. Several frameworks have emerged to describe the characteristics and critical success factors of effective business models (e.g., Zott and Amit, 2001; Morris et al., 2005; Bouwman et al., 2008; Casadesus-Masanell and Ricart, 2011). Yet, there is a bias towards growth and superior performance (e.g., Weill et al, 2005), whereas little attention has been dedicated to the vulnerability of business models and their inability to perform under certain conditions. Take the example of the low-cost airline business model. European low-cost carrier Ryanair has been studied as an example for effective business model design (e.g. Casadesus-Masanell and Ricart, 2007). The business model of Ryanair has been identified as successful since its components complement each other in a way that creates self-reinforcing feedback loops (Casadesus-Masanell and Ricart, 2007). Therefore, Ryanair has become a model for studying the design patterns and mechanics of successful business models (e.g., Gassmann et al., 2013). A different story has to be told about People Express, the first low-cost airline in the market. The business model of People Express allowed the company to rapidly grow to an important player in the market within only five years, building on strong reinforcing feedback loops (Morecroft, 2007). As opposed to Ryanair, People Express’ business model did not allow the carrier to deal with the demand fluctuations, leading to severe capacity planning problems. Instead of generating the lowest costs in the market, the company was confronted with a sharp decline in customer satisfaction, which led to a vicious cycle of demand (Hill and Jones, 2009; Senge, 1990). As a consequence, People Express’ dramatic growth was followed by an equally dramatic demise (Morecroft, 2008). Hence business models can only be 3

sustainably successful, if they not only allow the firm to grow, but are also robust to unexpected dynamics and conditions. Robustness has been proposed by different scholars as a universal criteria for business model effectiveness (Casadesus-Masanell and Ricart, 2007, 2010, 2011; Bouwman et al., 2008, 2012; Snihur and Zott, 2013). Robustness, as the “ability of the business model to sustain its effectiveness over time” (Casadesus-Masanell and Ricart, 2010, p. 148), has therefore been identified as an important phenomenon to study. However, only few scholars have approached it systematically. These approaches either focus on the business model’s robustness against competitive threats (e.g., Casadesus-Masanell and Ricart (2007), its ability to function despite changes in the environment (Bouwman et al., 2008), or its ability to be accepted by the market, while protected against imitation (Snihur and Zott, 2013). Consequently, Bouwman et al. (2012) conclude that “BM’s dynamics, and a prescriptive way of dealing with uncertainty and future robustness of BMs” are currently under-researched areas in the business model discussion. Thus, further research is required to reduce the research gap and most important to enable practitioners to design robust business models. The objective of this study is to advance our understanding of business model robustness to support business model design and analysis. The research aims at conceptualizing business model robustness to distinguish robust from non-robust business models. Therefore, the primary research question is: what constitutes a robust business model design? To approach this question, we consider a systems perspective and apply the principles and tools from ‘systems dynamics’. Prior research in the field of business models has proven the usefulness of system dynamics tools to the business model concept (Eurich et al., 2014). Recent literature has converged towards considering business models as systems (Zott and 4

Amit, 2013) that are inherently dynamic (Demil and Lecoque, 2010). Thus, system dynamics provides a promising perspective since it is concerned with analyzing dynamic systems. System dynamics allows one to make predictions about a system’s performance based on its structure and components. Consequently, the perspective can yield insights on how to recognize problematic design of dynamic systems and improve such design towards more robustness. Hence, the research further aims at answering the question: how can the principles and tools of system dynamics’ support the analysis of business model robustness? The paper is organized as follows. Section 2 reviews the extant literature on business model robustness. Section 3 proposes a conceptualization of business model robustness based on the system dynamics principles. Section 4 applies the conceptualization to the case of the failed business model of online grocery firm Webvan. Based on the analysis of the non-robust business model, it identifies four strategies of business model design to increase robustness. These strategies are illustrated by four alternative business models from the online grocery market. Section 5 then discusses the findings against literature and proposes questions to rapidly assess a business model’s robustness. Finally, section 6 summarizes the main insights of this work and provides directions for future research.

2

Literature review

The business model has been conceptualized, among others, as a system of activities and transactions (Zott and Amit, 2013), and as a system of choices and consequences (CasadesusMasanell and Ricart, 2010). The business model concept can be distinguished from other units of analysis – such as strategy – by its focus on “how the company communicates, creates, delivers, and captures value out of a value proposition” (Abdelkafi, 2012, p. 300). It has been 5

recognized as a complex construct since it integrates multiple dimensions (DaSilva & Trkman, 2013), spans across organizational boundaries (Amit and Zott, 2013), and is inherently dynamic (Casadesus-Masanell and Ricart, 2010). Thus, this research considers the business model as a dynamic system of interrelated components designed to create value for customers and partners and generate value to the firm. Business models innovation (BMI) refers to the process and outcome of searching new logics of the firm (Casadesu-Masanell and Zhu, 2012). BMI focuses on developing new value propositions for customers and partners, new ways to create, deliver and communicate value, and generate revenues (e.g., Magretta, 2002; Zott and Amit, 2007; Gambardella and McGahan, 2010; Markides, 2010; Teece, 2010; Abdelkafi et al., 2013). BMI literature has been concerned with developing, explaining and running a business (Spieth et al., 2014) and has recently developed several approaches to systematically design and evaluate business models (e.g., Schallmo, 2013; Bucherer, 2014; Eurich et al., 2014). Increasing focus has been devoted to designing artefacts as boundary objects for the business model ideation process (e.g., Eppler et al., 2011). However, little attention has been paid to evaluating different design options or comparing them with the current business model. Business model research has been concerned with the analysis of design characteristics and critical success factors of business models. Several scholars have proposed frameworks of successful business models (Afuah & Tucci, 2001, Pateli and Giaglis, 2004; Zott and Amit, 2007; Bouwman et al., 2008; Osterwalder and Pigneur, 2010; Weiner and Weisbecker, 2011; Casadesus-Masanell and Ricart, 2011). We follow the conceptualization of CasadesusMasanell and Ricart (2011), who argue that successful business models are (1) aligned with the strategic objectives of the firm, (2) self-reinforcing, and (3) robust. Building on the 6

conceptualization of these scholars, we consider robustness as the business model’s ability to remain effective over time despite uncertain variation of the business model components and the business model’s environment. While the criterion of self-reinforcement gains increasing attention (Täuscher and Abdelkafi, 2014), robustness has not been conceptualized sufficiently. Business model research has approached the concept of robustness from different theoretical perspectives. Casadesus-Masanell and Ricart (2011) have conceptualized robustness as the business model’s ability to fend off external threats from interactions with competitors and partners. Based on the work of Ghemawat (1991), they identify four main threats to robustness: imitation, hold-up, slack, and substitution (Casadesus-Masanell and Ricart, 2010). Snihur and Zott (2013) build on an institutional perspective and conceptualize robustness as the business model’s ability to provide a high familiarity to users and partners, while being sufficiently novel as a protection against imitation (‘legitimacy without Imitation’). The authors argue that firms can strategically design robust business models by increasing the legitimacy and barriers to imitation in the three business model levels of content, structure, and governance. Bouwman et al. (2008) approach the analysis of business model robustness as part of a business model design process. They define robustness as “the ability to cope with changes in the business environment […]. Typical examples of external influences are changes in user requirements, regulatory changes, emerging new target groups and changing scale of operation, the application of a different revenue model or the incorporation of a new technology” (p. 132). To assess the robustness, they make use of tools from future studies such as ‘what if-questions’ or scenario analysis. They demonstrate the usefulness of scenario analysis for the case of business models for mobile businesses in the field of digital television services over IP (IPTV). Overall, the scarce literature on business model robustness has not 7

converged towards a common understanding of the phenomenon and has subsequently not yet yielded a comprehensive perspective for designing robust business models. Every business model leads to certain risks for the firm. When innovating their business model, entrepreneurs and managers should therefore assess potential design alternatives also based the related risks (Chesbrough and Rosenbloom, 2002) and aim for business model design that yields advantageous risk structures to the firm (Teece, 2010). Existing business models can be improved by changing the components that create the most impactful risks to the business model (Girotra and Netessine 2011). A framework of business model risks has been proposed by Shi and Manning (2009). They propose a 3x3-matrix. On one axis the matrix distinguishes between the risks related to market size, market share, and the ability to defend the competitive advantage over time. The other axis refers to the related risk level: the individual business model components, compatibility between components, or the system the business model operates in. Girotra and Netessine (2011) argue that most risks can be assigned to one of two types of risks: information risks and incentive alignment risks. Information risks describe the risk that results from making decisions under uncertainty. Thus, a business model should aim at overcoming these risks by involving the necessary information early on. Incentive-alignment risks refer to a structural business model design that provides insufficient incentives for partners. The discussion section will show how our approach to business model robustness is compatible with earlier conceptualizations and can be combined with existing tools for assessing business model risks.

3 A system dynamics approach to business model robustness Existing approaches to business model robustness lack a dynamic view. System dynamics (SD) seems to be a promising methodology in this regard. In recent years, the principles and 8

tools of SD have been applied by a variety of business model researcher (Grasl, 2009; Casadesus-Masanell and Ricart, 2013; Schallmo, 2013; Bucherer, 2014; Eurich et al., 2013; Giesmann et al., 2014; Täuscher and Abdelkafi, 2014; Won et al., 2014). However, SD has not yet been used to analyze a business model robustness. 3.1 System dynamics System dynamics is a methodology for modelling and simulating dynamic systems (Sterman, 2002). The methodology extends beyond conventional systems perspectives by incorporating a system’s feedback loops and non-linear dynamics. The SD perspective builds on two major principles (Tang and Vijay, 2001). The first principle states that the system’s structure determines the system’s behavior. By understanding the system structure, it is therefore possible for researchers and practitioners to predict the business model’s dynamic behavior, since the “behavior of a system arises from its structure” (Sterman, 2000, p. 107). Consequently, SD is concerned with supporting the understanding of complex systems structures. SD approaches analyze how the elements of a system interrelate over time and how these dynamics shape the performance of the underlying system. The second principle of SD relates to human’s bounded rationality (Simon 1957). Empirical research in the field of SD has proven that humans mentally misrepresent the behavior of systems with non-linear relations. This limited cognition increases with complexity in the system. Simon illustrates bounded rationality with the metaphor of a pair of scissors, where one blade is the ‘cognitive limitations’ and the other the ‘structure of the environment’ (Tang and Vijay, 2001). The scissors metaphor points to the growing probability of cognitive errors in systems with higher complexity and dynamics. SD approaches aim to overcome these cognitive limitations through visualization and formulation of general cause-effect dynamics. As a consequence of 9

the combination of both principles, SD research is focused on enabling humans to understand a system’s structure based on the few key components that have the highest influence on the overall system. To facilitate the understanding of dynamic systems, SD scholars have developed a set of tools and instruments. This toolbox contains graphical notations and process models for qualitative modelling and for quantitative computer simulation. Business models are built on assumptions about how certain causes will lead to specific effects. Often, these assumptions are made implicitly by the entrepreneur or manager. Making these implicit mental models explicit can enhance the management team’s understanding of the structure and potential behavior. Explicit qualitative models thus reduce the risk of flawed business model design as a pure cause of cognitive limitations (as stated in principle 2; Pidd, 2003). Besides, qualitative modeling approaches provide a management team with a boundary artefact to collaboratively design, check, or improve a business model (Eppler et al., 2011). Recent SD research has stated that qualitative modelling approaches can convey most of the structure of the system and therefore provide value even without simulation (Scott et al., 2013). In this research, we apply causal loop diagrams (CLDs) for both their simple notation and its acceptance in business literature (Senge, 1990). Causal loop diagrams map the feedback loop structure of systems. A causal loop diagram is a simple map that is based on three elements: components, interactions between components, and feedback loops. Components can either be endogenous or exogenous to a system. Whereas the dynamics of endogenous components are explained by the model, exogenous components are independent from the model’s components. Interactions are marked by links from the influencing to the influenced component. A plus sign points to a positive influence, and a minus sign to a negative influence. Many complex systems are characterized by delays, 10

which have been identified as the major cause for problematic system behavior, as certain effects in a system are irreversible and not observed immediately, making any intervention to correct a system late and superfluous. Feedback constitutes one of the core concepts of SD. The potential of making better decisions by studying a system’s feedback loops is high, because humans generally tend to have flawed mental models of feedback systems (Sterman, 2002). Feedback loops exist, if components influence each other in a way that a change in one component will lead to a chain reaction of changes in other components that will ultimately influence the initial component. Since this interrelated chain of influencing components is circular, it has been conceptualized as a feedback loop. This dynamic either reinforces the direction of components in the feedback loop (self-reinforcing feedback loop) or will lead them to balance towards a goal value (balancing feedback loop). Building on SD principles, we can support the understanding of business model robustness by studying the business model’s components, interrelations, delays, and the resulting feedback loops. 3.2 Conceptualization of business model robustness The system dynamic principles can be applied to business models. We can build on two general sources of vulnerability: limited cognition of business model dynamics and inherent uncertainty of business model assumptions. Limited understanding of dynamic systems threatens the business model when it leads to cognitive errors about the business model structure (misconception of existing feedback loops) and how this structure will lead to a certain firm performance. Cognition errors are part of human thinking (Sterman, 2002). But, even a perfectly designed business model can fail due to the uncertainty of its components. The business model does not act in isolation, but depends on the dynamics of other systems such as customer and supplier markets. A business model designer has to make assumptions 11

on the factors affecting the success of the business model (e.g. market development, customer trends). Business model robustness therefore decreases, if the business model structure is intolerant to the uncertain dynamics related to its components. How well, for instance, does the business model maintain its effectiveness when management over-estimates the market development? Hence, a business model’s robustness increases with decreasing uncertainty about its components and their dynamics, and increasing tolerance to unpredicted component dynamics. Business model robustness can thus be analyzed based on four criteria: (1) level of uncertainty about key components, (2) tolerance to unpredicted component dynamics, (3) information feedback about the validity of the business model’s effectiveness; and (4) adaptability of the business model structure. The approach of causal loop diagramming supports a business model designer or analyst to qualitatively assess these criteria. The process of developing a CLD supports the modeler to identify the key components of the system; gain an understanding for the amount of variation that the system can absorb, while staying effective; recognize the important delays in the system; and identify components that produce path dependence. Therefore, causal loop modelling is a simple, yet powerful, tool to support the analysis of a business model’s robustness. In the next section, we develop a CLD for Webvan to analyze the business model’s robustness.

4

Application of the conceptualization

4.1

A causal loop diagram of Webvan

This section illustrates the applicability of the proposed conceptualization and the system dynamics tools by drawing on the case of Webvan, an online grocer which failed 12

spectacularly in the early 2000s. A failed business model can better illustrate the suitability of the robustness criteria proposed in the previous section. The case has been largely documented in the literature, allowing us to draw on a broad base of information (e.g., Afuah and Tucci, 2001; McAfee and Ashiya, 2006). Besides, the online grocery market is still highly interesting today. Online grocery, or e-grocery, describes the concept of ordering groceries over the internet. The market is considered to experience strongly growing demand in the next years (Esch and Klein, 2014). “If any industry is ripe for disruption by online shopping, it should be the grocery business” (Mitchell, 2014). Webvan was one of the first players in the market, creating high expectations among investors and consumers. After growing at high speed over some time, with investments into warehouses of over one billion US dollars, employing more than 2000 employees, and having raised more than 800 million US dollars from venture capital, the company crushed spectacularly (McDonald et al., 2014). The bankruptcy of Webvan is considered the biggest failure in the e-commerce history (Relan, 2013) and has frightened potential entrants into the markets for several years. Over the last years, however, several firms have entered the market with different business models (Mitchell, 2014). In this section, we illustrate the non-robust business model of Webvan and then discuss how the new entrants have learned from the Webvan experience, by focusing on how they increased the robustness level through different business model designs. Webvan started as an e-grocer with an innovative business model at that time. Founded in 1999, the company originally offered a home delivery service for groceries ordered on the internet, but soon expanded into other items such as apparel. The goal of Webvan was to become a single-source solution for the busy consumer today. As George Shaheen, former 13

CEO of Webvan, states the value proposition: “we (…) promise to give back your Saturday mornings” (The economist, 2000). To differentiate the value proposition from other grocers, Webvan promised the delivery of fresh products at competitive prices and a 30 minute delivery windows chosen by the customer. Webvan had raised an extraordinary amount of money during its short tenure as the largest and most promising e-grocer (Lunce et al., 2006) and had heavily invested into warehouses and logistics. During its short period of existence, Webvan successfully launched the company’s initial public offering (IPO), expanded into more than a dozen new markets, and added new products to the product line. Yet, in 2001 the firm filed for bankruptcy. The development of the causal loop diagram of Webvan follows the detailed guidelines of Sterman (2000, p. 137 - 190). The causal loop diagram is supposed to represent the core structure of Webvan’s business model as it was designed in 1999. We build the model on the management assumptions that have been referenced in the case studies through direct or indirect quotations. A first step towards developing a CLD is choosing the relevant components and model boundaries. Webvan’s business model builds on a number of components and can be considered from many perspectives. To select a set of key components, we use the approach of ‘reference modes’. ‘Reference modes’ refer to potential developments of the system and guide the modeler in choosing the model boundaries. In the case of Webvan, we focus on two problematic scenarios. In the first scenario, the company underestimates the amount of orders from its customers. The demand exceeds the firm’s capacity, which cannot fulfill the orders in the promised time frame (30 minute window selected by the customer), with the expected quality and quantity. To fulfill these orders, Webvan built up an infrastructure of warehouses, delivery vans, groceries, and service personnel. If the capacity is sufficient to fulfill the value 14

proposition, a high value to customers can be created. An increase in value to customers leads to an increase in orders, providing the company with cash that can be reinvested in its delivery system and product range. Thus, the first problematic scenario is a case in which the orders frequently exceed the capacity of Webvan to fulfill these orders. As a consequence, Webvan’s management could have been frightened by a rapidly falling demand due to unmet customer expectations. A second scenario refers to an overestimation of demand and consequently to overinvestments, especially in delivery capacity. Such a scenario would lead to a bottleneck of available cash due to a lack in customer orders, because of an unsuitable fixed cost structure (e.g. building of distribution centers). To represent these two anticipated problematic behaviors, we choose 10 components to represent the business model:          

Value creation capacity (Webvan’s infrastructure for creating and delivering the offered service) Cash (Webvan’s available financial means) Market expansion (Webvan’s investment in market development through advertising spending and construction of new distribution centers) Reputation (reputation of online grocery as a valuable alternative for conventional grocery shopping) Service quality (Webvan’s ability to fulfill the proposed service at any time) Customer expectations (Customer’s expectations about service quality) Value to customer (Satisfaction of Webvan’s customers) Orders (number of incoming orders to Webvan) Potential orders (number of overall orders in the areas in which Webvan has built a delivery infrastructure) Market demand (overall demand for online grocery services).

These components are interrelated to each other. In fact, every endogenous component is both a cause and an effect in the system. For instance, Webvan’s founder decided to highly invest in value creation capacity (distribution centers, local stations, vans, information technology, etc.) to enable a market leading service quality. The high service quality itself is the cause for assuming a higher value to its customers. The high value to its customers is assumed to lead 15

to a large number of orders. Management assumed that they could increase the e-grocery market demand with a superior value proposition compared to conventional grocery shopping. The management’s assumptions about these cause and effect relations are represented in figure 1. A link from one component to another represents the cause-effect-relationship between these two components.

Figure 1: Causal loop diagram of Webvan’s business model The CLD reveals several insights into the structure of Webvan’s business model. Webvan’s management described the advantage of their business model as being “highly cash generative” and therefore able to rapidly scale and expand (McAfee and Ashiya, 2006). To 16

represent this assumption, the cash component is placed in the center of the model. The management intended to rapidly scale the capacity in order to offer a superior value to customers, while achieving superior profit margins. At the same time, the firm intended to achieve economies of scale and scope by rapidly expanding into other markets, by expanding into new cities, entering into non-grocery categories, and heavily investing into marketing campaigns. The model represents the trade-off between investing the available cash either into the infrastructure for increasing the value creation capacity in existing markets or investing into market expansion. Both investment decisions are based on the assumptions of a reinforcing feedback loop. By increasing the value creation capacity, the firm can increase its service quality, which in turn increases the customer’s satisfaction (value to customer). An increase in the value to customer leads to more orders, leading to more cash. We call this feedback loop ‘Value to customer loop’. However, an increase in capacity does also increase the costs incurred by the firm. This has a negative influence on the available cash in every period (B2). We call this balancing loop ‘cost of value creation’. The second reinforcing feedback loop (R2) is called ‘market expansion loop’. Webvan’s business model is designed to rapidly increase the number of markets to generate economies of scales and scope (not explicitly in the model). More markets lead to more orders. Yet, the market expansion does not come for free. Investing in a new market induces costs that reduce the available cash. This balancing loop is called “costs of market expansion”. One additional reinforcing feedback loop exists between the two main reinforcing loops. By increasing the value to customers, Webvan expects to increase the overall market demand for e-grocery. This increases the number of potential orders the firm can expect when entering new markets. The loop is called ‘e-grocery reputation’ loop. A third balancing loop exists between the number of orders and Webvan’s service quality. A rising number of orders decrease the probability of being able to 17

perfectly deliver every order because of a lack in staff capacities or out-of-stock of certain products. Thus, we call the loop ‘too many orders to handle’. Overall, the CLD gives a visual overview of these dynamics that otherwise would have remained implicit. 4.2 Analyzing Webvan’s robustness The graphical representation of the elements now allows analyzing the robustness of Webvan’s business model. First, it focuses the analyst’s attention to a small set of components to assess their uncertainty. Several components in the business model have highly uncertain developments that are difficult to predict. The market demand is uncertain since the company operates in a newly established market that lacks reliable data about overall market demand and customer expectations. Case studies about Webvan show contrast some of the original assumptions about the component development with their actual dynamics and conclude that the business model was built on highly speculative assumptions about market demand, profitability, value to customers, or the ability to generate cash. The high uncertainty about key components therefore reduced the business model’s robustness. Next, we can assess the business model’s tolerance to component variations and unexpected dynamics. The business model is highly dependent on the continuous generation of cash by increasing its incoming orders, both in size and frequency. The generation of revenues depends on a large number of orders and Webvan’s ability to handle these orders efficiently. Thus, both the ‘market expansion’ loop and the ‘value to customer’ loop have to develop simultaneously. A rapidly growing demand, due to the market expansion loop, can lead to low service quality and low customer satisfaction (reference mode 1). Also, the 30 minute delivery window forced the firm to make accurate predictions about the timely order distribution. It turned out, that most orders were scheduled for the weekends and evenings, while the day 18

hours of the workweek produced large over-supply of capacity. After all, the business model structure had a low tolerance to unexpected development of components, which further reduced its robustness. Next, we assess how well the business model provides feedback about the validity of its assumptions. The CLD in figure 1 reveals the two most important delays in figure 1. Delays provide an important source of problematic system behavior (Sterman, 2000). Graphically highlighting such delays therefore allows the modeler to understand potential sources of problematic system behavior. The visual analysis reveals that the delays in the system effect both the ‘value to customer’ loop and the ‘market expansion’ loop. There exists a substantial delay between the firm’s investment in market expansion or capacity enlargement and the feedback about the effectiveness of these investments. In the case of the value creation capacity planning, the firm has to make its decision about the size of the infrastructure long before it receives feedback from customers about actual demand. The rapid expansion plan of Webvan did not allow the firm to include feedback from one market to another. As a consequence, the firm took a substantial risk of over- or under-estimating capacity requirements. Therefore, the business model’s structure prevented Webvan from directly incorporating feedback about misassumptions; this reduced the business models robustness substantially. Fourth, system dynamics analysis can reveal a system’s path dependence and inertia. Generally, path dependence results from components that accumulate and deplete over time (stocks). It is noteworthy that CLDs generally do not visually highlight components that build up over time, and those that are not. Such a distinction is made by the graphic notation of stock and flow diagrams, where stocks represent components with these attributes. However, we can use the principle of path dependence from stocks and flow models without visually 19

highlighting them in the model. In the business model of Webvan, three components fulfilled these criteria: (1) market demand, (2) cash, and (3) value creation capacity fulfill these criteria. First, the market demand component does not reduce the business model’s robustness significantly. This stock component could only backfire if the company would consider adapting the business model to offline groceries. Besides, the component is only partially shaped by Webvan and depends on external dynamics (not included in the CLD). Second, the cash component does provide the business model with vulnerability due to the uncertainty about the system to tolerate varying levels as discussed in factor 2. However, cash as a nonspecific asset does not limit the firm’s ability to change to another business model. Third, Webvan’s value creation capacity represents such a specific asset. Before starting their operations, Webvan already made the decision to build a large number of massive, highly automated warehouses with sophisticated inventory software. In July 1999, Webvan contracted with a construction to build up to 26 of these distribution centers for one billion US dollars (Hays et al., 2004). The high investment costs for the distribution centers, 35 million dollars each, tied the firm to this value creation decision. Thus, when the management realized that it over-estimated the demand dynamics, it was unable to adapt the value creation and delivery model or to change its expansion strategy. Webvan provides a good example of how a stock component leads to path dependence and therefore reduces the business model’s adaptability. In summary, the system dynamics-based approach to business model robustness demonstrates the factors that reduced the robustness of Webvan’s business model. The business model could have become highly successful, if the underlying assumptions had turned out perfectly correct. Yet, the business model did not tolerate much variation of its components, despite operating in a market environment with uncertain customer expectations and demand 20

development. The business model did not allow incorporating early feedback about the correctness of its assumptions. Moreover, it steered the company into a situation where it could have adapted its core logic of creating and delivering value only at immensely high costs. The combination of these four factors led to the spectacular demise of Webvan. 4.3 Strategies to design robust business models Building on the four factors that threaten a system’s robustness, we can formulate four strategies to increase robustness. First, the system can be stabilized by eliminating the components with the highest uncertainty. Second, a balancing loop can be added to increase the system’s tolerance to unexpected component variations. Third, the delays in the business model’s feedback loops should be reduced. Fourth, the structure of the business model should allow adaptability. In this section, we illustrate how different firms have used these strategies to improve the business model of Webvan. Eliminating components with high uncertainty A strategy consists in the elimination of vulnerable design elements. For instance, many successful business models have outsourced the risks of uncertain component development to their partners or customers. In the case of the e-grocery business model, market demand and consequently the number and amount of orders is highly uncertain. Every firm in this market has to deal with these uncertainties; however, they do not necessarily take responsibility for the uncertainty. Webvan’s business model forced the firm to invest large amounts of money into its capacity; thus, the firm had to anticipate the required capacity well in advance. The German startup ‘food.de’ has developed a business model that reduces the vulnerability from uncertain market demand. The firm offers a similar value proposition as Webvan (view the-hundert.com for detailed information). Yet, the company does not build on an own 21

warehouse infrastructure, but partners with large wholesales. Food.de utilizes the already existing infrastructure of a nationwide operating wholesaler as warehouses for the products it offers. Food.de therefore remains less vulnerable to changing customer preferences and demand fluctuations. The wholesaler bears most of the risk resulting from unexpected changes in customer preferences or changes in orders. They, not food.de, bear the risk of carrying groceries that might decay. Yet, the business model is attractive for its wholesale partner since it increases their revenues as well. Food.de still has to make assumptions about future orders to plan the delivery capacity for the last mile to the customer; however, the business model allowed food.de to rapidly expand its business from two test cities to 32 within 10 months. Adding a balancing feedback loop A second strategy to increase the business model’s robustness is to add a balancing feedback loop. A firm has to identify the most critical components and interactions in their business model. The Webvan model had a low tolerance to over- or underestimating the number and amount of orders. The overestimation led to idle capacity which decreased profit margins. A business model can be designed to reduce the threat from such an uncertainty by adding a balancing loop that stabilizes the uncertainty about required capacity. In fact, supermarkets that offer their groceries both online and offline build on such a balancing loop. These business models generally emerged from the traditional supermarket business model, but serve to illustrate the design logic behind such a business model. The model has been successfully applied by several established supermarket chains such as Tesco in the United Kingdom. Tesco offers its customers the possibility to either buy in their supermarket stores, or on their online platform Tesco.com. The advantage of the business model is its low 22

dependence on an unexpected development in either one of these channels. Generally, if the demand for online groceries increases, the demand for offline groceries decreases and vice versa. This is a classical balancing feedback loop. Adding this balancing feedback loop to the overall business model decreases the overall vulnerability to the system. As a consequence, the company is less likely to overstock or understock their groceries since it operates its online service out of its supermarket infrastructure. The warehouses of Tesco show less vulnerability to over- and understocking, enabling them to achieve higher margins. Since Tesco has already established warehouses all over the UK, it does not require large additional investment. The established logistics chain allows the company to achieve high efficiency in their delivery operations. Despite, the two value propositions enhance each other, resulting in more value to customers (Fehling, 2014, S. 8). Introducing a balancing feedback loop can therefore increase robustness if it hedges the uncertainty from its components. Reducing delays in feedback loops Every business model is a set of assumptions about how specific decisions will lead to certain consequences (Casadesus-Masanell and Ricart, 2010). These assumptions are made under uncertainty and thus lead to vulnerability of the business model. But, a business model can be adapted several times. The faster a business model designer receives feedback about false assumptions, the faster he can adapt the business model. Therefore, immediate feedback increases robustness. In the case of Webvan, two major delays existed in the business model: the investment decision in the value creation capacity was even before the company received feedback about capacity utilization in the first market. Secondly, Webvan rolled out the business model before validating the customer expectations about delivery time, price schemes, or service quality. Thus, a valid strategy to reduce these uncertainties is also an 23

approach that deliberately experiments with a business model until the required information feedbacks are available. The business model parameters should be set up in a way that managers can get feedback and are able rectify the business model, before it collapses. Obviously, the feedback that Webvan got was too late that managers did not have any other possibility to react than to close the business. Amazon has approached the field of online grocery delivery in such a feedback-focused approach. 1 In 2007, Amazon introduced AmazonFresh as a beta version to a selected number of customers in Seattle. But, unlike Webvan, Amazon expended slowly and carefully, to determine what works and what doesn’t before rolling out the business model to a large number of cities. In the seven years since introducing the service in Seattle, it has only entered few other markets, but at the same time experimented with some different business model designs. For instance, regarding the delivery system, the company has experimented with delivery to own retail stores, where customers could pick up their groceries themselves. After testing this option for several years, it decided to discontinue the pick-up service. Moreover, it tested and verified the value to customers of offering fresh products from several specialized partners such as restaurants or markets. Amazon has experimented with several delivery timing options, including next day delivery with a one hour delivery window, but has changed to a same-day delivery (MMR, 2014). The value communication changed from price and convenience, but changed their focus to a strong emphasis on convenience (“convenience, delivered”). Currently, Amazon is experimenting with a revenue mode building on subscription, charging $299 per year in return for free delivery, and tying it to Amazon Prime, its membership service which attracts loyal and wealthier customers (Mitchell, 2014).

1

Information about AmazonFresh are primarily based on McDonald et al. (2014). 24

Amazon could have entered the market rapidly given its large financial means. Yet, the company decided to first receive sufficient feedback from its first market before moving to the next. Thus, experimenting with different models allowed the firm to reduce the uncertainty of particular components and maintained the adaptability of the business model. The current business model therefore also reduces the variation of orders through a high entrance barrier, but higher incentives for regular orders. As a consequence, it produces a more predictable revenue stream, which makes the business model more robust. Reducing path dependence A fourth strategy aims at designing business models with higher adaptability. The business model of online grocery generally requires firms to build some delivery capacity. Even when firms do not operate their own warehouses or balance the online demand risk with offline grocery services, they remain vulnerable to uncertain demand since they have to provide a fleet of vehicles that can ensure last mile delivery to the customers at any time. Decisions about the fleet capacity can be taken faster than the decision of building new warehouses. Nevertheless, investments still have to be made in delivery vehicles and service personnel. Misassumptions about the utilization rate of the vehicles in the delivery service at a given day can lead to either lower value to customer or lower efficiency due to overstaffing. To cope with this problem, the German startup ALGEL (www.algel.de) has developed a business model based on ‘urban crowd based logistic services’. It offers a platform that matches online grocery customers with individuals who are willing to earn money by buying and delivering groceries for these customers. Every transaction between an online ordering customer and a delivery person generates a service fee for ALGEL. The incentive for the delivery persons consists in an order size-dependent fee from the online customer. So far, the company has 25

only started its operations in a limited geographic area. If the value proposition, the value creation and delivery, or the value capture logic do not provide the intended performance, the firm can simply change the business model without incurring large costs. It is to mention that firms with high path dependency are less likely to be imitated. The business model of ALGEL could be easily replicated. However, the business model design itself reduces the vulnerability to false assumptions by maintaining a high adaptability.

5

Discussion and Implications

The application of the system dynamics perspective and tools to the case of Webvan has provided generalizable insights into the factors that weaken business model robustness. This section discusses how the findings are compatible with the existing approaches presented in the literature review. Further, we propose a set of questions that can guide practitioners through a quick assessment of their business model robustness. The system dynamics perspective has yielded four factors of robust business model design. First, business models robustness is decreased when the development of components is highly uncertain. This factor relates to the element risks of Shi and Manning (2009) and the information risk of Girotra and Netessine (2011). Girotra and Netessine (2014) discuss several strategies how such information risks can be reduced. Firms should, for instance, outsource components with high uncertainty to partners that have a better information base. Outsourcing certain components to partners, however, introduces the firm to new risks (Euchner and Ganguly, 2014). Practitioners interested in assessing the vulnerability of their components should ask the following questions:   

How predictable are the key components of the business model? How much variation have the components shown in the past? How dependent is the business model to components outside the firm’s control? 26

The second factor for robustness stems from the business model’s tolerance to component variations. How vulnerable is the business model structure to unexpected component variations? Business model research has paid little attention to this question. Some approaches have been made to design tools that test a business model’s stability under different configuring of its components (Gordijn and Akkermans, 2001). Valuable insights for designing business models that are more tolerant to component variations can come from neighboring research fields such as supply chain management, business ecosystem research, systems engineering, or product design. In their seminal article on ‘robust quality’, Taguchi and Clausing (1990) argue that: “Design engineers take for granted environmental forces degrading performance […]. They try to counter these effects in product design—insulating wires, adjusting tire treads, sealing joints. But some performance degradations come from the interaction of parts themselves, not from anything external happening to them. In an ideal product—in an ideal anything—parts work in perfect harmony. Most real products, unfortunately, contain perturbations of one kind or another, usually the result of a faulty meshing of one component with corresponding components.”(p. 66) Business model researcher can build on build on approaches from these disciplines to advance the understanding of business model robustness. Practitioners can make use of the findings and tools from research on robust business ecosystems (Adner et al. 2013) on how to align components of a system in a way that reduces their vulnerability to variations. The authors point in particular to the vulnerability of relying on innovative developments outside the firm’s boundaries and develop strategies to overcome this risk for designing a robust ecosystem. Incorporating these findings, practitioners can assess their business model’s vulnerability to component variations by asking:   

How vulnerable is the business model to unexpected component dynamics? How vulnerable is the business model to misassumptions about component interrelations? How well does the business model deal with uncertainty about actions from its value partners? 27

The third factor for designing more robust business models relates to the business model’s ability to generate direct feedback about its effectiveness. The system dynamics perspective has pointed to the importance of recognizing delays in a dynamic system. System dynamics scholars have shown in several fields how delays lead to system instability and to flawed human decision making (Yasarcan and Barlas, 2005; Mosekilde and Laugesen, 2007). A famous exemplification for this phenomenon is incorporated by the so-called ‘beer game’ (Machuca and del Pozo Barajas, 1998). System dynamics has identified archetypical system structures based on feedback loops and delays. The literature on system archetypes (e.g., Senge, 1990; Braun, 2002; Wolstenholme, 2003) have identified how specific system structures lead to problematic system behavior and have proposed principles to resolve these problem-causing system structures. Studying the generic archetypes can improve the practitioner’s understanding of dynamic systems and thus reduce the risk of cognitive errors when designing business models. However, even perfectly designed business models have to build on uncertain assumptions. Thus, a business model becomes more robust if it enables the manager to receive rapid feedback on its effectiveness. In this regard, several business model scholars point to the importance of building business models that receive immediate feedback from customers and markets (Chesbrough, 2007). In industries with rapidly changing customer preferences, such as the fashion industry, reducing this information delay can generate an important competitive advantage. Girotra and Netessine (2011) describe how the fast fashion company ‘Zara’ has innovated the fashion industry’s business model by reducing the information delay about changing customer preferences by building a production and logistics infrastructure that strongly reduces the time between design decision, the feedback from customers in the store, and the firm’s ability to 28

react to this feedback. Thus, Zara’s business model design provides the firm with a competitive advantage since it receives faster feedback that leads to less vulnerability to wrong assumptions (either about structure or about component development). In this regard, analyzing business models from the system dynamics perspective can point to the major information delays in the system. Entrepreneurs and managers wishing to assess the feedback structure of their business model should ask:   

Does the business model generate direct feedback about the value to its customers? Does the business model generate direct feedback about its value creation and value delivery efficiency? Does the business model generate direct feedback about its ability to generate value to the firm (value capture)?

The system dynamics approach has emphasized the likelihood about making faulty assumptions about the structure or component development. Receiving direct feedback that verifies or falsifies these assumptions is only one side of the coin. The business model further has to be adaptable to new conditions. Business model research has argued for the importance of designing adaptable business models, since a “business model innovation is not a matter of superior foresight ex ante – rather, it requires significant trial and error, and quite a bit of adaptation ex post” (Chesbrough, 2010, p. 356). Adaptability can be built into the business model through “loosely fitting elements or introduction of new elements that change the dynamics among existing elements” (Morris et al., 2005, p. 732). Practitioners aiming for designing an adaptable business model can build on the method of discovery-driven business model design developed by McGrath (2010) or apply the tools from the lean startup literature (Ries, 2011; Blank, 2012). To assess the adaptability of their business model, practitioners can apply the scenario-analysis approach by Bouwman et al. (2008) aimed at stress-testing business model robustness under extreme scenarios. The authors develop possible scenarios 29

and assess whether the business model design allows effectively adapting to these changing environmental conditions. Generally, a business model designer should focus on three dimensions for assessing the adaptability of its business model:   

6

How easily can the value proposition be adapted? (Path dependence of image / reputation? Path dependence of customer base? How easily can the value creation and delivery logic be adapted? (Vendor lock-ins? Technological path dependence? Capacity How easily can the value capture logic be adapted? (fixed cost structure? compatibility of revenue sources? long-term pricing schemes?)

Conclusions and directions for further research

This research has analyzed business model robustness from a system dynamics perspective. Business model robustness is low if the business model builds on components that are highly uncertain; if it has a low tolerance to component variations, if the model generates feedback about its effectiveness only after substantial time delays and if the adaptability of the business model is low. We have shown four strategies to increase the robustness of the business model design based on principles and tools from system dynamics. First, a business model designer can increase robustness by eliminating components with high uncertainty. Second, she can design a balancing feedback loop that mitigates the vulnerability from unexpected component dynamics. Third, the designer can identify and reduce delays in the key feedback loops. Forth, she should identify and eliminate those components that provide the system with path dependence and inertia. The relevance of these theoretical criteria has been illustrated with examples from the online grocery market; however, further research should validate them with case studies from different industries. Besides, the next step would be the development of a universal measuring scale to assess a business model’s robustness.

30

The perspective and tools of system dynamics have proven useful for generating insights about business model robustness. The causal loop diagram approach enabled in a dynamic and systemic analysis of business model robustness. SD can provide further tools to analyze business models. In particular, SD provides guidelines and tools for computerized simulation. Computer simulation can enhance the qualitative analysis by stress-testing the business model under a large number of component variations and scenarios of external developments. Further research can therefore build on the qualitative findings and advance them through quantitative simulation models. This researched has focused on robustness from a design perspective. However, business models do not function in isolation (Casadesus-Masanell and Ricart, 2007). Having a robust business model design does not guarantee superior firm performance, since the business model is only one of many factors that will lead to the firm’s success. The success will depend on the firm’s competitive strategy to differentiate its business model from competitors or its operational efficiency to execute it efficiently, among many others. Therefore, we acknowledge that this research only considers the phenomenon from one possible perspective; combining this perspective of robust business model design with the approaches to analyzing robustness as the ability to fend off competitive threats such as imitation (Casadesus-Masanell and Ricart, 2010, 2011; Snihur and Zott, 2013) will further advance the understanding of business model robustness. It is important to recall that robustness is only one characteristic of successful business model design. Back in 1997, when Webvan received its first funding, a venture capitalist, talking to Webvan’s founder Louis Borders, claimed: “Louis, I think it is going to be a billion dollar

31

company.” Bolders responded: “Naw, it’s going to be $10 billion. Or zero.” 2 This statement underlines the fact, that a non-robust business model can promise superior firm performance, but comes at the risk of the firm’s complete demise. Especially for growth-oriented ventures in their early stages, robustness might be less important than self-reinforcement or scalability. Therefore, business model assessment should build both on criteria representing the opportunity and the risk side. Even more importantly, these two factors have to be balanced based on the firm’s current objectives (Casadesus-Masanell and Ricart, 2007). Future research could empirically analyze the relationship between these business model characteristics.

2

The dialogue has been reported in McAfee and Ayisha (2006, p. 17) based on Stross (2001). 32

References Abdelkafi, Nizar (2012): Open Business Models for the Greater Good – A Case Study from the Higher Education Context. In: Die Unternehmung 60 (3), pp. 279–316. Abdelkafi, Nizar; Täuscher, Karl (2014): Understanding the Basic Functioning of Business Models: A Graphical Approach for Theoretical Analysis and Tool Development. 25th ISPIM Conference. ISPIM, 08.06.2014. Adner, Ron; Oxley, Joanne E.; Silverman, Brian S. (Editors) (2013): Collaboration and Competition in Business Ecosystems. Advances in Strategic Management. Bingley: Emerald Group Publishing Limited. Afuah, Allan; Tucci, Christopher (2001): Internet business models and strategies. Text and cases. Boston: Irwin/McGraw-Hill. Algel UG: Crowdinvesting in Start-Ups. Online available at http://supercrowd.de/, last checked on 12.01.2015. Algel

UG:

Die

schönste

Art

zu

shoppen.

Online

available

at

http://www.algel.de/users/sign_in#firstBox, last checked on 12.01.2015. Amit, Raphael; Zott, Christoph (2001): Value Creation in e-Business. In: Strategic Management Journal 22, pp. 493–520. DOI: 10.1002/smj.187. Barnett, William; Bose, Katherine (2007): Webvan (B). Hg. v. Stanford Business School. Stanford, California. Barnett, William; Sigg, Keith (2007): WEBVAN. Hg. v. Stanford Business School. Stanford, California.

33

Blank, S. (2013): Why the Lean Start-Up Changes Everything. Harvard Business Review 91(5): 63-72. Bouwman, Harry; et al. (2012): Business Models. Tooling and a Research Agenda. 25th Bled eConference. University of Maribour. Bled, Slovenia, 17.06.2012. Bouwman, Harry; Faber, E.; Haaker, Timber; Feenstra, R. (2008): What’s Next? Some Thoughts and a Research Agenda. In: Harry Bouwman, Henny de Vos und Timber Haaker (Hg.): Mobile service innovation and business models. Berlin: Springer, pp. 137–150. Bouwman, Harry; Vos, Henny de; Haaker, Timber (Hg.) (2008): Mobile service innovation and business models. Berlin: Springer. Bouwman, Harry; Zhengjia, M.; van der Duin, P.; Limonard, pp. (2008): Robustness of IPTV Business Models. In: Harry Bouwman, Henny de Vos und Timber Haaker (Hg.): Mobile service innovation and business models. Berlin: Springer, pp. 205–216. Bucherer, Eva (2011): Business model innovation - guidelines for a structured approach. Dissertation. Aachen: Shaker. Carlos M. DaSilva; Peter Trkman (2013): Business Model: What It Is and What It Is Not. In: Long Range Planning (0), pp. 1-11. DOI: 10.1016/j.lrp.2013.08.004. Casadesus‐Masanell, Ramon; Ricart, Joan E. (2007): Competing through business models. IESE Business School (Working Paper, 713). Casadesus‐Masanell, Ramon; Ricart, Joan E. (2010): Competitiveness: business model reconfiguration for innovation and internationalization. In: Mgt Res: J of Iberoamer Ac Mgt 8 (2), pp. 123–149. DOI: 10.1108/1536-541011066470. Casadesus‐Masanell, Ramon; Ricart, Joan E. (2010): From Strategy to Business Models and onto Tactics. In: Long Range Planning 43, pp. 195–215. 34

Casadesus‐Masanell, Ramon; Ricart, Joan E. (2011): How to design a winning business model. In: Harvard Business Review 89 (1/2), pp. 100–107. Casadesus‐Masanell, Ramon; Zhu, Feng (2013): Business model innovation and competitive imitation: The case of sponsor-based business models. In: Strat. Mgmt. J. 34 (4), pp. 464–482. DOI: 10.1002/smj.2022. Demil, Benoît; Lecocq, Xavier (2010): Business model evolution. In search of dynamic consistency. In: Long Range Planning 43 (2/3), pp. 227–246. Eppler, Martin; Hoffmann, Friederike; Bresciani, Sabrina (2011): New Business Models through Collaborative Idea Generation. In: Int. J. Innov. Mgt. 15 (06), pp. 1323–1341. DOI: 10.1142/S1363919611003751. Esch, Franz-Rudolf; Klein, Jan (2014): Online und Offline harmonisch verzahnen: Lebensmittel per Click — Neue Kontaktpunkte für die Händlermarke nutzen. In: Lebensmittelzeitung, 27.06.2014 (26), pp. 42. Eurich, M.; Weiblen, T.; Breitenmoser, P. (2014): A six-step approach to business model innovation. In: International Journal of Entrepreneurship and Innovation Management 18 (4), pp. 330–348. DOI: 10.1504/IJEIM.2014.064213. Fehling, Jonas (2014): Im Check: Welcher ist der beste Online-Supermarkt? Available online at

http://www.focus.de/finanzen/news/rewe-tengelmann-edeka-so-gut-sind-dielieferdienste-

aus-dem-supermarkt_id_3674905.html last viewed. 09.04.2014). Gambardella, Alfonso; McGahan, Anita M. (2010): Business-Model Innovation: General Purpose Technologies and their Ghemawat, Pankaj (1991): Commitment. The dynamic of strategy. New York, Toronto, New York: Free Press; Maxwell Macmillan Canada; Maxwell Macmillan International. 35

Girotra, Karan; Netessine, Serguei (2011): How to Build Risk into Your Business Model. In: Harvard Business Review 89 (5), pp. 100–105. Girotra, Karan; Netessine, Serguei (2014): Four Paths to Business Model Innovation. In: Harvard Business Review (July/August). Grasl, Oliver (2009): Professional Service Firms: Business Model Analysis - Method and Case Studies. Dissertation. Universität St.Gallen, St. Gallen, Switzerland. Online available at http://www1.unisg.ch/www/edis.nsf/SysLkpByIdentifier/3686/$FILE/dis3686.pdf. Hays, T.; Keskinocak, P.; Lopez, V.: Strategies and Challenges of Internet Grocery Retailing Logistics. In: J. Geunes, E. Akçali, P. M. Pardalos, H. E. Romeijn und Z.-J. Shen (Hg.): Applications of Supply Chain Management and E-Commerce Research in Industry, Bd. 92 (Applied Optimization, 92). Hill, Charles W. L; Jones, Gareth R. (2010): Strategic management theory. An integrated approach. 9th ed. Mason, OH: South-Western/Cengage Learning. Johnson, Michael P.; Norman, Bryan; Secomandi, Nicola: Tutorials in operations research. Models, methods, and applications for innovative decision making. Institute for Operations Research and the Management Sciences. Hanover, Md. (TutORials). Johnson, Pontus et. al. (2013): Business Model Risk Analysis: Predicting the Probability of Business Network Profitability. In: Wil van der Aalst, John Mylopoulos, Michael Rosemann, Shaw, Michael J.; Clemens Szyperski, Marten van Sinderen et al. (Hg.): Enterprise Interoperability, Bd. 144. 5th International IFIP Working Conference. Enschede, Netherlands, 27.03.2013. Berlin, Heidelberg: Springer Berlin Heidelberg (144), pp. 118–130. Lee, H. (2003): The Triple-A Supply Chain. In: Harvard Business Review 82 (10), pp. 102– 112. 36

Lunce, S. E.; Lunce, L. M.; Kawai, Y.; Maniam, B. (2006): Success and failure of pure-play organizations: Webvan versus Peapod, a comparative analysis. In: Industrial Management & Data Systems 106 (9), pp. 1344–1358. Magretta, Joan (2002): Why Business Models Matter. In: Harvard Business Review 80 (5), pp. 86–92. Markides, Costas (2010): Creativity is not enough: How to create new market space through business model innovation. In: IPM Perspectives 2, pp. 73–81. MMR: Plastic Bag Prohibition Is Extended (2014). In: MMR 31, 18.08.2014 (13), pp. 56. McAfee, Andrew; Ashiya, Mona (2006): Webvan. Hg. v. Harvard Business Publishing. Boston. McDonald, Rory; Christensen, Clayton; Yang, Robin and Ty Hollingsworth (2014): AmazonFresh: Rekindling the Online Grocery Market. Harvard Business School Case 615013, July 2014. Revised August 2014. Mitchell, Dan (2014): Next up for disruption: The grocery business. In: Fortune, 04.04.2014. Online

available

at

http://fortune.com/2014/04/04/next-up-for-disruption-the-grocery-

business/. Morecroft, John (2007): Strategic Modelling and Business Dynamics: A Feedback Systems Approach. Hoboken, NJ: John Wiley & Sons. Pateli, Adamantia G.; Giaglis, George M. (2004): A research framework for analysing eBusiness models. In: European Journal of Information Systems 13 (4), pp. 302–314. DOI: 10.1057/palgrave.ejis.3000513. Pidd, Michael (2003): Tools for thinking. Modelling in management science. 2nd ed. Chichester, England, Hoboken, NJ: Wiley. 37

Pillai, P.; Dutta, S. (2004): Webvan: A Disaster on the Web. Hg. v. ISB Center for Management Research. Indian School of Business. Hyderabad, India. Relan, Peter (2013): Where Webvan Failed And How Home Delivery 2.0 Could Succeed. In: TechCrunch, 27.09.2013. Online available at http://techcrunch.com/2013/09/27/why-webvanfailed-and-how-home-delivery-2-0-is-addressing-the-problems/, last checked on 12.01.2015. Ries, Eric (2011): The Lean Startup: How Today's Entrepreneurs Use Continuous Innovation to Create Radically Successful Businesses. Crown Publishing. p. 103. ISBN 978-0-30788791-7. Stross, Randall (2001): Only a bold gamble can save Webvan now. The Wall Street Jounal, February 2, 2001. Rudarakanchana, Nat (2013): AmazonFresh’s (AMZN) Sneaky Business Strategy – Compete with Premium Grocers, Build Out Distribution Model, Enter Mass Market With Force. In: International

Business

Times,

01.10.2013.

Online

available

at

http://www.ibtimes.com/amazonfreshs-amzn-sneaky-business-strategy-compete-premiumgrocers-build-out-distribution-model. Schallmo, Daniel (2013): Geschäftsmodell-Innovation. Grundlagen, bestehende Ansätze, methodisches Vorgehen und B2B-Geschäftsmodelle. Wiesbaden: Springer Fachmedien Wiesbaden; Imprint: Springer Gabler (SpringerLink : Bücher). Scott, Rodney J.; Cavana, Robert Y.; Cameron, Donald (2013): Evaluating immediate and long-term impacts of qualitative group model building workshops on participants' mental models. In: Syst. Dyn. Rev. 29 (4), pp. 216–236. DOI: 10.1002/sdr.1505. Senge, Peter (1990): The fifth discipline. Mastering the five practices of the learning organization. 1st ed. New York: Doubleday/Currency. 38

Shi, Yuwei; Manning, Tom (2009): Understanding business models and business model risks. In: The journal of private equity : JPE 12 (2), pp. 49–59. Simon, Herbert (1957): Models of Man. Social and Rational: Wiley. Snihur, Yuliya; Zott, Christoph (2013): Legitimacy without Imitation: How to Achieve Robust Business Model Innovation. 35th DRUID Celebration Conference. The DRUID Academy. Barcelona, Spain, 17.06.2013. Spieth, P.; Schneckenberg, D.; Ricart, J. E. (2014): Business Model Innovation - State of the Art and Future Challenges for the Field. In: R&D Management 44 (3), pp. 237–247. Stanoevska-Slabeva, Katarina; Giessmann, Andrea; Kyas, Philipp; Tyrväinen, Pasi (2014): Towards a better Understanding of the Dynamics of Platform as a Service Business Models. Hawaii International Conference on System Sciences. University of Hawai'i at Manoa. Hawaii, 06.01.2014. Sterman, John (2000): Business dynamics. Systems thinking and modeling for a complex world. Boston, MA: Irwin/McGraw-Hill. Sterman, John (2002): System Dynamics: System Thinking and Modelling for a Complex World. Working Paper. Massachusetts Institute of Technology, Cambridge, Mass. Engineering Systems Division. Taguchi, Genichi and Clausing, Don (1990): Robust Quality. Harvard Business Review (1), pp. 65-75. Tang, Victor; Vijay, Samudra (2001): System Dynamics. Origins, development, and future prospects of a method. Massachusetts Institute of Technology, Cambridge, Mass. Research Seminar in Engineering Systems.

39

Teece, David (2010): Business Models, Business Strategy and Innovation. In: Long Range Planning 43, pp. 172–194. The economist: Shopping around the web: A survey of e-commerce. In: The Economist 354, pp. 5–54. Online available at Shopping around the web: A survey of e-commerce, last checked on 12.01.2015. The Hundert: food.de. Online available at http://the-hundert.com/de/startups/food-de/, last checked on 12.01.2015. Wolstenholme, E. F. (2003): Towards the definition and use of a core set of archetypal structures in system dynamics. In: Syst. Dyn. Rev. 19 (1), pp. 7–26. DOI: 10.1002/sdr.259. Won, Jong Hun; et al. (2014): Association of Product-Service Systems Design Concepts with Business Models and their Evaluation Method. International Conference on Engineering, Technology & Innovation. German University in Cairo. Cairo, Egypt, 19.04.2014. Zott, Christoph; Amit, Raffael (2007): Business model design and the performance of entrepreneurial

firms.

In:

Organization

Science

18

(2),

pp.

181–199.

DOI:

10.1287/orsc.1060.0232. Zott, Christoph; Amit, Raffael (2007): Business Model Design and the Performance of Entrepreneurial Firms. In: Organization Science 18 (2), pp. 181–199. Zott, Christoph; Amit, Raffael (2013): The business model: A theoretically anchored robust construct for strategic analysis. In: Strategic Organization 11 (4), pp. 403–411. Online available at http://www.scopus.com/inward/record.url?eid=2-s2.084887445607&partnerID=40&md5=1749066ed040118316c009c3ff8e05c7.

40