2 Three questions

RESEARCH PAPER

System dynamics for corporate business model innovation

Thomas Moellers 1 & Lars von der Burg 2 & Bastian Bansemir 3 & Max Pretzl 3 & Oliver Gassmann 1

Received: 15 September 2017 / Accepted: 3 January 2019

# The Author(s) 2019

Abstract

Business model innovation is a process that allows firms to build and maintain competitive advantages. However, it imposes

major challenges to managers who rely on incomplete cognitive representations while attempting to understand the environ-

mental dynamics that determine a business model’s prospective performance. System Dynamics is a computational approach

potentially useful in enhancing managerial understanding and decision-making during business model innovation, yet its effects

lack sound empirical evidence. This study of five cases inside BMW assesses the usefulness of System Dynamics along the

different phases of business model innovation processes. In order to develop a nuanced understanding we triangulate insights

from the multiple-case study approach with results from a Q-Sort exercise. Our emergent theory highlights that System Dynamics

enables managers to develop more accurate cognitive representations about their business models. Unexpectedly, we find that

this process leads to a cognition gap apparent in the communication with managers not involved in the modelling process. We

observe two strategies to overcome this gap. A second key insight is that System Dynamics has a tendency to consolidate mental

models by different managers that need to be managed cautiously. We develop a set of 11 propositions that represents the core of

our theoretical insights.

Keywords System dynamics . Corporate business model innovation . Managerial cognition

Introduction

Business model innovation (BMI) is a valuable mean to create

and maintain superior firm performance (Aspara et al. 2010;

Cucculelli and Bettinelli 2015; Desyllas and Sako 2013; Kim

and Min 2015). For instance, changing its business model

from selling tools, spare parts, and maintenance services

directly to a fleet management model, is widely considered

as the main driver of Hilti’s competitive success over recent

years (Casadesus-Masanell et al. 2017a). Not surprisingly, the

majority of top managers prioritizes or engages actively in

BMI (IBM 2006; Matzler et al. 2013).

However, BMI is complex and poses a variety of chal-

lenges for decision makers (Chesbrough 2010; Lucas and

Goh 2009; Teece 2010; Tripsas and Gavetti 2000).

Uncertainty concerning the viability of a new business model

and the dynamics arising from complex interactions between

its components cause significant difficulties to successful in-

novation. BMI in established firms requires additional consid-

erations as a potential business models may create or hinder

synergistic effects within a portfolio of multiple business

models (Aversa et al. 2017; Rumble and Mangematin 2015;

Sabatier et al. 2010). Even if not, the innovation of an

established business model is most often facing significant

inertia originating from existing resources or past success

(Chesbrough and Rosenbloom 2002; Leonard-Barton 1992;

Tripsas and Gavetti 2000). For instance, the introduction of

Hilti’s fleet management model was not an obvious success

story but required its top management to reflect on the poten-

tial consequences across a diverse set of issues ranging from

the financing of assets previously part of its customers’

This article is part of the Topical Collection on Design Science Research

in the Networked Economy

Responsible Editor: Alexander Mädche

* Thomas Moellers

[email protected]

* Bastian Bansemir

[email protected]

1 Institute of Technology Management, University of St. Gallen,

Dufourstrasse 40a, 9000 St. Gallen, Switzerland

2 BCG Switzerland,

Münstergasse 2, 8001 Zurich, Switzerland

3 BMW Group, Financial Services,

Heidemannstraße 164, 80939 Munich, Germany

https://doi.org/10.1007/s12525-019-00329-y

Electronic Markets (2019) 29:387–406

/ Published online: 26 February 2019

balance sheets, to a necessary restructuring of its sales force

(Casadesus-Masanell et al. 2017b). Whilst Hilti has been able

to successfully innovate its business model, scholars have

provided rich insight on firms whose top managers failed to

evaluate the complex and dynamic consequences triggered by

(potential) changes (cfr. Chesbrough and Rosenbloom 2002;

Lucas and Goh 2009).

Theory on managerial cognition assumes managers to

spend their time absorbing, processing, and disseminating in-

formation about their environment. It offers valuable insights

why they face cognitive challenges when exposed to contexts

characterized by high levels of complexity and dynamics

(McCall and Kaplan 1985; Prahalad and Bettis 1986; Simon

1957; Walsh 1995; Weick 1979). The central argument is that

the cognitive limitations of the human brain only allow for

simplified representations of reality (Gavetti 2012; Gavetti

and Levinthal 2000; Martignoni et al. 2016; Nadkarni and

Barr 2008) and poorly understood dynamic behaviour

(Forrester 1961; Sterman 2000; Stern and Deimler 2006)

causing misjudgements in decision-making. When innovating

business models, managers rely on simplified cognitive

models (Baden-Fuller and Morgan 2010), and hardly grasp

the entirety of interactions between different business model

components that give rise to complex dynamic behaviour

(Demil and Lecocq 2010). Their cognitive representations

are incomplete and the decisions regarding their innovation

are bounded rational.

One research stream receiving increasing managerial and

academic attention focuses on the modelling and simulation of

business models to counteract cognitive limitations and im-

prove managerial decision-making in the context of BMI

(cfr. Aversa et al. 2015; Baden-Fuller and Haefliger 2013;

Chesbrough 2010; Gambardella and McGahan 2010;

Rumble and Mangematin 2015). One established approach

to enhance understanding and facilitate effective decision-

making in the face of complexity represents System

Dynamics (SD) (Forrester 1961, 2007; Sterman 2000) and

hence, scholars have referred to SD as a promising tool for

managerial support during BMI (Markides 2015; Massa and

Tucci 2014).

The corresponding studies indicate that SD may be useful

during BMI because it can enhance managers when commu-

nicating, creatively designing, and evaluating new business

models (cfr. Cosenz 2017; Onggo et al. 2006), or assessing

strategic options for the optimization of existing ones (cfr.

Groesser and Jovy 2016). Research further indicates a positive

relation between the use of SD and managerial cognition in

BMI (cfr. Abdelkafi and Täuscher 2016; Barabba et al. 2002).

Yet, although these studies provide valuable insights, im-

portant issues remain. First, the empirical support and gener-

alizability for the highlighted benefits of SD for decision-

making in BMI is limited. The few existing publications ma-

jorly rely on the study of single cases. Although single case

studies enable rich descriptions of the phenomenon of interest

(Siggelkow 2007) the case variety inherent to multiple-case

studies enable the development of more robust and generaliz-

able theories (Eisenhardt and Graebner 2007). Further, the

mentioned studies focus nearly exclusively on business

models of entrepreneurial firms. Investigations into the use-

fulness of SD on BMI in corporate contexts remain an excep-

tion (cfr. Barabba et al. 2002; Moellers et al. 2017). Cosenz

(2017) and Groesser and Jovy (2016) concur on this, and

explicitly call for empirical research examining the usefulness

of simulation approaches for BMI. Second, so far there is

limited understanding of the constraints SD imposes when

used to support managers in understanding and making deci-

sions during BMI. This is surprising given the vital discus-

sions in the broader SD community (cfr. Featherston and

Doolan 2012; Royston et al. 1999).

Taken together, these issues suggest that despite the rele-

vance of simulation of business models for managerial sup-

port, scholars lack differentiated understanding of the useful-

ness of SD during BMI. We address this gap by asking: How

does System Dynamics support managers’ understanding and

decision-making along the different phases of BMI processes?

Given the theoretical and empirical limits of the existing

research we rely on an inductive multiple-case study approach

(Eisenhardt 1989; Yin 2014) which we complement with a Q-

Sort exercise (Stephenson 1953). Our setting comprises a

sample of five contextually diverse projects inside BMW.

Theoretical background

Business model innovation

A business model is most commonly considered as a struc-

tured and analytical model that defines the logic “by which the

enterprise delivers value to customers, entices customers to

pay for value, and converts those payments to profit” (Teece

2010, p.172). As an attribute of the real firm (Massa et al.

2017) a business model defines a distinct and holistic config-

uration of interrelated components (Baden-Fuller and

Haefliger 2013; Baden-Fuller and Mangematin 2013;

Baden-Fuller and Morgan 2010; Klang et al. 2014; Rumble

and Mangematin 2015). These components represent the

firm’s value proposition, the addressed market segment, the

value chain, the value capture mechanisms, as well as their

internal linkages but also the connections to the external en-

vironment (Al-Debei and Avison 2010; Ali-Vehmas and

Casey 2012; Foss and Saebi 2017; Saebi et al. 2016). Those

linkages impose feedback and circular causality to the busi-

ness model (Casadesus-Masanell and Ricart 2010; Sterman

2000) and cause non-trivial dynamic behaviour (Afuah and

Tucci 2003; Cavalcante et al. 2011; Demil and Lecocq 2010).

T. Moellers et al.388

Foss and Saebi (2017) define BMI as “designed, novel,

and nontrivial changes to the key elements of a firm’s BM

and/or the architecture linking these elements” (p.201).

There exist two complementary, but distinct conceptualiza-

tions of BMI. First, it may refer to an outcome, i.e. an innova-

tive business model. Contributions in this field comprise

contextually-embedded case studies that relate to specific sec-

tors (e.g., Abdelkafi et al. 2013; Hwang and Christensen 2008;

Karimi and Walter 2016; Kastalli et al. 2013; Souto 2015),

technologies (e.g., Richter 2013; Tapscott and Tapscott

2016), or socio-economic environments (e.g., Gebauer et al.

2017). Further, this line of research analyses the performance

implications of BMI (e.g., Fang et al. 2008; Suarez et al. 2013;

Visnjic et al. 2016).

Second, BMI refers to the (re)configuration process of

existing business models (Amit and Zott 2012; Berends

et al. 2016). Throughout this paper, we refer to this second

notion of BMI. Scholars concerned with BMI as a process

distinguish between two basic modes for innovation, i.e. ex-

perimental search and cognitive search (Gavetti and Levinthal

2000). For experimental search, new business models origi-

nate from experiments through which new configurations of

business model components are explored (McGrath 2010).

These actions and their real consequences are the source of

experience and guide the subsequent identification of new

combinations (Holland et al. 1986; Levitt and March 1988)

in a trial-and-error fashion (Mezger 2014; Sosna et al. 2010).

For cognitive search, managers instead rely on mental repre-

sentations to identify new business models and assess their

attractiveness according to established performance dimen-

sions (Aspara et al. 2011; Casadesus-Masanell and Ricart

2010; Doz and Kosonen 2010; Furnari 2015; Martins et al.

2015; Porac et al. 1989). These representations are conceptual

structures held by individuals that contain a simplified under-

standing of their environmental reality (Lakoff 1987).

Cognitive search therefore allows to assess multiple configu-

rations ‘off-line’ but the judgement of attractiveness underlies

errors in those representations or processing of their inherent

dynamics (Gavetti and Levinthal 2000; Lippman and McCall

1976). Berends and colleagues (Berends et al. 2016) find that

during BMI firms combine both modes with varying empha-

sis. For instance, when relying on leaping, firms start with a

focus on an cognition-centred mode before shifting to exper-

imental search.

In line with cognition-centred modes of innovation,

Frankenberger and colleagues (Frankenberger et al. 2013) de-

scribe four archetypical phases BMI entails, namely initiation,

ideation, integration, and implementation: The first phase of

the iterative process, named initiation, refers to managers’

attempts to gain understanding of the environmental develop-

ments and the corresponding identification of needs to inno-

vate. The ideation is a creativity-centred phase aiming to gen-

erate ideas for potential business model reconfigurations

responding to the observed developments. Whilst the ideation

is concerned with opening up the solution space of potential

business model configurations, the integration focuses on

elaborating the new business model along all components

and embedding it into the organizational context of the focal

firm. Finally, the implementation of the business model is

introduced to the market most dominantly through a series

of experiments.

System dynamics in the context of BMI

Formal conceptual business model representations play a cen-

tral role in mediating between operational reality and cogni-

tive representations during BMI. These refer to tangible ob-

jects through which managers explicate and materialize their

understanding relying on written, pictorial, mathematical, or

symbolic forms (Demil and Lecocq 2015; Massa et al. 2017;

Zoric 2011). The tangible property of conceptual representa-

tions enable their circulation between various individuals. By

using conceptual representations managers can consequently

enhance and guide communication, understanding, and coor-

dination within their firm when engaging in BMI (Doganova

and Eyquem-Renault 2009; Stigliani and Ravasi 2012).

Computational modelling has received increased attention

sought to produce conceptual representations that allow man-

agers to understand the consequences of changes in their busi-

ness models (e.g., Akkermans et al. 2004; Bouwman et al.

2008; Gordijn and Akkermans 2003). Previous studies relying

on cognition-centred BMI include stochastic modelling

(Euchner and Ganguly 2014) and quantitative scenario analy-

sis (Pagani 2009; Zoric 2011), amongst others. More recently,

scholars have called for research to connect System Dynamics

and BMI to accommodate for the dynamic interdependencies

inside business models (de Reuver et al. 2013; Kaplan 2012).

SD is a computer-aided approach to enhance analysis and

decision-making in complex systems (Forrester 2007). It re-

lies on formal modelling to develop a simplified and consis-

tent representation of the business model (Aversa et al. 2015;

Sterman 2000). Through quantification and computational

simulations of a system’s elements and interrelations of the

latter (Forrester 1961; Schwaninger 2006) this approach facil-

itates the consideration of feedback loops and dynamic behav-

iour (Sterman 2000). SD attempts to mitigate potential deci-

sion errors by providing users with means of learning about

complex systems and the potential consequences of their ac-

tions. Such ‘virtual worlds’ or ‘simulation games’ allow them

to explore the potential consequences of alternative business

model configurations (Aversa et al. 2015; Forrester 1961;

Sterman 2000).

SD has been applied to various domains including strategic

decision-making (Gary et al. 2009; Graham et al. 1992; Lyneis

1999), innovation processes (Galanakis 2006), and more re-

cently BMI. For instance, Abdelkafi and Täuscher (2016)

System dynamics for corporate business model innovation 389

applied SD to develop a conceptual business model for sus-

tainability. They found cognitive processes of entrepreneurs

and within established organizations to be important factors in

the design and evaluation of a “sustainable” business model.

Cosenz (2017) extended those insights through single-case

study applying SD in combination with the Business Model

Canvas (Osterwalder and Pigneur 2010). He concluded that

SD might be a valuable support tool for entrepreneurs

experimenting with different configurations of the business

model while keeping the costs of experimentation low. In

particular, he considered SD to enhance the design and per-

formance assessment for decision-making on new business

models but also the communication with external stake-

holders. The latter aspect may originate from the ability to

study how interactions of different stakeholders influence

market supply and demand as shown by Onggo and col-

leagues (Onggo et al. 2006) based on a hypothetical business

model for component-based simulation software. In the

course of a single case study, Groesser and Jovy (2016) gained

similar insights, referring to “simulation-based prototyping of

strategic initiatives” (p. 80) as a creative and iterative process

to analyse those effects on the current business model and to

optimize it. These findings are in line with Köpp and

Schwaninger (2014). Based on the study of Groupon’s busi-

ness model they applied SD to demonstrate its ability for “dy-

namic optimization, calibration and redesign of business

models” (p.3) to support entrepreneurs in the early phases of

business model design. The authors conclude that the model-

ling process enhances understanding of the underlying busi-

ness model logic allowing for possible redesigns through

model modification. Collectively, these studies indicate that

SD may support managers in cognition-centred BMI

processes.

A few authors also indicate limitations in the usefulness of

SD. For instance, SD models of business models tend to

quickly become complex leading potential users to feel

overwhelmed (Groesser and Jovy 2016; Groesser and

Schwaninger 2012). In addition, for high levels of uncertainty

regarding certain variables or cause-and-effect relationships,

the model behaviour may be difficult to track, validate or

falsify (Cosenz 2017). However, empirical evidence on the

usefulness of SD for managerial support during corporate

BMI remains scarce.

Methodology

This study addresses the research question ‘How does System

Dynamics support managers’ understanding and decision-

making along the different phases of business model innova-

tion processes?´ Given the limited current understanding of

this phenomenon, we opted for an inductive multiple case

study approach (Eisenhardt 1989; Yin 2014) of 5 embedded

cases inside BMW. Each case corresponds to one BMI project

in which managers heavily relied on SD for understanding and

decision support.

We initiated our research through a pre-study in which we

reviewed archival records of each project’s documentation.

For complementary means, we conducted three in-depth in-

terviews with managers heavily exposed to SD in their pro-

jects to identify initial theoretical constructs relating to the

relevance of SD in corporate BMI processes. In the course

of the main study, we conducted additional 18 interviews to

further corroborate and elaborate our initial findings. For an-

alytical support, we triangulated our insights from these inter-

views with those of a Q-Methodology (Jick 1979; Stephenson

1936, 1953). The Q-Methodology is a quantitative analysis

technique. It is centred around the forced ordering of ‘hetero-

geneous items’ that are subsequently analysed using an

inverted factor analysis (Stephenson 1936). This process al-

lows “to find qualitative ‘order’ even in domains where var-

iability and disparity seem initially to have prevailed” (Watts

and Stenner 2005, p.73) and can alleviate social desirability

bias (O’Reilly III et al. 1991). In the course of our study, the

Q-Methodology was used to sort statements concerning the

usefulness of SD in the context of the BMI projects our infor-

mants were involved in. This allowed us to understand differ-

ences in individual perspectives in sum facilitating more nu-

anced theoretical constructs. Overall, the triangulation of data

sources and research methodologies allows to create novel

theoretical perspectives of high internal validity (Gibbert and

Ruigrok 2010; Gibbert et al. 2008; Jick 1979).

Setting

The setting is BMW, a multinational automotive company with

more than 133′000 employees. BMW provides an attractive

research context for three main reasons: First, BMW has be-

gun to leverage SD for BMI in multiple projects (cfr. Moellers

et al. 2017). Second, BMW provided us with access to a vast

array of internal documents and informants. Notably, we were

able to make direct and continuous observations about the

development of two cases, including various internal discus-

sions. This richness of data allows for an accurate assessment

of the underlying phenomenon. Third, the long stability of the

automotive manufacturing business model and the looming

change the industry is confronted with – summarized under

variations of the acronym CASE (connected car, autonomous

driving, shared mobility, electrification) - constitutes an attrac-

tive setting to study BMI in an established firm.

BMW pursues a cognitive-centred approach to BMI, which

is exemplified by their paradigm “Fail in Virtuality to profit in

Reality!” often found in internal presentations on the topic.

Within BMW, the use of SD for BMI is referred to as

‘Business (Model) Simulation’ (BMS). BMS is an internally

applied method that leverages SD to support the design and

T. Moellers et al.390

evaluation of business models. It is divided into an iterative set

of six distinct phases, i.e. Sensing, Analysis, Transfer,

Aggregation, Simulation, and Decision. Sensing comprises

an assessment of the usefulness of SD for a specific business

model decision. Analysis aims to establish a common under-

standing of the business model between the modelling team

and the other managers. Transfer refers to the translation of

semantic descriptions into a model of cause-effect relation-

ships. Aggregation refines and substantiates the modelled re-

lationships to allow for testing of new business model config-

urations, which is part of the Simulation phase. The final

Decision phase aims to transform the deep understanding

gained about the business model into simple and precise de-

cision recommendations in respect to the specific context of

each manager. For a detailed description of the method we

refer to Moellers and colleagues (Moellers et al. 2017).

BMW operates a dedicated unit for the implementation of

BMS.

Grounding our analysis on a specific application method,

BMS, eliminates variance from several variables that affect

the usefulness of SD, e.g., the model development process,

the model usage and the integration of different managers into

the modelling process. Hence, it increases the reliability of our

analysis.

Data sampling and collection

This study relies on data from five BMI projects. We majorly

collected and analysed the following types of data: (1) tran-

scripts from semi-structured interviews; (2) archival data in-

cluding SD models, causal-loop-diagrams, brainstorming

maps, interim and final presentations, project management

artefacts; (3) observations from meetings, work periods, and

internal discussions; (4) an indicative survey answered by 59

senior executives of BMW Financial Services; (5) e-mails,

phone calls, and follow-up discussions with informants.

We began our sampling through a review of all past and

current BMI projects utilizing SD. This process revealed a

potential set of seven cases. By analysing archival records

and preliminary interviews, we generated brief write-ups of

each case containing project metadata (project dates, motiva-

tion, involved parties, outcomes) and a case chronology. We

found that two cases were similar in setup and context and

focused on the case for which more data was available.

Additionally, we excluded one case for confidentiality rea-

sons, leading to a final set of five BMI projects. Each of these

cases involves different sets of managers (mainly teams from

different departments) and project timelines. This high degree

of contextual heterogeneity across the cases enhanced the po-

tential to recognize clear pattern of the central constructs

(Eisenhardt and Graebner 2007).

Our main data body consists of interviews with 21 different

informants. We selected the informants based on their

involvement and decision-making power in the project. In

order to enhance the openness and truthfulness they were

granted anonymity. We ensured that we interviewed all in-

volved parties at least once for each project as well as the

respective decision-makers. We conducted the main inter-

views for this study between May and July 2017 in person

and with an average duration of 60 min. We chose German as

the interview language to avoid language barriers that might

prevent deeper reflection and understanding. Four types of

informants provided complementary viewpoints on the same

events:

1) modellers, who are trained in SD and lead the modelling

process;

2) input providers, who commonly work on issues closely

related to the BMI project and provide subject-specific

inputs;

3) decision makers, who bear the strategic and financial re-

sponsibility for the project in their respective department;

4) project owners, who are operationally in charge of the

project – typically the project owner(s) comprise one or

more employees of the decision maker.

The interviews were semi-structured and consisted of five

main segments: (i) background information on the informant,

(ii) project setup, involved parties, and primary goals, as well

as questions regarding the informant’s contributions to the

project; (iii) chronology of the major events and work phases;

(iv) direct questions related to insights derived from the appli-

cation of SD and its perceived usefulness in the project con-

text; (v) the Q-Sort, with follow-up questions about particular

sort choices.

During the interviews, we consistently followed an event-

based approach in line with the ‘courtroom style’ advocated

by Eisenhardt (1989) to reduce the impact of recall bias

(Huber 1985; Miller et al. 1997). We let the informant identify

key events in the project’s chronology and challenged the

informant’s statements only when inconsistent with our previ-

ously developed case write-ups. Table 1 provides an overview

of the five cases and the informants including their respective

roles, corresponding units, and details of the Q-Sort exercise

that we discuss in the subsequent section. Figure 1 outlines the

project development of the Regulation case along the different

BMI phases. Collectively, our sample largely represents the

firm’s experience with SD for BMI.

Data analysis

We transcribed all interviews aiming for the generation of a

rich base of textual data. We then employed a two-stage with-

in-case coding process to distil our data into higher order

codes (Gioia et al. 2012): First, we relied on open coding to

‘open up’ the text-based sources (Flick 2009). In order to stay

System dynamics for corporate business model innovation 391

Table 1 Case overview, informants, and Q-Methodology details

Case # Project Role Unit Q-Sort Factor loadings

(f1/f2/f3/f4)

Flagginga

Sharing

Car sharing platform based on a recently developed

technology

1 Modeller Business Simulation

2 Input Provider Service Business Models x 0.28/0.07/0.89/0.08 f3

3 Project Owner/ Input Provider Service Business Models x 0.83/0.05/0.16/0.02 f1

4 Project Owner Service Business Models x 0.07/−0.33/0.62/0.10 f3

Mobility

Mobility-as-a-service business model

5 Modeller Car Development for Mobility Services

6 Project Owner Car Development for Mobility Services x 0.14/0.21/0.17/0.79 f4

7 Decision Maker Corporate Strategy

Regulation

Electrification of fleet triggered by looming

regulatory changes

8 Modeller Business Simulation

9 Project Owner Market x 0.03/0.79/−0.02/−0.03 f2

10 Decision Maker Financial Services / Market Organization

11 Decision Maker Market Organization x −0.39/−0.11/−0.16/0.78 f4

Insurance

Business model reconfiguration proposal by

an insurance company

12 Modeller Business Simulation x 0.29/0.85/−0.01/0.24 f2

13 Project Owner Financial Services x 0.93/−0.01/0.04/−0.08 f1

14 Project Owner Aftersales x −0.35/0.40/−0.21/−0.29 –

15 Input Provider Aftersales x 0.47/−0.09/−0.60/0.16 f3

Assets

Business model reconfiguration to enable the

management of vehicles over multiple

selling/usage cycles

16 Decision Maker Financial Services

17 Decision Maker (Senior Management) Financial Services

18 Decision Maker Financial Services

19 Project Owner Business Simulation x −0.41/0.64/ -0.01/0.10 f2

20 Input provider Financial Services x 0.38/0.44/0.50/0.49 –

21 Decision Maker BMW Bank

a loadings at p < 0.05

T. Moellers et al.392

close to the data and avoid a premature introduction of theo-

retical ideas, we aligned with Charmaz’ (2014) suggestion of a

line-by-line coding, leading to code formulations very close to

the source (1773 codes in total). We then performed axial

Month 1 Month 2 Month 3 Month 4 Month 5

First Management

Board Decision (line

management):

Topical relevance is

confirmed based on

initial ideas about the

key mechanics of the

business model

(presented in the form of

a Causal Loop Diagram).

Second Management Board

Decision (line management):

Key ideas concerning the

management of the business

model and especially the

improvement of the current

situation are confirmed (presented

using a simplified simulation

concept showing causes and

resulting effects).

Presentation of results: All

global directors are informed

about the results of the business

model simulation and its

implications. Directions for

activities are proposed.

Presentation of

results:

All sales and

marketing directors

are informed.

Directions for

activities are pro-

posed.

Informal update to

management

(no decision required)

2. Ideation: Based on the initial understanding to prolong

the sales period, it was important to collect and evaluate

alternative ways to make use of the vehicles while they

were waiting to be sold (parking vehicles was seen as one

possible however not most promising option). Within the

ideation phase four alternative usage models are

exemplarily modelled in a simplified manner. The initial

modelling was used to help decision makers to reflect (1) if

the underlying cause-effect relationships resulted in the

expected manner or (2) if severe and fundamental questions

arose that needed further discussions and modelling.

Luckily, the initial ideation had been perceived with great

acceptance and hence, laid the groundwork for further

integration in the next step.

4. Implementation: Finally, the results of the analysis had

been aligned with actual numbers from selected markets to

make the generalized results of the analysis more tangible

for each addressed market. Also, the plausibility of the

simulation model was checked against ‘real’ numbers from

past experiences as well as a calibration of specific factors

had been conducted. For this data analytics on few and

carefully selected factors had been executed. On this basis

mathematical optimization was calculated and call for

actions had been derived and prepared for communication

on two internal conferences.

1. Initiation: Based on anecdotal evidence from several

markets, the issue of so called ‘fire sales’ occurred. A ‘fire

sale’ describes a situation in which a significant tranche of

young used cars of a specific series is put on the market for

sale. In line with economic theory, a considerable reduction

of price levels occurred striking the whole used year fleet

and not only the tranche at sale. In addition, prices do not

rebound instantly as soon as the tranche of vehicles is

absorbed by the market to the original level, but rebounds

with a considerable time delay and not to the original level.

In practice, such ‘fire sales’ have a recognizable impacts on

the financial performance of the business. The initial

analysis suggest to put a fraction of the initial tranche on

the market at once and to stretch the sales period. This

stabilizes prices and reduces negative financial impacts of

‘fire sales’. The key mechanics of the business model

reconfiguration was captured in a Causal Loop Diagram,

especially for communication purposes to the board of

managers.

Legend: Detailed explanation of working

sessions and official meetings

Flow of activities, traversing communication

and work sessions interactively

3. Integration: The four alternative usage models were

analyzed in greater detail, adding market and headquarter

expertise on each of the usage models, adding knowledge

on financial consequences of each usage model and

especially considering developments of each usage model

over time. In sum, three factors exerted dominant relevance

for implementing business strategies of each usage model:

(1) vehicle capacity, (2) financial implications and (3) the

role of time a vehicle spends in one usage models. For

instance, it became clear that one usage model that used the

vehicles for short term rent seemed to be financially

superior to the other usage models, however, the number of

vehicles that the usage model could absorb was strongly

limited. On the other hand, vehicles that were parked lost

significant value after reaching a certain threshold of days

being parked. Based on the analysis, it was possible

(already in this step) to ‘put numbers’ on each usage model.

While the key mechanic of each usage model had been well

understood on a qualitative level, showing the impact of the

usage models in financial terms helped decision makers

and management draw conclusions about the importance of

the topic.

(ii) Working sessions:

(i) Communication of milestones/Official meetings:

Fig. 1 Project development of the Regulation case

System dynamics for corporate business model innovation 393

coding by aggregating these first order concepts into code

groups, which we labelled by second order themes (Flick

2009; Gioia et al. 2012). In order to increase construct validity,

we triangulated our interview data with archival records and

case histories throughout this stage. This allowed us to har-

monize different accounts of the same case. Following

Eisenhardt (1989), we subsequently performed a cross-case

analysis based on the generated hierarchical code structure.

Before turning to the findings the next section provides

details relating to the derivation of statements and evaluation

of responses of the complementary Q-Methodology.

Derivation of Q-methodology statements

At the core of the Q-Methodology lies a set of multifaceted

items on the phenomenon of interest. In the context of our

study, we relied on the Q-Methodology as a complementary

analytical approach to analyse the established or emerging

theoretical constructs. We formulated these constructs as hy-

pothetical statements and presented them as a Q-sort exercise

to 12 of our informants.

We relied on two sources for the generation of statements:

First, we searched the BMI literature for concepts that have

been associated with managerial understanding and decision-

making and generated one statement for each identified

concept. For instance, scholars have found that the capability

of managers to initiate BMI strongly depends on their ability

to interpret external developments as threats or opportunities

and further that perceived threats and opportunities have dif-

ferent effects in initiating BMI (Gilbert 2005; Osiyevskyy and

Dewald 2015; Saebi et al. 2016). In this case, we formulated

two statements relating to the concepts of both threat and

opportunity:

1. “In this project, Business Simulation helped me to

recognize and assess external threats to BMW”

2. “In this project, Business Simulation helped me to

recognize and assess opportunities to BMW”

Second, we used the interview data from the first phase

to identify relevant cognitive mechanisms that were not

covered by the previous statements. We subsequently re-

fined this set through discussions among the researchers

and informants. In total, this process yielded a set of 16

statements, which is consistent with prior literature (Cross

2005). Table 2 lists the 16 statements and provides foun-

dational literature references where applicable. It further

lists the z-scores for each factor correspondingly to the Q-

Sort statements. We briefly discuss their development in

the following section.

Evaluation of Q-Methodology responses

The Q-Methodology ‘inverts’ the axes of the Q-Sorts and thus

undertakes a factor analysis of the individual respondents as

dependent variables and the questions as samples (Stephenson

1936). This produces groups of respondents who tend to rank

the same statements similarly. Our analysis of the 16 re-

sponses followed a two-step approach in line with established

Q-Methodology practices (cfr. Watts and Stenner 2005). In a

first step, we reduced the data dimensionality using a principal

components analysis. We rotated the factors using the stan-

dard varimax rotation method to maximize the variance of

factor loadings. For the selection of the major factors to be

included, we imposed two standard Q-Methodology require-

ments (cfr. Watts and Stenner 2005): a) eigenvalues for each

factor are greater than 1.00; b) features at least two significant

loadings. The factor loadings and flagging for each Q-sort

respondent are displayed in Table 11

. Subsequently, to identify

those statements that distinguish the different vectors, we cal-

culated the difference in the normalized factor score (z-score)

for each statement (Table 2).

Overall, our insights from the various data sources of the

multiple case study and the results of the Q-methodology

served as the foundation for a set of 11 propositions that col-

lectively provide a testable entry to our emergent theory. Our

inference from empirical data to theoretical explanation is

based on inductive reasoning (Ketokivi and Mantere 2010).

We sought to bridge the unavoidable logical gap of inductive

reasoning, i.e. a set of empirical data can be used to formulate

different theoretical explanations, each coherent with the data

(Goodman 1954; Maher 1998), by means of inter-researcher

reasoning (Eisenhardt and Graebner 2007).

Findings

Our research seeks to assess how SD supports managers along

the BMI process. The qualitative analysis of five cases sub-

stantiated the idea that these effects depend on the model

properties, the model development process, and the way the

insights are presented. The importance of each of these aspects

varies depending on the respective phase of the BMI process.

The analysis of factor loadings from the Q-Sort further indi-

cated that informants’ perception of the usefulness of SD in a

BMI context vary, even within the same case and between

managers of the same role. Correspondingly, the four respon-

dent groups – each represented by one of the factors (f1–4) – do

not reveal a distinct character (in terms of role, unit, or case).

For instance, project owners are represented in each factor but

each factor comprises respondents of other roles.

1 Complementary parts of the Q-Sort analysis can be found in the Appendix.

T. Moellers et al.394

In the following, we present our findings structured along

the four BMI phases (Frankenberger et al. 2013). Thereby, our

main focus lies on the insights generated from primary inter-

view data, observations and archival records of the cases. We

complement these by results from the Q-Sort analysis.

Initiation

In the context of the initiation the informants reported in-

creased predictability of the impact of external developments

on the existing business model based on insights from the SD

model. We observed this in the Regulation and Insurance

cases, in which managers were able to better understand the

consequences of changes in the regulative environment or a

proposed business model change from a collaboration partner.

In both cases, the cross-functional approach and the extensive

amounts of collected data facilitated the detailed identification

of change drivers. Furthermore, the SD model enhanced the

interpretation of the ‘general direction of outcomes’ (16 quo-

tations) in the Insurance case. Similarly, in the regulation case

enabled managers to quantify the impact of climate regulation

on BMW’s business model in the long-term planning horizon

(5 quotations). In this context, one manager stated:

We know the dynamic trajectories of the volume-models

and the individual units and you do rough estimations.

When you see [redacted: country-specific regulatory

changes] you simply know, ‘this is going to be extremely

Table 2 Q-Sort statements

# Statement: “In this project, Business Simulation

helped me to…”

Corresponding literature f1 f2 f3 f4

Initiation

1 recognize changes in the environment that will impact

BMW’s business in the future

(de Reuver et al. 2009; Demil and Lecocq

2010; Doz and Kosonen 2010;

Frankenberger et al. 2013)

−1.78 0.01 −0.62 2.26

2 recognize and assess external threats to BMW (Gilbert 2005; Osiyevskyy and Dewald 2015;

Saebi et al. 2016)

−1.49 −0.23 −0.79 1.12

3 recognize and assess opportunities for BMW (Gilbert 2005; Osiyevskyy and Dewald 2015;

Saebi et al. 2016)

0.47 0.60 0.30 −0.34

4 understand the longer-term dynamics of BMW’s

existing business model

Emerged from preliminary interviews −0.65 2.17 0.20 −1.51

Ideation

5 understand how a business model (BM) from a

different context/industry could be applied to BMW

(Gentner 1983; Martins et al. 2015) −0.47 −1.92 −1.00 −1.52

6 combine multiple components (e.g. a certain sales

approach or a printer-and-ink revenue model) of

existing business models in a novel way

(Gassmann et al. 2014; Martins et al. 2015;

Wisniewski 1997)

−0.18 −0.71 1.97 0.00

7 leverage my professional experience at BMW to

generate new BM ideas

(Bingham and Eisenhardt 2011; Loock and

Hacklin 2015)

−0.47 −0.72 −0.23 −1.14

Integration

8 include the customer dimension in the business model

properly

Emerged from preliminary interviews 0.37 −0.60 0.61 0.38

9 grasp the complexity of a business model Emerged from preliminary interviews 0.84 1.32 1.26 0.77

10 communicate the simulated business model to

stakeholders (e.g. involved employees, partners,

decision makers)

(Demil and Lecocq 2015; Doganova and

Eyquem-Renault 2009; Täuscher and

Abdelkafi 2017)

1.59 0.37 −1.97 −0.37

11 group the business model into relatively independent

components

(Aversa et al. 2015) 1.12 −1.21 0.72 −0.74

12 identify areas of conflict between BMW’s current

business model and potential configurations of

BMW’s future business model.

(Kim and Min 2015) −0.37 −0.13 −0.61 −0.38

13 give my recommendations about business model

changes more credibility within the organization

(Gavetti 2012; Sosna et al. 2010) 0.18 1.20 −0.96 0.74

14 critically reflect on the adequacy of BMW’s dominant

“way of doing business”

(Prahalad and Bettis 1995) −1.12 −0.12 0.11 −0.01

Implementation

15 formulate concrete actions for implementing a

business model

(Baden-Fuller and Morgan 2010; França

et al. 2017; Petrovic et al. 2001)

0.47 −0.12 −0.12 −0.01

16 test and compare different possible configurations of a

Business Model

(Andries et al. 2013) 1.49 0.11 1.13 0.75

System dynamics for corporate business model innovation 395

expensive’ and the System Dynamics method was ideal

to better understand and quantify this ‘extremely expen-

sive’. - #9: Project Owner

Related to the same case another interviewee further

highlighted how SD facilitated attention among top manage-

ment for strategic issues:

To us [the anticipated financial impact on the business

model] seemed somehow obvious, however in regards

to the fines, the level of the peak in 2020 was quite a

surprise […] and when you suddenly start talking about

these amounts, then people start listening. […] of course

BMW knows this somehow. Somewhere some people

know that this exists […] But it is about the creation of

this connection between the headquarter and the mar-

ket. - #8: Modeller

Such a long-term horizon of the quantified forecasts was high-

ly unusual in the context of the Regulation case that is char-

acterized by a short-term sales orientation (8 quotations). In

return, project members were able to raise attention among

decision makers for possible solutions. Specifically, the pro-

ject team could argue that investments for activities that re-

duce (potential) fines would best match the level of these

avoided fines. In sum, the detailed forecasts proved crucial

for the approval of two related BMI projects.

The Q-Sort results provide complementary findings.

Interestingly, f1 (−1.78, −1.49) and f4 (2.26, 1.12) assign high

but contrary scores to the statements that relate to the recog-

nition and assessment of opportunities.2 The identified envi-

ronmental developments related to the Regulation and

Mobility case where rather perceived as threats, whereas the

Insurance and Mobility case this assessment was neutral to

positive. Under this consideration the results indicate that

managers felt supported to gain a clear understanding of the

impact of the identified developments, whereas the SD model

did not change or reveal opposite perceptions. Respondents of

f2 put the most significant scores on the statement “In this

project, Business Simulation helped me to understand the

longer-term dynamics of BMW’s existing business model.”

(2.17) Considering their affiliation with either the Market, or

the Business Simulation unit, the SD model provides them

with a holistic understanding of how the individual parts of

the business model collectively drive performance along the

parameters of interest.

Proposition 1. Quantifiable results from system dynamics

models are useful in guiding attention of decision makers

towards relevant environmental developments.

Proposition 2. System dynamics models provide holistic

business understanding for domain experts.

Ideation

In the context of the ideation, our cases showed only lim-

ited effectiveness in applying SD for the generation of new

business model ideas. Our thematic coding supports this

finding on a more granular level: In contrast to 11 quota-

tions relating to the code ‘No new operational ideas gen-

erated through BMS’, only two quotations related to the

code ‘New BM idea.’3 The Q-Sort analysis majorly sup-

ports these findings. The groups of respondents consistent-

ly put negative loadings on the statement “In this project,

Business Simulation helped me to understand how a busi-

ness model from a different context/industry could be ap-

plied to BMW.” Similarly, respondents reported difficulties

to “leverage [their] professional experience at BMW to

generate new BM ideas” and to “combine multiple compo-

nents of existing business models in a novel way.” The

loadings for these two statements are, however, lower.

Surprisingly, one group of informants (f 3 ) put very high

scores (1.97) on the latter statement. This group entails

two input providers and one project owner.

In an attempt to consolidate these findings, we interpret that

SD is not well suited for the generation of innovative business

model ideas. The standardized notation built on systems of

cause-and-effect relationships hinders creative cognition.

However, when the innovation heavily relies on the reuse/

reconfiguration of components or mechanisms that are well

understood, since they are based on existing business models.

In these cases managers may perceive that their professional

experience is useful, but they cannot leverage it themselves.

However, when the underlying cause-and-effect relationships

of knowledge domains, such as business model components

well are understood some managers might be able to translate

this domain knowledge into new component combinations

using SD.

Proposition 3. System Dynamics models are of limited

use in generating new business model ideas that rely on

creative cognition.

Proposition 4. If the underlying cause-and-effect relation-

ships of business model components are well understood,

this knowledge can be used for new component

combinations.

2 f1 (project owner: Insurance, project owner: Sharing); f4 (project owner:

Mobility, decision maker: Regulation)

3 Both quotations relate to an idea for a business model innovation in the

context of the Insurance case. This new idea, however, quickly proved to be

unviable after minimal investigation.

T. Moellers et al.396

Integration

During the integration our informants reported that the SD

model facilitated the elaboration of business model ideas and

the creation of shared understanding among managers. In the

following both aspects are outlined separately. Across all

cases the interviewees attributed a more holistic elaboration

of the business model to the application of SD modelling (18

quotations). This effect can be divided into support in ‘identi-

fying own knowledge gaps’ and ‘understanding the business

model as a whole.’ First, the SD modelling process allowed

managers to identify knowledge gaps for instance, by

highlighting business model elements that require a more dis-

tinguished understanding such as motives for customer behav-

iour (Sharing case). Second, the SD model itself allowed man-

agers to understand the business model holistically by provid-

ing those involved in its development with the perception to

capture the real-life complexity (‘capturing complex relation-

ships in a business model’, 12 quotations; ‘BMS encourages

realistic modelling’, 9 quotations; ‘including circular rela-

tions in the business model, 3 quotations). For instance, in

the Regulation case, managers modelled parameter variations

for multiple model elements and ran simulations to analyse

their effects over time.

Essentially, the development of the SD model demands

managers to form explicit and precise definitions about the

existence of business model elements and their interrelation-

ships and to reflect critically on the development of certain

model parameters by running simulations. These mechanisms

guide them towards deeper levels of understanding, which

outsiders may lack of (‘information gap between modellers

and non-modellers’, 3 quotations). In summary, we posit

Proposition 5. System Dynamics modelling processes in-

duce deep reflection and a fine-grained understanding of

the complex relationships in new business models.

Throughout all cases, we found evidence that the applica-

tion of SD had a consensus-forming effect/created shared un-

derstanding among those managers directly involved model

development. We observed three properties to facilitate this

effect: First, the development of a holistic model integrating a

diverse set of managers and units consolidates different un-

derstanding by enforcing a consistent terminology. For in-

stance, in the Insurance case there originally existed different

mental models, and even abbreviations for the same business

model:

Regarding the claims, it was already a revelation in

terms of wording: When we [as Aftersales] talk about

«claims», then my primary focus is the claimed parts,

because we are assessed based on revenue on parts …

‘What is actually the KPI of an insurance?’ They are

assessed based on CR, the Combined Ratio, which we

internally refer to as Cost of Retail. Already the wording

is different between Financial Services and Sales. You

really need to realize what you and the others talk about.

- #14: Project Owner

Second, the SD model facilitates the consolidation of mind-

sets between managers by providing a neutral frame that can

be highly effective at “de-politicizing” otherwise divisive

topics:

I think [using SD] is meaningful, once I approach cross-

departmental issues, because […] I can de-emotionalize

those using such a tool. - #13: Project Owner

Overall [applying SD] was very, very constructive, and

the political discussions fade into the background or

disappeared completely. - #12: Modeller

Third, the SD model creates shared understanding through the

explicit integration and quantification of causal relationships.

Whereas it is traditionally difficult to track the consideration

of certain business model elements in decision-making pro-

cesses, SD allows managers to transparently ensure that all

relevant ones are taken appropriately into account. Thereby,

discussions tend to favour global maxima in the use of con-

flicting assets, such as financial resources within the business

model. Surprisingly, the z-scores from the Q-Sort exercise

reveal consistently negative factor loading on the statement

‘BMS helped me to identify areas of conflict between BMW’s

current business model and potential configurations of

BMW’s future business model’. We interpret this result to orig-

inate from the definition of the model boundaries. In order to

manage the complexity of the SD model, elements from adja-

cent business models inside BMW are only taken into account

if they are immediately affected by the new model. This in

return allows managers to resolve potential asset conflicts

across departments within the modelled system.

Proposition 6. System Dynamics facilitates shared under-

standing (especially on divisive issues and conflicting

assets) within the model boundaries by providing a neu-

tral and consistent frame for discussion.

However, the integration of a diverse set of managers into

the modelling process can also foster a type of ‘least common

denominator’-consensus, leading to a simulation model large-

ly consisting of simple, often linear relations among variables.

We observed this in the Sharing project: Originating from

different mental models the informants reported that they per-

ceived the SD model to differ markedly from the “true” logic

of the underlying business model. Our triangulation with ar-

chival records revealed that shared consensus was only creat-

ed between a set of variables in linear relation to customer

System dynamics for corporate business model innovation 397

satisfaction, which was used as a proxy for demand and rep-

resents the dominant business case logic inside BMW. While

the process of building the simulation model helped the

Sharing project members to understand each other’s mental

model, the final SD model did not represent real consensus of

the business model logic.

Proposition 7. When managers’ mental models are highly

diverse, system dynamics models tend to reflect the dom-

inant logic without creating a new shared understanding.

Evidence from the Regulation and Sharing cases suggests

that the modelling process that determines the access and uti-

lization of data moderates the extent to which the SD model

creates new shared understanding. For instance, in the evolu-

tion of the Sharing case, single elements that reflected an

integral part of a non-dominant mental model in the business

model logic, such the importance of a ‘word-of-mouth effect,’

were eventually excluded from the simulation model due to a

lack of underlying data. Eventually, this reduced the simula-

tion model back to set of established financial variables and

their relationships for which data was more readily available.

In contrast, a systematic data gathering process from inter-

nal and external sources characterized the Regulation case.

Consequently, the single code with the most related quotations

was ‘utilize expert input/feedback’ (12x). The resulting simu-

lation model differed significantly from the market organiza-

tion’s dominant logic. Still, decision makers judged its results

credible, largely because data was readily available or obtain-

able by the Regulation project team and the managers had

confirmed the high data quality. The focus on acquiring and

using data outside of immediate reach in the course of the

modelling process (e.g., data located in another organizational

unit within BMW) contributed much to the overall credibility

and revelatory potential of the simulation. One decision maker

states:

It is quite uncommon in the context of sales to disassem-

ble a project on such a level of detail before

reassembling it again. Hence, I think it is worth sharing

that there was definitely admiration for this system. -

#10: Decision Maker

We summarize:

Proposition 8. The data handling (determined in the

modelling process) moderates the extent of shared under-

standing created by the system dynamics model.

In contrast to the consensus created among managers in-

volved in the modelling process, multiple informants reported

that “outsiders” sometimes perceive the simulation model as

‘black boxes’ (13 quotations) and start to mistrust the

simulation results. Essentially, the inherent model complexity

inhibits the effective communication of insights to non-

modellers. For instance, one informant in the Sharing case

stated:

Everybody looks at the front end and presses a button.

Then something happens and you start thinking ‘hm,

somehow this relationship looks odd’. But in this mo-

ment you can barely grasp if a careless mistake slipped

in somewhere, if a relationship is actually non-existent

or not appropriately represented… - #4: Project Owner

Informants across all cases attributed three factors to the failed

communication. First, informants perceived a ‘liability of

newness’ that applied to the use of SD simulations for decision

support per se. This applied especially to managers from the

controlling department who are regularly confronted with

business case calculations. One interviewee shared:

Some people did not fully understand why we try to

develop a simulation model and the added value of this

approach, if it is meaningful or why we can’t just do it in

Excel. People confront us quite often with such ques-

tions and it is difficult to provide a concise explanation.

Especially, if people have never used a simulation soft-

ware, it is difficult to say ‘You know there are tables and

you can adjust graphs manually’ … that is not tangible

and people do not understand it they haven’t tried it

themselves.” - #5: Modeller

Second, informants reported difficulties to convert the simu-

lation model into a simple narrative for presentation purposes.

This, however, was reported as crucial due to the decision-

making setting, which involves a large audience and short

time slots to pitch innovation initiatives (cfr. Bartel and

Garud 2009). One interviewee concludes:

[A] presentation using the business model simulation is

quite complex for a larger audience and rather leads to

questions than results - #2: Input Provider

Third, the multitude of mathematically specified relations in

the model seems to be difficult also for informants to recon-

struct when facing questions on model details by decisions-

makers.

The corresponding results from the Q-Sort consistently

confirmed the impression that SD models help to make sense

of the real-life complexity inherent to any business model. All

informants ranked the statement ‘BMS helped me to grasp the

complexity of a business model’ (statement #10) high (0.84/

1.32/1.26/0.77). Interestingly, when asked about the useful-

ness of SD for the communication of the business model to

outsiders (statement #11) only one group of informants put

T. Moellers et al.398

significantly positive z-scores on this statement (f 1 , 1.59),

whereas another group in contrast used highly negative scores

(f3, −1.97).

We interpret this result as follows: When confronted with a

SD model managers understand the complexity of the real

business model. However, only those who have the possibility

to thoroughly analyse the underlying dynamics gain deeper

levels of understanding. Others remain ‘outsiders’: they feel

overwhelmed by the level of complexity and unable to closely

follow the line of reasoning.

Proposition 9. The use of System Dynamics models in

business model innovation projects complicates the com-

munication of insights to non-modellers, especially to

larger audiences.

Elaborating on means to improve communication when

relying on insights from SD, the informants frequently re-

ferred to ‘translation’ of the insights suiting the decision-

making context as a critical component (14 quotations relating

to this code group in total). How can the insights created from

the simulation model be ‘translated’ for the audience in pre-

sentation contexts? We identified two distinct mechanisms:

First, by integrating conventional metrics in the SD model

and exclusively focusing on these when communicating to

decision-makers. Among others, we observed how this ap-

proach enhanced the communication in the Mobility case.

Here, initial discussions concerning the value of on-demand

mobility and its implications for BMW remained abstract and

did not sufficiently take the insights from the SD model into

account. Eventually, managers integrated a metric that is long-

established within the manufacturing division of BMW: pro-

duction costs per unit. In spite of a minor and trivial mathe-

matical operation to transfer to this metric, this step signifi-

cantly improved the discussions with managers from other

units.

Second, multiple informants from the Sharing case

highlighted a complexity reduction in visual interfaces as a

valuable approach to suit the underlying model to the commu-

nication context with non-modelling managers. By hiding vi-

sual tools to define key settings of the simulation (e.g.,

sliders), managers can reduce the perceived complexity of

the model. For presentations to larger audiences managers

may even refrain from presenting the simulation at all and

communicate only static key scenarios. For instance, in their

communication to decision-makers in the Regulation case

managers began their presentation by outlining the function-

ing of SD in general terms. They continued to highlight the

activities to ensure and exhaustiveness and rigor in the gath-

ering and aggregation of the used data. This included to elab-

orate on the data’s origin that had been gathered from internal

market experts. The combination of a sophisticated, theoreti-

cally appealing methodology and rigorous execution created

credibility for the method. For the actual presentation of re-

sults to the decision-makers, the managers did not leverage the

SD model. Instead, they focused on the resulting insights re-

lying on established terms and the formulation of concise core

statements such as ‘potential regulatory fine payments.’ This

indeed proved to change the mind-set of the decision-makers

who became aware of the acuity of impeding regulatory

changes.

Proposition 10. The communication of insights from the

system dynamics model to managers not involved in the

modelling can be improved through the integration of

established metrics and terms and complexity reduction

in visual interfaces.

Implementation

In the context of the implementation we observed the notion of

‘risk’ as a dominant issue across the majority of interviews.

This heavily guided the ways by which managers utilized SD

throughout this phase. Here, our cases revealed two distinct

modes: First, instead of engaging directly in experimentation,

managers leveraged model simulations for quasi-experiments,

through which they intended to test broadly, efficiently, and at

low risk levels. Q-Sort respondents consistently assigned pos-

itive scores to the statement “BMS helps me to test and com-

pare different possible configurations of a business model.”

One interviewee elaborates on this issue:

Of course, you could say ‘let’s do it as a startup’ or you

can order it from a startup or invest in one through

BMW ventures, or alternatively you say: ‘I have this

group that can simulate systems, that just costs me

man power, IT and brain.’ And thereby I can already

start testing. […] It’s faster and cheaper. And you get

more variants. Within the firm I cannot say: ‘It didn’t

work out, so let’s try it a different way.’ I have my targets,

my KPIs, my revenue objectives, my profit objectives,

this needs to have some kind of flow. You can always

conduct some minor tests, but not too many. - #16:

Decision Maker

Second, managers made use of the SD model to identify the

most critical levers for subsequent testing. Hence, SD was

leveraged as preparatory means for experimentation. By test-

ing various scenarios, it enabled managers to develop sugges-

tions for concrete steps of subsequent action.

However, the results from the Q-Sort reveal that the use-

fulness of SD for such subsequent operationalization is rather

limited. The informants consistently put relatively neutral fac-

tor loadings on the statement ‘BMS helped me to formulate

concrete actions for implementing a business model’ (0.47/

System dynamics for corporate business model innovation 399

−0.12/−0.12/−0.01). This may explain why in some cases the

simulation results were only used as a preparatory means to

steer mutual agreement on a series of actions. For instance,

one interviewee stated:

We came up with very concrete suggestions for action,

but developed it again in a workshop […] So we pre-

pared [the workshop] a bit, but jointly formulated [con-

crete actions]. - #8: Modeller

In sum, we interpret that whilst individual managers attribute

great value to SD models in partly replacing or preparing

experimentation-related actions, others do not. Yet though,

they may agree on the same set of actions when confronted

with them through conventional means.

Proposition 11. System Dynamics models accommodate

managerial risk aversion by providing an efficient mean

for (pre-)experimentation guiding subsequent steps of

action.

Figure 2 provides an overview of the SD model used for the

Regulation case.

Discussion and conclusion

We add to theories of system dynamics and managerial cog-

nition. Prior theory suggests that SD has positive effects on

managers’ perceived ability to understand and make decisions

in BMI. But empirical evidence is limited and does not suffi-

ciently differentiate between the different activities within the

same BMI process. Addressing this gap, we explored how

managers at BMW used SD throughout their BMI processes

to support their understanding and decision-making across

five contextually different cases. We condensed our insights

into a set of eleven propositions thereby “bridging the often

wide gulf between qualitative and quantitative researchers”

(Gioia et al. 2012, p.25) to facilitate future theory testing.

Prior research has emphasized the difficulties for managers

to cope with dynamic behaviour of systems imposed by the

cognitive limitations of the human brain (Cyert and March

1963; Weick 1979). When reconfiguring their business

models these lead managers to make decisions based on in-

complete, lower dimensional cognitive representations of rel-

evant interdependencies within the firm and to its external

environment (Gavetti and Levinthal 2000; Tripsas and

Gavetti 2000). Consolidating our findings, we conclude that

SD modelling enables corporate managers to reflect deeply

about a business model’s component architecture and conse-

quently to develop higher dimensional cognitive representa-

tions. SD-based simulations enhance their processing capabil-

ities and allow them to better quantify the impact of

environmental changes. This is particularly valuable when

initiating BMI because managers can better anticipate the

existing business model’s prospective performance and raise

decision makers attention. Along the BMI process the SD

model integrates the input from various managers into a sin-

gle, fine-granular model containing mathematical descriptions

of cause-and-effect relationships. SD thereby allows individ-

uals to overcome cognitive limitations by leveraging the intel-

ligence from multiple managers and computational simula-

tion. Such combinations of computational processing and hu-

man cognition capabilities have been demonstrated to facili-

tate valuable solutions in the context of business model vali-

dation (Dellermann et al. 2018).

However, our findings also highlight that this integrative

property needs to be handled with caution. When elaborating

on potential business model configurations the SD models

guide managers towards new shared understanding. Yet, if

the individual mental models are highly diverse and support-

ive data for new representations may be out of immediate

reach, the shared understanding tends to reflect and enforce

the conventional business model logic. A mitigating strategy

is the purposeful gathering of data reflecting alternative

configurations.

Additionally, SD may impose what we would call a

cognition gap among managers. This gap refers to the

challenges arising from different levels of dimensionality

in the cognitive models between managers that are in-

volved and those not involved in the modelling process.

During the modelling process managers gain thorough

understanding of fine-grained dynamics inherent to the

business model, for instance by varying the graphs of

individual relationships and observing the consequences

on other parts of the model in simulations. Among those

being involved in this kind of practice the SD model pro-

vides a neutral frame in which different mental models

become transparent and can be openly discussed. This is

what our informants referred to when they described SD

as tool to de-emotionalize and facilitate constructive dis-

cussions. ‘Outsiders’ however, that hold lower dimension-

al cognitive representations grasp the underlying com-

plexity of the models, yet they feel overwhelmed by it

and perceive them as ‘black boxes’. Such varying levels

of dimensionality impede constructive discussions. Due to

the sheer amount of variables, their value ranges and re-

lations, modelling managers may fail to recall every mod-

el property in the moment of discussion when confronted

with requests on a specific detail leading non-modelling

managers in extreme cases to generally suspect the model

of holding faulty assumptions.

We observed different strategies aiming to overcome the

cognition gap. First, an integration of conventional metrics

into the SD model aims at building a connection between

the dominant cognitive representations and the simulation

T. Moellers et al.400

model. Second, a simplification in visual interfaces reduces

the perceived complexity and is achieved by simplifying the

visible architecture of elements and by focusing on a selection

of static scenarios.

Notwithstanding these insights, there are some limitations

that apply to our research. First, our findings derive from a

specific application of System Dynamics, BMS. The BMS

method represents a distinct approach used by BMW to lever-

age SD during BMI. Any transfer from BMS to SD is to be

taken carefully. Nevertheless, since our approach aligns close-

ly with the best practices in SD modelling formulated by

Martinez-Moyano and Richardson (2013) we expect that our

insights and corresponding propositions on the usefulness of

SD in the context of corporate BMI possess external validity.

Second, BMW’s BMI processes and culture are also distinct

and may limit the transferability into other corporate contexts,

for instance with lower levels of risk aversion. Third, our cases

relate to a cognition-centred approach for BMI. As outlined

before, this approach differs significantly from experimental

modes of search often associated with innovation of the entre-

preneurial firm (Blank 2013; Sarasvathy 2001). Therefore, we

are not able to draw conclusions about the usefulness of SD to

support experimental search in general. However, considering

that BMW deploys significant resources (in terms of time,

human and financial resources), we assume that SD is better

suited for corporate environments.

Despite these limitations, we conclude that our study is

amongst the first empirical inquiries into the application of

SD for BMI in a corporate environment. It provides a differ-

entiated understanding of the benefits and challenges when

managers leverage it as a tool to support their understanding

and decision-making along the different phases of the innova-

tion process. We invite further research to build on our insights

to extend and test theory in this domain.

Usage Model Weighting Mechanism

Flash Sale Parameter

#

Potential off-lease Stock

UM REMA Stock Cost Rate Remarketing

Subvention

UN REMA assumed loss

Buffer Parking CpDay

Buffer Parking Cost

Rate

Buffer Parking assumed loss

UM DN Stock

DriveNowFleetIn

UM UCL Stock

Used Car Leasing

Rate

UM DN CpVcl

UM DRNW Cost Rate

UM DRNW assumed loss

UM UCL CpC

UM UCLS Cost Rate

UM UCLS assumed loss

Total Cost over all

Usage Models

Assumed Loss All Usage Models

Mkt off-lease vcl

price avg

Apply Usage Models

Apply Usage ModelsAccumulated Activity

Remarketing

Percentage on Flash

Sale

DriveNow Percentage

on Flash Sale

UsedCar Percentage on

Flash Sale

UM DriveNow

weighting

UM Used Car

Leasing weighting

UM REMA weighting

Additional Volume from

Max Capacity Drive Now

UM DN Inflow Rate

Real Flash Sale

Real Flash Sale

Additional Volume

ReMarketing Max Cap

Additional Volume

ReMarketing Max

Cap

DriveNow

Percentage on

Flash Sale

UsedCar

Percentage on

Flash Sale

Buffer Parking

Inflow Rate

BufferParking Buffer Parking

Outflow Rate

Buffer Parking Vcls

UM LTCarPark

weighting

Longterm Parking

Rate

UM LTP Stock

Longterm Ready for

DeParking rate

Longterm parked

Vehicles

Longterm Parking

Percentage on Flash

Sale

UM LTCP Duration Min

UM LTCP CpDay

UM LTCP Cost Rate

UM LTCP assumed loss

Maximum In-flow Rate

Longterm Parking

UM UCL Inflow Rate

UM REMA Inflow Rate

Additional Volume from

Max Capacity Used Car

Leasing Rate

Additional Volume from

Max Capacity Longterm

Parking Rate

Additional Volume from

Max Capacity

Remarketing Rate

Overall additional

Volume from Max Cap

Swith Longterm Parking

Swith Used Car Leasing

Swith Drive Now

Swith Remarketing

Accumulated Activity

Max Cap

Accumulated Activity

Max Cap

Additional Volume

Dri veNow Max Cap

Additional Volume Used

Car Leasing Max Cap

Additional Volume

Longterm Parking Max

Cap

UM LTCP Outflow Rate

Longterm DeParking

Rate

Longterm Parking Ready for

DeparkingLongterm Standby

Deparking

Longterm Starndby

Deparking Rate

Swith Longterm Parking

Lots

UM LTCP Inflow Rate

UM LTCP Capacity Max

Drive Now Cost per km

Usage Duration per Day

Average speed

Yearly KM

Revenue per Usage Min

Revenue rate per Drive

Now Vehicle

Earnings Drive Now

Earning Drive Now

Average Depriciation in

Percent

UM DN Depreciation

UM DN Duration AVG

Remarketing rate

Additional Volume

Remarketing

Flash Sale volume

Flash Sale rate

FS sell-off durationMkt addtl vol max

FS impulse vol

FS impulse interval

FS impulse first

occurance

Minimum Time for

Reduction of Sales

Flash in

time cycle

FS impulse duration

one year only

tine cycle one year

cycle in

step in

input Flash

FS Switch 1 - 4

Off-lease Inflow

Auxiliary_1

Remarketing rate - 2

market price

Longterm Parking

Percentage on

Flash Sale

Apply pot Revenue

Market Price Delay

817161

22.000

23.000

24.000

25.000

EUR/vcl

indicated Price market price

V. Financial evaluation of

business model options

Presentation modelling: Based on the implementation of part ‘IV: Usage model distribution’, it is shown how the actual

distribution of vehicles on the usage models is conducted. This part serves as an example – more complex, and hence also less

understandable, modelling can be found throughout the simulation model. However, the simulation model itself is used to

show complexity of an issue, but never to discuss actual modelling or implementation.

II. Vehicle volumesI. ‘Fire sale’

IV. Usage model distribution

III. Usage models

Usage Model Weighting Mechanism

Accu mulated Activity

Remarketing

Percentage on Flash

Sale

DriveNow Percentage

on Flash Sale

UsedCar Percentage on

Flash Sale

Additional Volume from

Max Capacity Drive Now

Real Flash Sale

Additional Volume

ReMarketing Max Cap

Longterm Parking

Percentage on Flash

Sale

Additional Volume from

Max Capacity Used Car

Leasing Rate

Additional Volume from

Max Capacity Longterm

Parking Rate

Additional Volume from

Max Capacity

Remarketing Rate

Overall additional

Volume from Max Cap

Swith Longterm Parking

Swith Used Car Leasing

Swith Drive Now

Swith Remarketing

Accumulated Activity

Max Cap

Additional Volume

Dr iveNo w Max Cap

Additional Volume Used

Car Leasing Max Cap

Additional Volume

Longterm Parking Max

Ca p

Additional Volume

Remarketing

Fig. 2 Excerpt from the SD model of the Regulation case

System dynamics for corporate business model innovation 401

Appendix: Q-Methodology statistical data

Table 3 Q-Methodology

design choices Characteristic Choice/Value

Original data 16 statements

Forced distribution TRUE

Number of factors 4

Rotation varimax

Flagging automatic

Correlation coefficient pearson

Table 4 Abbreviations for Q-Sort statements

Abbreviated Statement Full statement: BMS helped me to…

EnvChanges recognize changes in the environment that will impact BMW’s business in the future

Threats recognize and assess external threats to BMW

Opportunities recognize and assess opportunities for BMW

Longer-termDynamics understand the longer-term dynamics of BMW’s existing business model

AdequacyOfStatusQuo critically reflect on the adequacy of BMW’s dominant “way of doing business”

Analogy understand how a business model from a different context / industry could be applied to BMW

ComponentCombination combine multiple components (e.g. a certain sales approach or a printer-and-ink revenue model) of existing

business models in a novel way

Heuristics leverage my professional experience at BMW to generate new business model ideas

CustomerDimension include the customer dimension in the business model properly

Complexity grasp the complexity of a business model

Communication communicate the simulated business model to stakeholders (e.g. involved employees, partners, decision makers)

Componentialization group the business model into relatively independent components

Conflicts identify areas of conflict between BMW’s current business model and potential configurations of BMW’s

future business model.

Credibility give my recommendations about business model changes more credibility within the organization

ImplementationActions formulate concrete actions for implementing a business model

ConfigurationComparison test and compare different possible configurations of a business model

Table 5 Original Q-Sort data

19: PO 12: M 20: IP 6: PO 9: PO 11: DM 3: PO 2: IP 4: PO 13: PO 14: PO 15: IP

EnvChanges 6 4 5 7 3 7 2 3 3 1 7 4

Threats 5 3 2 5 4 6 1 2 5 2 2 3

Opportunities 4 5 5 6 5 1 4 5 5 5 5 7

Longer-term dynamics 7 7 4 2 7 2 3 4 3 3 6 1

Adequacy of status quo 4 3 4 3 5 5 3 4 4 2 5 3

Analogy 3 1 1 1 1 3 4 2 6 3 5 6

T. Moellers et al.402

Open Access This article is distributed under the terms of the Creative

C o m m o n s A t t r i b u t i o n 4 . 0 I n t e r n a t i o n a l L i c e n s e ( h t t p : / /

creativecommons.org/licenses/by/4.0/), which permits unrestricted use,

distribution, and reproduction in any medium, provided you give

appropriate credit to the original author(s) and the source, provide a link

to the Creative Commons license, and indicate if changes were made.

Publisher’s note Springer Nature remains neutral with regard to jurisdic-

tional claims in published maps and institutional affiliations.

Table 8 Standard error

of differences between

factors

f1 f2 f3 f4

f1 0.47 0.43 0.43 0.47

f2 0.43 0.39 0.39 0.43

f3 0.43 0.39 0.39 0.43

f4 0.47 0.43 0.43 0.47

Table 9 Statement factor scores

f1 f2 f3 f4

EnvChanges 1 4 3 7

Threats 2 3 3 6

Opportunities 5 5 5 4

Longer-term dynamics 3 7 4 2

Adequacy of status quo 2 4 4 4

Analogy 3 1 2 1

Component combination 4 3 7 4

Heuristics 3 2 4 2

Customer dimension 4 3 5 5

Complexity 5 6 6 6

Communication 7 5 1 3

Componentialization 6 2 5 3

Conflicts 4 4 3 3

Credibility 4 6 2 5

Implementation actions 5 4 4 4

Configuration comparison 6 5 6 5

Table 5 (continued)

19: PO 12: M 20: IP 6: PO 9: PO 11: DM 3: PO 2: IP 4: PO 13: PO 14: PO 15: IP

Component Combination 5 3 5 4 2 4 3 7 6 4 2 2

Heuristics 4 2 3 2 4 3 4 4 2 3 6 4

Customer dimension 3 4 4 5 2 4 6 5 4 4 4 3

Complexity 5 6 7 6 6 4 6 6 7 5 4 5

Communication 6 5 4 4 3 3 5 1 2 7 4 6

Componentialization 2 2 3 4 3 2 5 5 4 6 3 2

Conflicts 1 4 2 3 5 4 2 3 4 4 4 5

Credibility 4 6 3 4 6 6 5 3 1 4 3 5

Implementation actions 3 4 6 3 4 5 4 4 3 5 1 4

Configuration comparison 2 5 6 5 4 5 7 6 5 6 3 4

Table 7 Correlation between factor z-scores

f1 z-score f2 z-score f3 z-score f4 z-score

f1 z-score 1.00 0.08 0.20 −0.19

f2 z-score 0.08 1.00 0.02 0.17

f3 z-score 0.20 0.02 1.00 0.06

f4 z-score −0.19 0.17 0.06 1.00

Table 6 General factor characteristics

av_rel_coef nload eigenvals expl_var reliability se_fscores

f1 0.8 2 2.5 21 0.89 0.33

f2 0.8 3 2.3 19 0.92 0.28

f3 0.8 3 1.9 16 0.92 0.28

f4 0.8 2 1.7 14 0.89 0.33

System dynamics for corporate business model innovation 403

References

Abdelkafi, N., & Täuscher, K. (2016). Business models for sustainability

from a system dynamics perspective. Organization & Environment,

29(1), 74–96.

Abdelkafi, N., Makhotin, S., & Posselt, T. (2013). Business model inno-

vations for electric mobility — what can be learned from existing

business model patterns? International Journal of Innovation

Management, 17(1), 1–41.

Afuah, A., & Tucci, C. L. (2003). Internet business models & strategies:

Text and cases (2nd ed.). New York: McGraw-Hill.

Akkermans, H., Baida, Z., Gordijn, J., Peña, N., Altuna, A., & Laresgoiti,

I. (2004). Value webs: using ontologies to bundle real-world ser-

vices. IEEE Intelligent Systems, 19(4), 57–66.

Al-Debei, M. M., & Avison, D. (2010). Developing a unified framework

of the business model concept. European Journal of Information

Systems, 19(3), 359–376.

Ali-Vehmas, T., & Casey, T. R. (2012). Evolution of wireless access

provisioning: a systems thinking approach. Competition and

Regulation in Network Industries, 13(4), 333–361.

Amit, R., & Zott, C. (2012). Creating value through business model

innovation. MIT Sloan Management Review, 53(3), 40–50.

Andries, P., Debackere, K., & van Looy, B. (2013). Simulatenous exper-

imentation as a learning stratgy: business model development under

uncertainty. Strategic Entrepreneurship Journal, 7(1), 27–47.

Aspara, J., Hietanen, J., & Tikkanen, H. (2010). Business model innova-

tion vs replication: financial performance implications of strategic

emphases. Journal of Strategic Marketing, 18(1), 39–56.

Aspara, J., Lamber, J.-A., Laukia, A., & Tikkanen, H. (2011). Strategic

management of business model transformation: lessons from Nokia.

Management Decision, 49(4), 622–647.

Aversa, P., Haefliger, S., Rossi, A., & Baden-Fuller, C. (2015). From

business model to business modelling: modularity and manipula-

tion. Advances in Strategic Management, 33, 151–185.

Aversa, P., Haefliger, S., & Reza, D. G. (2017). Building a winning

business model portfolio. MIT Sloan Management Review, 58(4),

47–54.

Baden-Fuller, C., & Haefliger, S. (2013). Business models and techno-

logical innovation. Long Range Planning, 46(6), 419–426.

Baden-Fuller, C., & Mangematin, V. (2013). Business models: a chal-

lenging agenda. Strategic Organization, 11(4), 418–427.

Baden-Fuller, C., & Morgan, M. S. (2010). Business models as models.

Long Range Planning, 43(2–3), 156–171.

Barabba, V., Huber, C., Cooke, F., Pudar, N., Smith, J., Paich, M., …

Cooke, F. (2002). A multimethod approach for creating new busi-

ness models: the general motors OnStar project. Interfaces, 31(1),

20–34.

Bartel, C. A., & Garud, R. (2009). The role of narratives in sustaining

organizational innovation. Organization Science, 20(1), 107–117.

Berends, H., Smits, A., Reymen, I., & Podoynitsyna, K. (2016). Learning

while (re)configuring: business model innovation processes in

established firms. Strategic Organization, 14(3), 181–219.

Bingham, C. B., & Eisenhardt, K. M. (2011). Rational heuristics: the

“simple rules” that strategists learn from process experience.

Strategic Management Journal, 32(13), 1437–1464.

Blank, S. (2013). Why the lean start-up changes everything. Harvard

Business Review, 91(5), 63–72.

Bouwman, H., de Vos, H., & Haaker, T. (2008). Mobile service innova-

tion and business models. Berlin Heidelberg: Springer-Verlag.

Casadesus-Masanell, R., & Ricart, J. E. (2010). From strategy to business

models and onto tactics. Long Range Planning, 43, 195–215.

Casadesus-Masanell, R., Gassmann, O., & Sauer, R. (2017a). Hilti (A):

Fleet management? Harvard Business School Press (pp. 1–19).

Casadesus-Masanell, R., Gassmann, O., & Sauer, R. (2017b). Hilti Fleet

management (B): Towards a new business model. Harvard Business

School Press (pp. 1–9).

Cavalcante, S., Kesting, P., & Ulhøi, J. (2011). Business model dynamics

and innovation: (re)establishing the missing linkages. Management

Decision, 49(8), 1327–1342.

Charmaz, K. (2014). Constructing grounded theory: A practical guide

through qualitative analysis (2nd ed.). Thousand Oaks: Sage

Publication.

Chesbrough, H. (2010). Business model innovation: opportunities and

barriers. Long Range Planning, 43(2–3), 354–363.

Chesbrough, H., & Rosenbloom, R. S. (2002). The role of the business

model in capturing value from innovation: evidence from Xerox

Corporation’s technology spin-off companies. Industrial and

Corporate Change, 11(3), 529–555.

Cosenz, F. (2017). Supporting start-up business model design through

system dynamics modelling. Management Decision, 55(1), 57–80.

Cross, R. M. (2005). Exploring attitudes: the case for Q methodology.

Health Education Research, 20(2), 206–213.

Cucculelli, M., & Bettinelli, C. (2015). Business models, intangibles and

firm performance: evidence on corporate entrepreneurship from

Italian manufacturing SMEs. Small Business Economics, 45(2),

329–350.

Cyert, R. M., & March, J. G. (1963). A behavioral theory of the firm.

Englewood Cliffs: Prentice-Hall.

de Reuver, M., Bouwman, H., & MacInnes, I. (2009). Business models

dynamics for start-ups and innovating e-businesses. International

Journal of Electronic Business, 7(3), 269–286.

de Reuver, M., Bouwman, H., & Haaker, T. (2013). Business model

roadmapping: a practical approach to come from an existing to a

desired business model. International Journal of Innovation

Management, 17(1), (134006-)1–18.

Dellermann, D., Lipusch, N., Ebel, P., & Leimeister, J. M. (2018). Design

principles for a hybrid intelligence decision support system for busi-

ness model validation. Electronic Markets. https://doi.org/10.1007/

s12525-018-0309-2

Demil, B., & Lecocq, X. (2010). Business model evolution: in search of

dynamic consistency. Long Range Planning, 43(2–3), 227–246.

Demil, B., & Lecocq, X. (2015). Crafting an innovative business model in

an established company: the role of artifacts. Advances in Strategic

Management, 33, 31–58.

Desyllas, P., & Sako, M. (2013). Profiting from business model innova-

tion: evidence from Pay-As-You-Drive auto insurance. Research

Policy, 42(1), 101–116.

Doganova, L., & Eyquem-Renault, M. (2009). What do business models

do?. Innovation devices in technology entrepreneurship. Research

Policy, 38(10), 1559–1570.

Doz, Y. L., & Kosonen, M. (2010). Embedding strategic agility: a lead-

ership agenda for accelerating business model renewal. Long Range

Planning, 43(2–3), 370–382.

Eisenhardt, K. M. (1989). Building theories from case study research.

Academy of Management Review, 14(4), 532–550.

Eisenhardt, K. M., & Graebner, M. E. (2007). Theory building from

cases: opportunities and challenges. Academy of Management

Journal, 50, 25–32.

Euchner, J., & Ganguly, A. (2014). Business model innovation in prac-

tice. Research-Technology Management, 57(6), 33–39.

Fang, E. (. E.)., Palmatier, R. W., & Steenkamp, J.-B. E. (2008). Effect of

service transition strategies on firm value. Journal of Marketing,

72(5), 1–14.

Featherston, C. R., & Doolan, M. (2012). A critical review of the criti-

cisms of system dynamics. In The 30th International Conference of

the System Dynamics Society (pp. 1–13).

Flick, U. (2009). An introduction to qualitative research (4th ed.).

Thousand Oaks: Sage Publication.

Forrester, J. W. (1961). Industrial dynamics. Cambridge: MIT Press.

T. Moellers et al.404

Forrester, J. W. (2007). System dynamics—The next fifty years. System

Dynamics Review, 23(2/3), 359–370.

Foss, N. J., & Saebi, T. (2017). Fifteen years of research on business

model innovation: how far have we come, and where should we

go? Journal of Management, 43(1), 200–227.

França, C. L., Broman, G., Robèrt, K. H., Basile, G., & Trygg, L. (2017).

An approach to business model innovation and design for strategic

sustainable development. Journal of Cleaner Production, 140, 155–

166.

Frankenberger, K., Weiblen, T., Csik, M., & Gassmann, O. (2013). The

4I-framework of business model innovation: a structured view on

process phases and challenges. International Journal of Product

Development, 18(3/4), 249–273.

Furnari, S. (2015). A cognitive mapping approach to business models:

representing causal structures and mechanisms. Advances in

Strategic Management, 33, 1–40.

Galanakis, K. (2006). Innovation process. Make sense using systems

thinking. Technovation, 26(11), 1222–1232.

Gambardella, A., & McGahan, A. M. (2010). Business-model innova-

tion: general purpose technologies and their implications for indus-

try structure. Long Range Planning, 43(2–3), 262–271.

Gary, M. S., Kunc, M., Morecroft, D. W., & Gary, M. S. (2009). System

dynamics and strategy. System Dynamics Review, 24(4), 407–429.

Gassmann, O., Frankenberger, K., & Csik, M. (2014). The business mod-

el navigator: 55 models that will revolutionize your business.

Harlow Pearson, UK: FT Press.

Gavetti, G. (2012). PERSPECTIVE — toward a behavioral theory of

strategy toward a behavioral theory of strategy. Organization

Science, 21(1), 267–285.

Gavetti, G., & Levinthal, D. (2000). Looking forward and looking back-

ward: cognitive and experiential search. Administrative Science

Quarterly, 45(1), 113–137.

Gebauer, H., Haldimann, M., & Saul, C. J. (2017). Business model inno-

vations for overcoming barriers in the base-of-the-pyramid market.

Industry and Innovation, 24(5), 543–568.

Gentner, D. (1983). Structure-mapping: a theoretical framework for anal-

ogy. Cognitive Science, 7, 155–170.

Gibbert, M., & Ruigrok, W. (2010). The “what” and “how” of case study

rigor: three strategies based on published work. Organizational

Research Methods, 13(4), 710–737.

Gibbert, M., Ruigrok, W., & Wicki, B. (2008). What passes as a rigorous

case study? Strategic Management Journal, 29(13), 1465–1474.

Gilbert, C. G. (2005). Unbundling the structure of inertia: resource versus

routine rigidity. Academy of Management Journal, 48(5), 741–763.

Gioia, D. A., Corley, K. G., & Hamilton, A. L. (2012). Seeking qualitative

rigor in inductive research: notes on the Gioia methodology.

Organizational Research Methods, 16(1), 15–31.

Goodman, N. (1954). Fact, fiction, and forecast. Cambridge: Harvard

University Press.

Gordijn, J., & Akkermans, J. M. (2003). Value-based requirements engi-

neering: exploring innovative e-commerce ideas. Requirements

Engineering, 8(2), 114–134.

Graham, A. K., Morecroft, J. D. W., Senge, P. M., & Sterman, J. D.

(1992). Model-supported case studies for management education.

European Journal of Operational Research, 59(1), 151–166.

Groesser, S. N., & Jovy, N. (2016). Business model analysis using com-

putational modeling a strategy tool for exploration and decision-

making. Journal of Management Control, 27(1), 61–88.

Groesser, S. N., & Schwaninger, M. (2012). Contributions to model val-

idation: hierarchy, process, and cessation. System Dynamics Review,

8(2), 157–181.

Holland, J. H., Holyoak, K. J., Nisbett, R. E., & Thagard, P. (1986).

Induction: Processes of inference, learning, and discovery.

Cambridge: MIT Press.

Huber, G. P. (1985). Temporal stability and response-order biases in par-

ticipant descriptions of organizational decisions. Academy of

Management Journal, 28(4), 943–950.

Hwang, J., & Christensen, C. M. (2008). Disruptive innovation in health

care delivery: a framework for business-model innovation. Health

Affairs, 27(5), 1329–1335.

IBM. (2006). Expanding the Innovation Horizon: The Global CEO Study

2006, 1–61.

Jick, T. D. (1979). Mixing qualitative and quantitative methods: triangu-

lation in action. Administrative Science Quarterly, 24(4), 602–611.

Kaplan, S. (2012). The business model innovation factory: How to stay

relevant when the world is changing. Hoboken: Wiley.

Karimi, J., & Walter, Z. (2016). Corporate entrepreneurship, disruptive

business model innovation adoption, and its performance: the case

of the newspaper industry. Long Range Planning, 49(3), 342–360.

Kastalli, I. V., Van Looy, B., & Neely, A. (2013). Steering manufacturing

firms towards service business model innovation. California

Management Review, 56(1), 100–123.

Ketokivi, M., & Mantere, S. (2010). Two strategies for inductive reason-

ing in organizational research. Academy of Management Review,

35(2), 315–333.

Kim, S. K., & Min, S. (2015). Business model innovation performance:

when does adding a new business model benefit an incumbent?

Strategic Entrepreneurship Journal, 9, 34–57.

Klang, D., Wallnöfer, M., & Hacklin, F. (2014). The business model

paradox: a systematic review and exploration of antecedents.

International Journal of Management Reviews, 16(4), 454–478.

Köpp, S., & Schwaninger, M. (2014). Scrutinizing the Sustainability of

Business Models: System Dynamics for Robust Strategies, 1–55.

Lakoff, G. (1987). Women, fire and dangerous things: What categories

reveal about the mind. Chicago: University of Chicago Press.

Leonard-Barton, D. (1992). Core capabilities and Core rigidities: a para-

dox in managing new product development. Strategic Management

Journal, 13, 111–125.

Levitt, B., & March, J. G. (1988). Organizational learning. Annual

Review of Sociology, 14, 319–340.

Lippman, S. A., & McCall, J. J. (1976). The economics of job search: a

survey. Economic Inquiry, 14, 155–189.

Loock, M., & Hacklin, F. (2015). Business modeling as configuring heu-

ristics. Business Models and Modelling, 33, 1–19.

Lucas, H. C., & Goh, J. M. (2009). Disruptive technology: how Kodak

missed the digital photography revolution. Journal of Strategic

Information Systems, 18(1), 46–55.

Lyneis, J. M. (1999). System dynamics for business strategy: a phased

approach. System Dynamics Review, 15(1), 37–71.

Maher, P. (1998). Inductive inference. In E. Craig (Ed.), Routledge ency-

clopedia of philosophy (pp. 755–759). London: Routledge.

Markides, C. C. (2015). Research on business models: challenges and

opportunities. Advances in Strategic Management, 33, 1–16.

Martignoni, D., Menon, A., & Siggelkow, N. (2016). Consequences of

misspecified mental models: contrasting effects and the role of cog-

nitive fit. Strategic Management Journal, 37(13), 2545–2568.

Martinez-Moyano, I. J., & Richardson, G. P. (2013). Best practices in

system dynamics modeling. System Dynamics Review, 29(2), 102–

123.

Martins, L. L., Rindova, V. P., & Greenbaum, B. E. (2015). Unlocking the

hidden value of concepts: a cognitive approach to business model

innovation. Strategic Entrepreneurship Journal, 9(1), 99–117.

Massa, L., & Tucci, C. L. (2014). Business model innovation. In M.

Dodgson, D. M. Gann, & N. Phillips (Eds.), The Oxford handbook

of innovation management (pp. 420–441). Oxford, UK: Oxford

University Press.

Massa, L., Tucci, C., & Afuah, A. (2017). A critical assessment of busi-

ness model research. Academy of Management Annals, 11(1), 73–

104.

System dynamics for corporate business model innovation 405

Matzler, K., Bailom, F., von den Eichen, S. F., & Kohler, T. (2013).

Business model innovation: coffee triumphs for Nespresso.

Journal of Business Strategy, 34(2), 30–37.

McCall, M. W., & Kaplan, R. E. (1985). Whatever it takes: Decision

makers at work. Upper Saddle River: Prentice Hall.

McGrath, R. G. (2010). Business models: a discovery driven approach.

Long Range Planning, 43(2–3), 247–261.

Mezger, F. (2014). Toward a capability-based conceptualization of busi-

ness model innovation: insights from an explorative study. R&D

Management, 44(5), 429–449.

Miller, C. C., Cardinal, L. B., & Glick, W. H. (1997). Retrospective

reports in organizational research: a reexamination of recent evi-

dence. Academy of Management Journal, 40(1), 189–204.

Moellers, T., Bansemir, B., Pretzl, M., & Gassmann, O. (2017). Design

and evaluation of a system dynamics based business model evalua-

tion method. DESRIST 2017. Lecture Notes in Computer Science,

10243, 125–144.

Nadkarni, S., & Barr, P. S. (2008). Envirnonmental context, managerial

cognition, and strategic action: an integrated view. Strategic

Management Journal, 29, 1395–1427.

O’Reilly, C. A., III, Chatman, J., & Caldwell, D. F. (1991). People and

organizational culture: a profile comparison approach to assessing

person-organization fit. Academy of Management Journal, 34(3),

487–516.

Onggo, S., Soopramanien, D., & Pidd, M. (2006). A dynamic business

model for component-based simulation software. In Proceedings of

the 2006 Winter Simulation Conference (pp. 954–959).

Osiyevskyy, O., & Dewald, J. (2015). Explorative versus exploitative

business model change: the cognitive antecedents of firm-level re-

sponses to disruptive innovation. Strategic Entrepreneurship

Journal, 9, 58–78.

Osterwalder, A., & Pigneur, Y. (2010). Business model generation: A

handbook for visionaries, game changers, and challengers.

Hoboken: Wiley.

Pagani, M. (2009). Roadmapping 3G mobile TV: strategic thinking and

scenario planning through repeated cross-impact handling.

Technological Forecasting and Social Change, 76(3), 382–395.

Petrovic, O., Kittl, C., & Teksten, R. D. (2001). Developing Business

Models for Ebusiness. SSRN Electronic Journal, 1–6.

Porac, J. F., Thomas, H., & Baden-Fuller, C. (1989). Competitive groups

as cognitive communities: the case of Scottish knitwear manufac-

turers. Journal of Management Studies, 26(4), 397–416.

Prahalad, C. K., & Bettis, R. A. (1986). The dominant logic: a new

linkage between diversity and performance. Strategic Management

Journal, 7(6), 485–501.

Prahalad, C. K., & Bettis, R. A. (1995). The dominant logic: retrospective

and extension. Strategic Management Journal, 16(1), 5–14.

Richter, M. (2013). Business model innovation for sustainable energy:

German utilities and renewable energy. Energy Policy, 62, 1226–

1237.

Royston, G., Dost, A., Townshend, J., & Turner, H. (1999). Using system

dynamics to help develop and implement policies and programmes

in health care in England. System Dynamics Review, 15(3), 293–313.

Rumble, R., & Mangematin, V. (2015). Business model implementation:

the antecedents of multi-sidedness. Business Models and Modelling,

33, 97–131.

Sabatier, V., Mangematin, V., & Rousselle, T. (2010). From recipe to

dinner: business model portfolios in the european biopharmaceutical

industry. Long Range Planning, 43(2–3), 431–447.

Saebi, T., Lien, L., & Foss, N. J. (2016). What drives business model

adaptation? The impact of opportunities, threats and strategic orien-

tation. Long Range Planning, 50(5), 567–581.

Sarasvathy, S. D. (2001). Causation and effectuation: toward a theoretical

shift from economic inevitability to entrepreneurial contingency.

Academy of Management Review, 26(2), 243–263.

Schwaninger, M. (2006). System dynamics and the evolution of the sys-

tems movement. Systems Research and Behavioral Science, 23,

583–594.

Siggelkow, N. (2007). Persuasion with case studies. Academy of

Management Review, 50(1), 20–24.

Simon, H. A. (1957). Administrative behavior: A study of decision-

making processes in administrative organizations. New York:

Macmillan.

Sosna, M., Trevinyo-Rodríguez, R. N., & Velamuri, S. R. (2010).

Business model innovation through trial-and-error learning: the

naturhouse case. Long Range Planning, 43(2–3), 383–407.

Souto, J. E. (2015). Business model innovation and business concept

innovation as the context of incremental innovation and radical in-

novation. Tourism Management, 51, 142–155.

Stephenson, W. (1936). The inverted factor technique. British Journal of

Psychology, 26(4), 344–361.

Stephenson, W. (1953). The study of behavior; Q-technique and its

methodology. Chicago: University of Chicago Press.

Sterman, J. D. (2000). Systems thinking and modeling for a complex

world. Boston: McGraw-Hill.

Stern, C. W., & Deimler, M. S. (2006). The Boston consulting group on

strategy. Hoboken: Wiley.

Stigliani, I., & Ravasi, D. (2012). Organizing thoughts and connecting

brains: material practices and the transition from individual to

group-level prospective sensemaking. Academy of Management

Journal, 55(5), 1232–1259.

Suarez, F., Cusumano, M., & Kahl, S. (2013). Service and the business

models of product firms: an empirical analysis of the software in-

dustry. Management Science, 59(2), 420–435.

Tapscott, D., & Tapscott, A. (2016). Blockchain revolution. New York:

Penguin.

Täuscher, K., & Abdelkafi, N. (2017). Visual tools for business model

innovation: recommendations from a cognitive perspective.

Creativity and Innovation Management, 26(2), 160–174.

Teece, D. J. (2010). Business models, business strategy and innovation.

Long Range Planning, 43(2–3), 172–194.

Tripsas, M., & Gavetti, G. (2000). Capabilities, cognition, and inertia:

evidence from digital imaging. Strategic Management Journal,

21(10), 1147–1161.

Visnjic, I., Wiengarten, F., & Neely, A. (2016). Only the brave: product

innovation, service business model innovation, and their impact on

performance. Journal of Product Innovation Management, 33(1),

36–52.

Walsh, J. P. (1995). Managerial and organizational cognition: notes from

a trip down memory lane. Organization Science, 6(3), 280–321.

Watts, S., & Stenner, P. (2005). Doing Q Methodology: theory, method

and interpretation. Qualitative Research in Psychology, 2(1), 67–91.

Weick, K. E. (1979). The social psychology of organizing (2nd ed.).

Reading: Addison-Wesley.

Wisniewski, E. J. (1997). When concepts combine. Psychonomic Bulletin

and Review, 4(2), 167–183.

Yin, R. K. (2014). Case study research: Design and methods (5th ed.).

Thousand Oaks: Sage.

Zoric, J. (2011). Connecting business models with service platform de-

signs: exploiting potential of scenario modeling. Telematics and

Informatics, 28(1), 40–54.

T. Moellers et al.406

© 2019. This work is published under

http://creativecommons.org/licenses/by/4.0/(the “License”). Notwithstanding

the ProQuest Terms and Conditions, you may use this content in accordance

with the terms of the License.