The evolution of innovation strategy: Studied in the context of medical device activities at the pharmaceutical company Novo Nordisk A/S in the period 1980-2008

Stjernholm Madsen, Arne Doctoral Thesis The evolution of innovation strategy: Studied in the context of medical device activities at the pharmaceutical company Novo Nordisk A/S in the period 1980-2008 PhD Series, No. 21.2012 Provided in Cooperation with: Copenhagen Business School (CBS) Suggested Citation: Stjernholm Madsen, Arne (2012) : The evolution of innovation strategy: Studied in the context of medical device activities at the pharmaceutical company Novo Nordisk A/S in the period 1980-2008, PhD Series, No. 21.2012, ISBN 9788792842695, Copenhagen Business School (CBS), Frederiksberg, https://hdl.handle.net/10398/8453 This Version is available at: https://hdl.handle.net/10419/208817 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by-nc-nd/3.0/ Arne Stjernholm Madsen The PhD School of Economics and Management PhD Series 21.2012 PhD Series 21.2012 The evolution of innovation strategy copenhagen business school handelshøjskolen solbjerg plads 3 dk-2000 frederiksberg danmark www.cbs.dk ISSN 0906-6934 Print ISBN: 978-87-92842-68-8 Online ISBN: 978-87-92842-69-5 The evolution of innovation strategy Studied in the context of medical device activities at the pharmaceutical company Novo Nordisk A/S in the period 1980-2008 1 The evolution of innovation strategy Studied in the context of medical device activities at the pharmaceutical company Novo Nordisk A/S in the period 1980-2008 Arne Stjernholm Madsen PhD thesis. Date of submission: 2012-03-16. Supervisor: Professor Jens Frøslev Christensen, Department of Innovation & Organizational Economics (INO). Copenhagen Business School, Doctoral School of Organisation and Management Studies. Arne Stjernholm Madsen The evolution of innovation strategy Studied in the context of medical device activities at the pharmaceutical company Novo Nordisk A/S in the period 1980-2008 1st edition 2012 PhD Series 21.2012 © The Author ISSN 0906-6934 Print ISBN: 978-87-92842-68-8 Online ISBN: 978-87-92842-69-5 “The Doctoral School of Economics and Management is an active national and international research environment at CBS for research degree students who deal with economics and management at business, industry and country level in a theoretical and empirical manner”. All rights reserved. No parts of this book may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or by any information storage or retrieval system, without permission in writing from the publisher. 3 Preface and acknowledgements This DBA (Doctor of Business Administration) project has been underway for six years. Although working in the field of innovation management, my background was an education as a graphic designer and I therefore had to start by writing a research proposal, in order to obtain dispensation from CBS for the lack of academic degree. Luckily, I succeeded, and thus I could begin the adventures of academic research by yearend 2006. Having my normal work to do on the side, and trying to maintain rudiments of a private life, the DBA project sometimes felt like being in a maelstrom of challenges. So much greater was the excitement, when the pieces of the puzzle began to form a picture – such peak experience makes the whole journey worth the efforts. A number of persons and institutions have contributed to making this research project possible. I would first like to thank my employer, Novo Nordisk A/S, for generous financial support and for letting me use the company as a case. Furthermore, app. 50 former and present colleagues have served as my informants – these remain anonymous, in accordance with the corporate policies. However, I think it is fair explicitly to thank the three executive informants: Former CEO Mads Øvlisen, CSO Mads Krogsgaard Thomsen and CEO Lars Rebien Sørensen. Without the numerous internal informants at Novo Nordisk, there would have been no story to tell… Several academic researchers and educators have supported me. First of all I wish to thank my supervisor, Professor Jens Frøslev Christensen, for believing in my project and guiding me patiently through my ups and downs. “I can see the diamond in the coalmine”, he persistently encouraged me. Along the way I have also received valuable input from Professor John Bessant, University of Exeter Business School, who served as external supervisor in the beginning of the project. I have received helpful advice from Professor Richard A. Bettis, The Kenan-Flagler Business School, University of North Carolina at Chapel Hill, at a PhD course at CBS in 2009. Professor Andrew Van de Ven, Carlson School of Management, University of Minnesota, opened my eyes for the principles of engaged scholarship at the EDEN doctoral seminar in 2011. For the theoretical frame, I am in debt to Professor Robert A. Burgelman, Stanford University Graduate School of Business, who has also kindly commented my diagrams of his theories. Professors Christian Knudsen and Peter Karnøe, both CBS, gave very constructive feedback at my pre-defense. Last, not least, I thank the dissertation committee for very thorough and constructive comments, which have sharpened the argumentation in the final text: Professors Christian Knudsen (Chairman) CBS; Fredrik Tell, Linköping University; Robert A. Burgelman, Stanford University. I owe all of these scholars a lot. On the private side, I have been gratified by seeing both my children undertake PhD projects and thereby encourage their father: Tabita, my daughter, took her PhD from The Niels Bohr Institute, University of Copenhagen, in 2009 and Mathias, my son, currently conducts a PhD project at Institute for Logic, Language, and Computation at the University of Amsterdam. The one person, however, that I am most in debt to, is my wife, Kirsten, who has supported me and my project throughout the last six years, regardless how stressed or absentminded I have been. 4 5 Summary Increased globalization in business competition makes the ability to innovate and to redefine strategy crucial to a company. An interesting question however is if a management team can control innovation and strategic renewal of the company at all; or do such changes emerge, driven by external events or by bottom-up processes in the organization? The present research project addresses some of these issues through the overall research question “How does innovation strategy evolve?” The research question is examined in a specific empirical context. Since 2001, I have worked as an internal innovation consultant at Novo Nordisk A/S; a pharmaceutical firm founded in 1923 operating in a well established industry (insulin for diabetes treatment), characterized by intensive investments in Research and Development. I took advantage of this unique access to the internal life of an organization and consequently set up my research project as a longitudinal in-depth case study of the medical device innovation activities at Novo Nordisk A/S covering the period 1980-2008. The study specifically analyzes the relationship between the classic core product of the firm (insulin) and complementary products (medical devices, such as insulin ‘pens’), which hold the potential to either enhance the value of the core product, or to become a distinct business of its own. Burgelman’s evolutionary theory of strategy making, especially his ‘internal ecology model’ (Burgelman 1991, 2002), has been chosen as the basic theoretical framework for the project. Some expansions of this framework, however, were needed. First, the present study puts greater emphasis on analyzing the external environment and its influence on internal strategy processes. Second, the analysis includes the role of management cognition, especially the notion of the corporate dominant logic (Prahalad & Bettis, 1986; Bettis & Prahalad, 1995), understood as an enduring top management worldview or mindset based on reinforcement of experiences from the past. With regard to results, the present study identifies a more entrepreneurial role of the top management driven induced strategy process than traditionally described in evolutionary theory. In this case study, strategic variation and trial-and-error learning is not restricted to the autonomous initiatives in the ‘internal ecology’; on the contrary, top management cognition creates strategic visions or hypotheses, which are enacted as experiments in the market, for example in the form of new product categories. External feedback determines the destiny of these strategic experiments. Thereby innovation strategy (in case, for medical devices) serves as a strategic laboratory at corporate level, so to speak. The device-based strategic experiments face the challenge of escaping the gravity of the dominant logic, which repeatedly pulls the strategy back towards the well-known success formula, centered on the drug itself (i.e. the insulin). Thus, the induced strategy process mediates core assets (pharmaceutical drugs) and complementary assets (medical devices), by swinging the pendulum between cycles of innovation strategy which define the devices as core or complementary 6 respectively. Hence, the balance between what is defined as core and what is defined as complementary in the corporate innovation strategy seems to be dynamic and negotiable. As a consequence of the cycles of strategic experimentation, the corporate induced strategy process acts as a force of strategic entrepreneurship, seen over extended time. The implications for research point towards a new paradigm of strategic research in the ‘middle ground’ between rational choice theory and evolutionary theory, as proposed by Gavetti & Levinthal (2004). The present research project suggests that a firm’s ability for strategic adaptation depends both on strategic context determination of autonomous initiatives in the ‘internal ecology’ and on ability to enact induced strategic experiments with alternating innovation strategies in the market. This theory of ‘inbound’ and ‘outbound’ strategic search establishes a dynamic understanding of the corporate induced strategy process. In this understanding, innovation strategies act as hypotheses, which create strategic dissonance between vision and reality and thereby drive strategic learning. The implications for management practice are first recognition of how fortunate it has been for Novo Nordisk to sustain the core business strategy, protected by the dominant logic. This fact relates to a background where the core market proved to hold immense growth potential, and the industry was relatively stable compared to for instance the IT industry. On the other hand, Novo Nordisk’s success is partly due to cycles of strategic experiments with complementary assets for innovation, in case medical devices. Top management initiated these explorative experiments and the learning was utilized for expansion of the position within the core business. Hence, one can conclude that a company should explore and utilize the value of complementary assets, since these are perfect tools for strategic experimentation without risking the core business. 7 Resumé I den stigende globale konkurrence har virksomheders evne til innovation og til at redefinere deres forretning fået livsvigtig betydning. Men kan man som ledelse overhovedet styre innovation og strategisk fornyelse, eller opstår fornyelsen organisk, så at sige af sig selv, drevet af ydre omstændigheder eller nedefra i organisationen? Dette forskningsprojekt indkredser en del af denne problematik gennem den overordnede problemformulering “Hvordan udvikler innovationsstrategi sig?” Problemstillingen undersøges empirisk i en specifik kontekst. Siden 2001 har jeg arbejdet som intern innovationskonsulent i Novo Nordisk A/S; et firma etableret i 1923, som opererer i den forskningstunge farmaceutiske industri med insulin til diabetesbehandling som sit hovedprodukt. Jeg har valgt at udnytte min unikke tilgang til organisationens indre liv ved at udforme forskningsprojektet som et dybdegående langtids-casestudie i konteksten udvikling af medicinske devices (dvs. insulinpenne og lignende udstyr) hos Novo Nordisk A/S i perioden 1980-2008. Specielt analyseres relationen mellem devices som komplementære produkter i forhold til virksomhedens klassiske kerneprodukt (insulin): Understøtter devices salget af kerneproduktet, eller udnyttes devices til selvstændig forretningsudvikling? Som teoretisk ramme er primært valgt Robert A. Burgelmans evolutionære teori om strategiudvikling kaldet strategiudviklingens interne økologi (Burgelman 1991, 2002). Desuden er inddraget nogle supplerende perspektiver, som Burgelmans teori gør mindre ud af: For det første er der lagt større vægt på omverdenens indflydelse på organisationens strategiudvikling; for det andet inddrages ledelseskognition i analysen, herunder begrebet dominerende logik (Prahalad & Bettis, 1986; Bettis & Prahalad, 1995), der betegner et relativt statisk ledelsessyn på virksomhedens succesfaktorer, som er opbygget gennem erfaring. Som resultat påviser casestudiet en mere innovativ rolle for den topstyrede, inducerede strategiproces, end man normalt ser beskrevet i evolutionær ledelsesteori. I mit studie er strategisk fornyelse og prøven-sig-frem ikke begrænset til de såkaldte autonome initiativer i den interne økologi; tværtimod fører topledelsens kognition til eksperimenter med strategiske visioner eller hypoteser, der prøves af i markedet f.eks. med introduktion af nye produktkategorier. Reaktionen fra omverdenen bestemmer disse eksperimenters skæbne. Innovationsstrategien kan altså siges at være en slags strategisk laboratorium for virksomheden som helhed. De strategiske eksperimenter med devices har desuden den konstante udfordring, at den dominerende logik trækker strategien tilbage imod den velkendte formel, baseret på selve insulinen som forretningens basis. Således balancerer strategiudviklingen mellem kerneprodukter (i dette tilfælde insulin) og komplementære produkter (devices), som på skift danner basis for vekslende innovationsstrategier. Dermed synes grænsen mellem kerne og komplementær at være dynamisk eller m.a.o. til konstant forhandling. 14 the vision had died. – Why? Didn’t top management know what they were doing when they launched PDS? As employees, we got no real explanation for the turbulence in innovation strategy. I didn’t expect such turbulence, since Novo Nordisk was a very mature company (founded 1923), acting in a relatively stable industry (product lifecycles of typically 20 years), based on a tradition of science and intensive R&D investments – not the kind of company, from which you expect impulsive fluctuations in the innovation strategy. Furthermore, when I began examining the history of Novo Nordisk, I discovered that this was actually the second time that such visionary innovation strategy had bloomed and withered. At a much earlier stage in 1988, a division was established for medical devices, called Medical Systems Division (MSD), on a similar vision of full circle homecare for the patient, based on medical devices including devices for blood sugar monitoring. In 1988, the technologies were not mature for envisioning an automatic ‘closed loop’ system – the patient would still have to do everything manually. But the mere vision of providing the patient with all the drugs and devices, he/she needed for taking care of the disease at home (not at the hospital) was very progressive. However, MSD also had a lifetime of only four years, followed by a strategic turnaround in 1992: focus now should solely be on injection devices, in order to support the pharmaceutical drugs. During my research, I’ve talked with many veteran colleagues about the evolution of the medical device activities at Novo Nordisk, and it was a great eye-opener for me, when a manager from the device area, who had been part of this venture for two decades, first described the development and then concluded: “It’s really funny to see how we started in MSD by diverging – then we converged again – then came PDS, where we went off at countless tangents – we even included our inhalable insulin project, that was really extreme – and now we have narrowed down again and have become focused again; our area must secure that the next generation prefilled [insulin pen] simply gets launched. That is our key function within Novo Nordisk today. It’s quite funny – it has all happened within 20 years” (interviewed in 2008). First, I should explain the use of the words ‘divergent’ and ‘convergent’: in this context, these concepts refer to the latitude of the device innovation activities. By “diverging” the manager means expanding the scope of the activities through adding new areas; by “converging” he means contracting the scope of activities to fewer areas. The point, however, is that this manager sees the MSD period from 1988 to 1992 as similar to the PDS period from 2001 to 2005: both periods were “divergent”. In contrast, the period between MSD and PDS (1992-2001) is likened with the period after 2005: in these periods the activities “converged”. When I dug more into the history of the medical device activities at Novo Nordisk, it struck me that the fluctuations in strategy, as outlined in the quotation above, contrasted the actual development in the market. Until 1985, insulin was sold in vials, to be injected with traditional syringes. Novo Nordisk introduced the first injection device, the NovoPen®, in 1985. The NovoPen® was well received by the market and was followed by the introduction of several other ‘insulin pens’, as these devices were called (because they imitated the look of a fountain pen). The insulin pens contained the insulin in a customized insulin cartridge, and therefore the market penetration of the devices can be measured by the volume of insulin sold in these cartridges compared to the total volume of insulin 15 sold. Since 1985, insulin sold in cartridges customized for devices has steadily increased market share, compared to insulin sold in traditional vials, and today (2012) accounts for nearly two thirds of all insulin sold globally (source: internal Novo Nordisk statistics). This development has been initiated and driven by Novo Nordisk. In other words, the introduction of insulin devices (‘insulin pens’) has gradually changed the market towards the new systems, displaying a steady growth. Metaphorically, the market impact of the new insulin devices since 1985 has slowly increased like a train, which mile for mile crosses the landscape, following the laid down track. In contrast to this steady increase in market impact, the explicit innovation strategy of Novo Nordisk has unfolded two very “divergent” waves, as described in the above quotation from a veteran device manager. The two epochs which he labeled “divergent” (Medical Systems Division, MSD, 1988-92; Protein Delivery Systems, PDS, 2001-05), held ambitious visions about how Novo Nordisk could create large revenue streams based on medical device systems, beyond the revenue from the insulin itself; in other words making medical devices a business of its own. Outside these two ‘waves’ were periods, in which the explicit strategy defined the insulin drug as the sole revenue generator, and the role of the medical devices was seen more or less as “advanced packaging” for the drug (as it was called by several informants from the device area). One can say that there have been two waves of visionary ideas about revolutionizing diabetes care through the means of medical devices – and in between, the role of medical devices in the explicit strategies has been almost understated, if we consider the actual market success. Metaphorically spoken, the strategy twice has attempted to lift off from the laid down track in a helicopter. As a consequence of the above observations and reflections, I asked myself about the reasons for these fluctuations in explicit innovation strategy, which seemed to contrast the actual development in market impact. To what extent were the fluctuations planned and deliberate or, alternatively, emerging ‘by themselves’, either driven by external trends and events or by bottom-up organizational processes? “How does innovation strategy evolve?” became the overall research question, which guided my longitudinal case study of the medical device innovation activities at Novo Nordisk. First, I mapped and analyzed each transformation of innovation strategy, which had occurred. From the concrete understanding of each transformation, the next level of reflection arose: Is there a pattern in the changes of innovation strategy? Can we, in other words, extract a generalized understanding, i.e. a theory? These puzzles resulted in the formulation of the research questions outlined in the next section. 16 Research questions General questions The topic of this research project is the evolution of innovation strategy, stated in the overarching question “How does innovation strategy evolve?” For examining the topic, evolutionary theory on strategy-making forms a theoretical basis (see next section, and Chapter 2, about the theoretical framework), especially Robert A. Burgelman’s (1991; 2002) theory on the ‘internal ecology’ of induced and autonomous strategies. Consequently, the topic is addressed with a general research question based on the ‘internal ecology’ model:  What is the role of induced and autonomous strategy processes for the evolution of innovation strategy? Specific questions To study the general questions empirically, I focus on the evolution of the innovation strategy for a specific area in a specific organization; namely, the medical device innovation activities at the pharmaceutical company Novo Nordisk A/S in the period 1980-2008. Following a logic of lifecycles (see Definitions below) the evolution of innovation strategy can be split into sub-questions about the creation, the growth and the eventual change of a strategy. Consequently, the following specific research questions are stated: 1. How did the innovation strategy for medical devices at Novo Nordisk come into being? 2. Through which processes was the device innovation strategy substantiated or retained, once established? 3. Through which processes was the device innovation strategy altered or reconfigured? a. Phase by phase, transformation by transformation b. Patterns of change across the individual phases and transformations. Inspired by literature (see next section and Chapter 2) some underlying themes will be examined. One important theme is the role of management cognition for the evolution of innovation strategy. An element hereof is the perception of the medical devices as either core assets for innovation, i.e. seen as means of creating a business of its own – or as complementary assets for innovation, i.e. seen as enhancers of the sales of pharmaceutical drugs. Another theme is the already mentioned relationship between induced and autonomous strategy-making processes, which overlaps with the question of deliberateness versus emergence in the evolution of strategy. And a third theme is the interplay between internal and external ‘ecologies’ in the evolution of innovation strategy. 17 Definitions The following definitions of the concepts used in the research questions are applied (mentioned in alphabetic order): Autonomous and induced strategies: Burgelman (1991) contrasts two sorts of strategy making processes: “The induced process concerns initiatives that are within the scope of the organization’s current strategy and build on existing organizational learning; the autonomous process concerns initiatives that emerge outside of it and provide the potential for new organizational learning” (p. 241). This definition, as can be seen, links to the content or scope of the strategy process and resembles the two modes of organizational learning described by March (1991): exploitation and exploration. However, Burgelman (1991) also links the concepts to the organizational hierarchies: in the normal case, he claims that autonomous strategies work their way up from beneath the organization, whereas induced strategies are exactly ‘induced’ from the top. And further, Burgelman analyzes the process: Does the strategy develop out of local experiments, i.e. vision ‘ex post’, or does it develop out of cognition, i.e. vision ‘ex ante’? The normal distribution of these three parameters in Burgelman (1991, 2002) is shown in Table I-1. Autonomous strategy Induced strategy Content (per definition) Explores opportunities outside current strategy Expands within current strategy Actor Lower or middle managers Top management Process Action-based: learning from experiments (vision ex post) Cognition-based: plan before action (vision ex ante) Table I-1. The normal distribution of parameters in Burgelman (1991, 2002) for characterizing respectively autonomous and induced strategy. Cognition: thinking. “The mental action or process of acquiring knowledge and understanding through thought, experience, and the senses” 1 . In the context of this research, cognition is seen as part of the mental aspect of strategy making, e.g. formulation of visions and strategies in words and models, as opposed to the physical aspect, e.g. trial-and-error experiments or other action in the ‘real world’. Core and complementary assets: builds on Teece (1986), who describes innovation as comprising core technical knowledge, needed for the invention itself, and complementary assets, needed for the successful commercialization of the invention. Teece mentions processes such as marketing, 1 http://oxforddictionaries.com/definition/cognition, accessed 2011-07-16 18 manufacturing and after-sales support as examples of complementary assets. For this research project, however, it is essential that he also mentions: “when the innovation is systemic, the complementary assets may be other parts of the system. For instance; computer hardware typically requires specialized software” (ibid, p. 288). In this understanding, the medical devices may be seen as complementary assets, being part of the pharmaceutical product offering as a whole – but also potentially being seen as core technical knowledge, needed for innovation on par with the other components, such as knowledge about the insulin. Evolution: “the gradual development of something” 2 . The term is here used to label changes of an organization (or its environment) over time, unfolding as cycles of variation-selection-retention. Van de Ven (2007) defines evolution by stating: “An evolutionary model explains change as a recurrent, cumulative, and probabilistic progression of variation, selection, and retention among entities in a designated population” (p. 204). The present research project studies only at one entity, namely Novo Nordisk – however, such organization can in itself be seen as a population or an ecosystem of initiatives (as in Burgelman, 1991) or of multiple cognitive frames (as in Kaplan, 2008). Innovation: the invention and market introduction of new products. Hence, compared to literature on management of innovation in general (example: “Innovation is about knowledge – creating new possibilities through combining different knowledge sets”, in the textbook by Tidd, Bessant & Pavitt 2005), the definition of innovation is here narrowed down to product innovation only, since that is the focus of the present study. Innovation strategy: the strategy for the individual innovation activities with the objective to create product or business innovations (strategy understood as a plan for the future or a storyline of the past). Life cycles: “the series of changes in the life of an organism including reproduction” 3 ; i.e. the sequence of birth, growth and decline which an organization, a technology or a new product follows. In this context meaning that also strategies are born (i.e. formulated), grow (i.e. are implemented and substantiated) and decline (i.e. lose momentum and eventually are formally cancelled, to be replaced by new strategies). However, social phenomena like strategies do not evolve in predefined phases, as we often observe in nature; ‘evolution’ in social science is not deterministic. Strategy: the pattern of activities of a company over time, before action (strategy as plan) or after action (strategy as pattern). This definition builds on Mintzberg (2007), who identifies the two meanings of the word, the forward-looking plan or commitment to action, or the backward-looking 2 http://oxforddictionaries.com/definition/evolution, meaning 2, accessed 2011-07-16 3 http://oxforddictionaries.com/definition/life+cycle, accessed 2011-07-16 19 “pattern in a stream of decisions” (p. 2). In general, whenever the word ‘strategy’ in this thesis is not attached to ‘innovation’, the word refers to business strategy; i.e. it is the corporate storyline about the company’s product-markets – it’s customers, offerings, value proposition, profit model etc. Theoretical framework The present research project belongs to the research field of strategic management and explores the topic of ‘strategy making’. Concepts from the field of management of innovation have been included as well, because of the specific focus on innovation strategy. As the overall research question is “How does innovation strategy evolve?”, we need a basic understanding of what strategies are, how they come into existence and how they evolve. This basic understanding is taken from general strategic management literature, from research on strategy making and, since strategies both exist in observable reality and in managers’ minds, from literature on strategic cognition. The literature on strategic management unfolds several schools of thought (see for example Gavetti & Levinthal, 2004, for a review). In broad terms, these can be placed in a continuum between rational choice theory and behavioral decision theory: a) Rational choice theory builds on a premise of the organization or the management team as a rational agent, where the actor is supposed to make informed and deliberate decisions based on analysis of the consequences of alternative options. This rational premise is consistent with mainstream (neo-classic) economics. Examples within strategic management comprise both the early, operational approach to strategy as for instance in Ansoff (1965), as well as the later positioning school with Porter (1980) as the principal representative. b) In opposition to the rational premise in economic and management theory, a paradigm of behavioral decision theory evolved based on the research of H. Simon (e.g. 1955), who analyzed the limitations of rationality in actual decision behavior. These limitations became known as ‘bounded rationality’. Bounded rationality implies that even if managers are “intendedly” rational, they don’t have full access to information and they only have limited resources and capabilities for processing information. In alignment with the notion of bounded rationality, several schools of thought developed. These schools of thought shared their focus on actual organizational behavior, in contrast to the ideal of rational choice. One such school is the stream of research in management cognition (see Walsh, 1995, and Kaplan, 2011, for reviews), which provides evidence for the constraints and biases of the cognitive representations used in decision making. Another school consists of evolutionary economic theory (Nelson & Winter, 1982), which analyzes how knowledge and experience become embedded in enduring organizational routines. Within strategic management theory, models of emergent strategy and evolutionary strategy making were developed by researchers such as Mintzberg (1994) and Burgelman (1991). If we turn towards the topic ‘evolution of strategy’, one should notice that the rational choice models, such as the positioning school, mainly are concerned with explaining the causes of superior 20 business performance as reflected in the competitive landscape (the outcome of strategy), and less with analyzing the underlying process of strategy formation. The main question for rational choice theory on strategy can be stated as: ‘what kind of strategy should we build?’ Conversely, the evolutionary theories are more focused on the actual behavior in the process of strategy making. The main question of such theories could be stated as: ‘where do strategies come from?’ Thus, Gavetti & Rivkin (2007) summarize the two research strands as “the content-oriented rational-choice class and the process-centered learning class” (p. 422). For the present research project, behavioral theories (or learning models) of strategy-making provide the best foundation for understanding the evolution of innovation strategy. The theoretical framework will be analyzed in detail in Chapter 2. However, some core elements should be presented here. First, the word strategy (see Definitions) has two meanings or two dimensions: Strategy can be a forward-looking plan or vision or it can be a backward-looking “pattern in a stream of decisions” (Mintzberg, 2007). This dualism comprises a time-bound dimension: strategy as perceived before or after action. Furthermore, strategies exist as concepts or theories in the minds of managers (in the form of forward-looking visions and plans; or backward-looking rationalization of experience) and they also manifest as patterns in actual organizational behavior or practice. This dualism comprises a spacebound dimension spanning the two realms, in which strategies exists: the mental and the physical (Gavetti & Rivkin, 2007). These two basic dualisms (time-bound: before/after action and space-bound: mental/physical realms) can be synthesized into a simple model of a learning cycle, see figure I-1, inspired by similar learning cycles in literature on strategic search and managerial cognition (Burgelman, 1988, Figure 2; Gavetti & Levinthal, 2000, Figure 1; Walsh, 1995, Figure 1; Prahalad & Bettis, 1986, Figure 1). The model shows how strategy making unfolds in a learning cycle of theory application into practice, from which experiences fuel further theorizing. This basic learning cycle synthesizes the dualism between the mental and the physical realms in the vertical axis (between Theory and Practice) and the dualism between the forward-looking and backward-looking search processes in the horizontal axis. Going from theory to practice is normally associated with forward-looking and deliberate strategy making, because it springs out of a strategic intent or vision. Going from practical experience to theory is normally seen as the basis of backward-looking and emergent strategy making, because the strategy is formed via trial-and-error, where the understanding develops en route (Mintzberg, 1994), or even backwards, as rationalization of experience (Burgelman, 1988). (The theoretical foundation for this simplified model is presented in detail in Chapter 2). 21 Figure I-1. Strategy making shown as a learning cycle between theory and practice, inspired by similar learning cycles in literature on strategic search and managerial cognition – see Chapter 2. Even the learning cycle is here depicted as one loop, the learning process of course is recursive and in principle endless; loop after loop unfolds. The interplay between deliberate or theory-driven strategy making on the one side and emergent or experience-based strategy making on the other has been very thoroughly analyzed in Burgelman’s evolutionary theory on the ‘internal ecology of strategy making’ (Burgelman 1991, 2002). He basically models how long-term adaptation of the corporate, induced strategy is achieved via continuous integration of local, autonomous initiatives from the ‘internal ecology’ of the organization. The learning cycle in figure I-1 integrates the mental and the physical aspects of strategies in the Theory-Practice dimension. The ‘theory’ part includes strategy formulation, visioning and reasoning in general. Research on managerial cognition informs about the pervasive influence of cognitive representations (or, mental models) of reality. For the present research project, I especially apply the concept of dominant logic (Prahalad & Bettis, 1986; Bettis & Prahalad, 1995), which terms a set of mental models, which have settled as a top management worldview or mindset, created via reinforcement of experiences from the past. Such dominant logic, once established, is difficult to change but has strong effect on the strategy making of a firm. Strategy making Theory Strategy formulation; vision; cognition Practice Strategic action; experiments Theory in Use Experience Forwardlooking search Backwardlooking search 22 Since the present research project addresses the evolution of innovation strategy, we need a basic characterization of this concept as compared to ‘strategy’ in general. First, what is innovation? – “The word innovation derives from the Latin word innovatus, which is the noun form of innovare "to renew or change," stemming from in—"into" + novus—"new". ” 4 . In a business context, “newness can refer to anything that affects customers, manufacturing, sales or service” (Foster & Kaplan, 2001, p. 24). Literature on innovation distinguishes innovation from invention: “Invention is the first occurrence of an idea for a new product or process, while innovation is the first attempt to carry it out into practice” (Fagerberg et al, 2005, p. 4); “Innovation is invention that has produced economic value. Without economic value there can be no innovation. Invention precedes innovation” (Foster & Kaplan, 2001, p. 24). For the present research project, the concept of innovation is narrowed down to the introduction of new products (and the concurrent new business). What is then innovation strategy? – Strategies in general are seen as overall patterns in a company’s activities, either in form of forward-looking plans or backward-looking storylines. Corporate strategy as a forward-looking plan may (but must not) envision the creation of new product-markets. In such case, the corporate strategy comprises a strategy for innovation. However, innovation can be organized as a distinct set of activities within a company. Thus, Christensen (2002) defines: “Management of innovation signifies the management and organization of the individual innovation processes with the objective to produce product or process innovations” (p. 1318). Accordingly, I define innovation strategy as such:  Innovation strategy is the strategy for the individual innovation activities with the objective to create product or business innovations (strategy understood as a plan for the future or a storyline of the past). Clearly, the innovation strategy may be overlapping with the general business strategy – often, the overarching corporate strategy comprises an underlying specific innovation strategy. In large organizations, where innovation is organized as a distinct set of activities, the strategy levels will be equal to organizational levels; the innovation strategy will be the responsibility of the organizational unit for product innovation. 4 http://en.wikipedia.org/wiki/Innovation, accessed 26-05-2012 23 Research design For understanding the evolution of innovation strategy we need a process research model rather than a variance research model (Van de Ven, 2007): “In general terms, a variance model explains change in terms of relationships among independent variables and dependent variables, while a process model explains how a sequence of event leads to some outcome” (p. 148; my emphasis). The fundamental difference between the two approaches is shown in figure I-2. The process study approach does not exclude the search of causality; but the way to causality goes through “a narrative describing how things develop and change” (ibid, p. 148). To establish such narrative, I chose the format of a longitudinal case study of the medical device area at the pharmaceutical company Novo Nordisk A/S, analyzing the medical device innovation activities since the beginning of these activities around 1980 to year-end 2008. The study also includes a combined qualitative and quantitative tracking of the portfolio of product innovation projects within the studied timeframe. Such case study design opens for a ‘thick description’ of the events, contexts and interpretations (Stake, 2000, p. 437). The ‘thickness’ is achieved both via in-depth interviews and studies of archival data resulting in: • Analysis at multiple levels: external industry dynamics; corporate events and top management cognition; local device level events (cognition, strategy and structure); and concrete innovation activities (innovation projects and product launches). • Mapping long term evolution – across more lifecycles of strategy; this opens for seeing generic patterns. Figure I-2. “Two approaches to explaining strategic change” - from Van de Ven, 2007 (p. 149). 30 ‘global’ rationality. “Broadly stated, the task is to replace the global rationality of economic man with a kind of rational behavior that is compatible with the access to information and the computational capacities that are actually possessed by organisms, including man, in the kinds of environments in which such organisms exist” and “we are interested in models of “limited” rationality rather than models of relatively “global” rationality” (Simon 1955, p. 99 and 113). Simon called his theory a ‘behavioral model’ and several schools of research in strategic management subscribed to the behavior oriented paradigm erected by the Carnegie School, which besides Herbert Simon included researchers such as R. Cyert and J. March. Their interest in actual decision behavior as opposed to the ideal or ‘global’ rationality also opened for studying strategic management in a process theory perspective (Van de Ven, 2007 – see Introduction), where the interest moved from studying conditions and variables behind different strategic outcomes to studying the processes leading to the formation and change of a strategy. Basic dualities of strategy One of the researchers to build on Herbert Simon and the Carnegie School was Henry Mintzberg, who has researched within the field of strategy making since the 1960’s (e.g. “The Science of Strategy- Making”, in Sloan Management Review 1967). Mintzberg noted that Simon saw the concept of strategy as a “pattern: consistency in behavior over time” (Mintzberg, 2007, p. 1). This led Mintzberg to pursue the question: “because there is a pattern, must there necessarily be a plan? In other words, must strategies always be deliberate? Or can they emerge: that is, can patterns just form out of individual actions?” (Mintzberg, 2007, p. 4 – the emphasized words are italics in the original). Mintzberg’s strong orientation towards a behavioral or learning oriented model is clearly expressed in the following statement: “If deliberate strategy is about control, then emergent strategy is about learning. It suggests that anyone, so-called formulators and implementers alike, can learn their way into strategies – action by action, perhaps also decision by decision. Indeed, strategies can form without people even realizing it, although they may recognize these strategies once they have formed” (Mintzberg, 2007, p. 5). This quotation perfectly contrasts the rational choice theory, and Mintzberg’s concept of ‘emergent strategy’ opens for new perspectives on strategy making, which has been further explored by other researchers, such as Burgelman (1991). The dualism between deliberate and emergent strategy is one aspect of strategy making. Another aspect – where Mintzberg’s work similarly can illuminate the understanding – is the dualism between the mental and the physical realms of strategy making. Corporate strategies can be seen as managers’ way of making sense of the world and conceptualizing a storyline about the organization in the form of a perceived pattern in the activities over time – looking backward, as a rationalization of past behavior, or forward, as a plan or vision. Strategies are top managers’ tools for understanding the corporate situation – they give language to complex matters, such as dynamics in the internal and external environment. In this sense, strategies belong to the mental realm of perception, reasoning, theory and cognition. This mental aspect of strategies has been well expressed by Mintzberg (1994): “No one has ever seen or touched a strategy. Strategies, in other words, do not exist as tangible entities. 31 They are abstract concepts, in the minds of people. And the best of them seem to be gestalt in nature, tightly integrated, whether intended strategies as synthesized patterns of preferences prior to the taking of actions or realized strategies as synthesized patterns already formed among actions. Thus, serious change in strategy generally means shift in gestalt – the conception of a new worldview, generally based on a permanent change in conditions, or at least the perception of such a change” (p. 240 – the emphasized words are italics in the original). Aligned with the above thinking, Burgelman (1983) states: “the concept of corporate strategy represents the more or less explicit articulation of the firm's theory about its past concrete achievements. This theory defines the identity of the firm at any moment in time. It provides a basis for the maintenance of this identity and for the continuity in strategic activity. It induces further strategic initiative in line with it” (p. 66). Thus, the emphasis in the understanding is here on strategy as a ‘theory’; and this theory is both backward-looking (building on past achievements) and forwardlooking (inducing further initiatives). However, even if strategies are abstract concepts in the minds of people, or theories, they also often manifest in action – for example launch of new products in new markets etc. Thus, strategies have a dual nature, both belonging to the intangible sphere of thinking and to the physical and observable sphere of behavior or action. Identification of a theoretical ‘middle ground’ Gavetti & Levinthal (2004) analyze the first nearly 50 years of research on business strategy. They map the strands of research according to two dimensions: a) the view on choice processes (rational choice models versus behavioral models) and b) the level of analysis. This second dimension in their map concerns whether researchers describe context and situation specific events, or rather seek for general and more permanent structural factors. When applying this map, Gavetti & Levinthal (2004) place the positioning school (e.g. Porter, 1980) in a so-called ‘structure-conduct-performance’ paradigm, which is characterized as being both highly rational and structural. Researchers like Henry Mintzberg and Robert Burgelman are placed in a cluster of ‘emergent views’ on strategy, which are behavioral rather than rational and situational rather than structural: “These authors argued that a firm’s articulated strategy was often an ex post construction, occurring subsequent to the emergence of patterns of behavior that had, de facto, already configured the firm’s strategy” (Gavetti & Levinthal, 2004, p. 1312). Gavetti & Levinthal (2004) propose a ‘middle ground’ between rational and behavioral views and between situational and structural levels of analysis. They suggest the new paradigm ‘in the middle’ to build on evolutionary economics, going back to Nelson & Winter (1982). However, they identify some limitations in the evolutionary theory which should be overcome in order to serve as a broader paradigm. One such basic limitation is the evolutionary theory’s focus on organizational routines as the holder of the “DNA” of the company: “…our general term for all regular and predictable behavioral patterns of firms is “routine”…In our evolutionary theory, these routines play the role that genes play in biological evolutionary theory. They are a persistent feature of the organism and determine its possible 32 behavior” (Nelson & Winter, 1982, p. 14). Such premise nearly excludes elements of deliberation and reasoning. Thus, the evolutionary theory is mostly behavioral: organizations learn through their activities, and this learning becomes imbedded in the organizational routines, displaying large degree of tacit knowledge. If a new ‘middle ground paradigm’ should be able to embrace the rational perspective, then “the current challenge is to identify other genetic traits [than organizational routines], as well as search mechanisms operating on those traits, consistent with more deliberate forms of thinking” (Gavetti & Levinthal, 2004, p. 1314). The dilemma is between remaining faithful to the evolutionary logic, implying that “the property of firm-level behavioral continuity must be maintained” (ibid, p. 1314, my emphasis), and at the same time embracing deliberation and reasoning. “The question then arises as to what extent the evolutionary framework can incorporate elements of deliberation and cognition” (ibid, p. 1314). As an example of ‘middle ground’ research, they mention amongst others Tripsas & Gavetti’s (2000) case study of Polaroid, because it “illustrates the power of cognitive representations as carriers of behavioral continuity, thus giving them a similar status as routines as part of the organization’s genetic material” (Gavetti & Levinthal, 2004, p. 1314). In accordance herewith, research on strategic cognition has been included in the theoretical framework of the current research (see later section). Besides the focus on organizational routines as the holder of the ‘corporate DNA’, traditional evolutionary theory also is biased towards seeing the routines and capabilities as developing “from below” in the organization, and Gavetti & Levinthal (2004) therefore suggest to “pay more attention to the linkages across actors within the organizational hierarchy” (ibid, p. 1315). This perspective is salient in the research of Robert A. Burgelman, which I shall present in the next sections. 33 Where do strategies come from? Mintzberg (1994) defines three sources of strategy:  Intended strategies, which are plans formulated ahead of time (some of which are never realized)  Deliberate strategies, which are intended strategies realized via use of formal control systems  Emerging strategies, which are formed (not formulated) ‘en route’, as adaptation to the real world: “where a realized pattern was not expressly intended” (p. 25). Together, these sources result in the realized strategy. See figure II-1, which shows two strategy processes, respectively Deliberate and Emergent Strategy Process (taken from Mintzberg, 2007). Figure II-1.The sources of strategy, as described by Mintzberg (1994, 2007) (here from Mintzberg, 2007, p. 6). Strategy making as social learning Burgelman’s (1988) case study on Internal Corporate Venture units in large firms analyzes the emergent strategy process in detail. Burgelman (1988) describes strategy making as a social learning process of interplay between an action realm (in the form of entrepreneurial activities) and a cognitive realm (in the form of strategy formulation). Strategic renewal is seen as a result of gradual development, where each step depends on emerging activities in previous phases. Burgelman suggests that in the beginning of a new business, action and strategy are narrowly based and grow together (learning by doing). Gradually, a strategic (cognitive) framework can be abstracted. In the end, strategy can be separated from action – as it becomes institutionalized. Burgelman (1988) describes this institutionalizing of a strategy as a shift from ‘doing well is basis for planning well’, to ‘planning is basis for doing’ (p. 81). Thus, Burgelman differentiates the two phases, respectively before and after the institutionalization (ibid, p. 83):  The emergent state: Planning is retro-active rationalization of autonomous strategic activities. 34  The steady state: A strategy is distilled from experiences to induce further strategic activities through a planning process. What starts as “opportunistic search in the stream of ongoing work” at operational level, manifested in local experiments, can – in case of success – gradually gain impact at yet higher management levels, ultimately to be recognized at corporate management level and thereby change the firm’s ‘concept of strategy’. Hereafter, the institutionalization of the new strategy begins. Figure II-2 synthesizes Burgelman’s (1988) theory of interplay between the two aspects of strategy making: the physical aspect (experiments and implementation in the ‘realm of action’) and the mental aspect (strategy formulation in the ‘realm of cognition’). Figure II-2. Strategy making as a social learning process, based on Burgelman (1988). The process is here depicted as a ‘sound wave’ swinging in the polarity between the realms of cognition and action, gradually gaining resonance at yet higher organizational levels. Forward-looking cognitive search and backward-looking experiential learning Similarly to Burgelman (1988), Gavetti & Levinthal (2000) analyze how organizational search processes can either be forward-looking, based on cognition, or backward-looking, based on experience: “Cognition is a forward-looking form of intelligence that is premised on an actor's beliefs COGNITION REALM ACTION REALM Local management Middle management Top management Opportunistic search Local experiments Medium-range Implementation Company-wide implementation Interpretation backwards (‘retroactive rationalizing’) Corporate plans Medium-range plans Local plan Business idea Trial & error, experiments, practical implementation Evaluation, selection, interpretation, strategy formulation Time Organizational impact Institutionalization: New ‘concept of strategy’ 35 about the linkage between the choice of actions and the subsequent impact of those actions on outcomes. Such beliefs derive from the actor's mental model of the world (Holland et al., 1986)” (p. 113). “In contrast, experiential wisdom accumulates as a result of positive and negative reinforcement of prior choices (Levitt and March, 1988). Choices that have led to what are encoded as positive outcomes are reinforced, while the propensity to engage in actions that have led to negative outcomes is diminished. In this sense, experiential learning offers a form of backward-looking wisdom” (ibid, p. 114). They summarize this understanding in a learning or reinforcement cycle – see figure II-3. Figure II-3. Gavetti & Levinthal’s (2000) model of “intelligence of action” as a learning or reinforcement cycle comprising forward-looking cognitive search and backward-looking experiential search. Gavetti & Levinthal (2000) state that the forward- and backward-looking search processes are complementary, because experience-based search is limited to the number of experiments you can actually engage in, and cognitive search is constrained by the incompleteness of the mental models in use. Their theory is not focused on strategic search; however, their theory can be directly applied to strategy making. Cognitive and experience-based search in a perspective of organizational maturation In a more recent study, Gavetti & Rivkin (2007) develop a time-bound theory on action and cognition in strategy making. Their research question is: “Where do strategies come from?”, and the title of the paper addresses the dualism between the mental and the physical aspect of strategy: “On the Origin of Strategy: Action and Cognition over Time”. They conduct a longitudinal case study of the Internet portal company Lycos and its search for strategies. For building their theory, they contrast two strategy models: “The positioning model portrays strategic search as cerebral and top-down; the core search mechanism is deductive application of economic logic to a firm’s activities—the central elements that are searched. It assumes firms are highly plastic, conforming readily to the shape desired by the 36 management team. The evolutionary model posits that managers are intendedly but boundedly rational (Simon 1957a, p. xxiv). As a result of cognitive bounds, much behavior in organizations is based on semiautomatic rules and routines (Nelson and Winter 1982). The core search mechanism is local search: Actors seek solutions that entail incremental change to existing routines—the key elements searched—often through trial and error (Cyert and March 1963) and with limited deliberation. Managers rely on local search because they are cognitively limited and know their firms are not fully plastic” (p. 421, my emphasis). Based on their longitudinal case study, Gavetti & Rivkin (2007) suggest that positioning and evolution models have different plausibility or likelihood depending on the maturity of both the company and the industry: For young companies in immature industries, local search and experiments is most plausible. This corresponds to the evolution model; action comes before cognition. Conversely, rationality gains plausibility especially as the industry matures. This corresponds to the positioning model; cognition and planning comes before action. Gavetti & Rivkin (2007) conclude: “The world of action, the world of cognition, and their interplay are sensitive to time, and our models need to incorporate this sensitivity” (p. 436). They summarize their contribution such:  “Over time, the cognitive and physical elements that make up a strategy become less plastic, while mechanisms to search rationally for a strategy become more available. This generates a fundamental tension in the origin of strategy: Managers struggle to understand their environment well enough to search rationally for an effective strategy before their firms lose the plasticity [plasticity: see quotation above] necessary to exploit that understanding. A focus on time allows us to synthesize and extend the evolutionary and positioning models of strategic search” (ibid, p. 420). Gavetti & Rivkin (2007) thus propose that the plasticity (“conforming readily to the shape desired by the management team”, p. 421) of the firm decreases as result of the maturation of the company – the organization settles in its competencies, routines etc. By contrast, the possibilities for rationality in strategic search increase with the maturation of the industry. This theory integrates the evolutionary and positioning models of strategy as depending on the stage of maturation of the company and its industry. Conclusion on origin of strategies Based on the presented research, there seems to be strong evidence showing that the strategy processes evolve in phases. Just as we all learn our ways through the childhood via endless trial-and- error processes, so does strategy making seem to begin with local experimentation leading to retroactive rationalizing, based on which top management can induce further strategic activities through a planning process (Burgelman, 1988). Similarly, strategic search in a young organization seems more likely to be based on action or behavior (the evolutionary model), where cognition and rationality (the positioning model) increases likelihood as the organization and especially its industry matures (Gavetti & Rivkin, 2007). Put differently: 37  In the emergent state (a strategy before institutionalization; or an immature organization) strategy making is dominated by experience-based learning (trial-and-error practice and backward-looking search).  In the steady state (a strategy after institutionalization; or a mature organization) strategy making is dominated by cognition-based planning (beliefs and forward-looking search). From the understanding of strategy making as a learning process, I build on previous approaches in literature on strategic search and managerial cognition to model strategy making as a learning or reinforcement cycle. Burgelman (1998) does so in his Figure 2, p. 84. Gavetti & Levinthal (2000) do it in their Figure 1 (shown above as Figure II-3). Prahalad & Bettis (1986) similarly show the dominant logic as being created in a reinforcement cycle in their Figure 1 (shown below as Figure II-9). Walsh (1995) summarizes the body of literature on managerial cognition in a recursive learning cycle in his Figure 1 (p. 282), displaying respectively the development and the use of knowledge structures. ‘Knowledge structures’ are cognitive representations of the information environment, built up from past experience (“mental templates” for interpretation of the environment; ibid p. 281 – i.e. theories about the world). The use of the cognitive representations leads to consequences in practice; which again feed back to further development of the mental templates (the theories). Applying this thinking, my Figure I-1, shown again here as figure II-4, synthesizes theory-based and experiencebased processes of strategy making into a basic, recursive learning cycle. The conclusion from this section is that the two processes (respectively the right and the left side of the learning cycle) have different emphasis respectively before and after the institutionalization of strategy. 38 Figure II-4. The basic learning cycle of strategy making, repeated from figure I-1, inspired from Walsh (1995) and others. The experience-based process seems dominant before strategic institutionalization; while the theory-based process gains importance after institutionalization of strategy. How strategies evolve The internal ecology of strategy making The basic idea that strategy processes are not all planned by top management, but also emerge from lower levels in the organization, is central in the theoretical framework developed by R.A. Burgelman (1991, 1996, 2002). Central here are the concepts of induced and autonomous strategy. “Induced strategy exploits initiatives that are within the scope of a company’s current strategy and that extend it further in its current product-market environment. Autonomous strategy exploits initiatives that emerge through exploration outside of the scope of the current strategy and that provide the basis for entering into new product-market environments” (Burgelman, 2002 - p. 327). This definition clearly reflects the two modes of organizational learning proposed by March (1991), exploitation and exploration: induced strategy as exploitation of existing capabilities and positions; autonomous strategy as exploration outside existing capabilities and positions. According to Burgelman (2002), induced and autonomous strategy should be balanced, since induced strategy reduces variation whereas autonomous strategy increases variation (p. 354). In Strategy making Theory Strategy formulation; vision; cognition Practice Strategic action; experiments Theory in Use Experience Forwardlooking search Backward -looking search 39 other words: the strategic renewal (variation) comes from emerging autonomous strategies, whereas selection and retention are obtained via the induced strategy. This interplay between induced and autonomous strategy is framed as the ‘internal ecology of strategy making’. Burgelman’s theory is depicted in a simplified form in figure II-5. Figure II-5. A simplified image of Burgelman’s model of ‘internal ecology of strategy making’ (based on Burgelman, 2002). The concept ‘strategic context determination’ is the process by which corporate management recognizes a legitimate role of an autonomous initiative in the induced strategy. Burgelman’s theory comprises in-depth analysis of the evolution of strategy and provides many keys to answering my research questions, although the theory concerns business strategy in general – therefore we shall go into more detail with this framework. Burgelman (1991) builds his theory on a longitudinal case study of Intel Corporation. He notes that when Intel in 1985 made the strategic decision to exit the DRAM (memory) business and devote its main activities to the microprocessor business, this seemingly dramatic decision was a formal articulation of a de facto development in Intel’s activities over more than 10 years. In fact, at the time of decision, the memory business accounted for less than 25% of Intel’s revenue, whereas the microprocessor business had gradually climbed to app. 75%. How can such development take place, even though Intel’s formal strategy still saw Intel as a “memory company”? To explain the development, Burgelman (1991) contrasts the two sorts of strategy making processes within an evolutionary framework of variation-selection-retention: The induced strategy process is driven by top management’s intent within the established domain. Even if this might result in planned variation (Burgelman 1991, p. 246), the induced process generally has “a variation-reduction effect on the set of strategic initiatives” (ibid, p. 245). Variation (i.e. alternative and novel opportunities) in the induced process implies that top management, based on its ex ante vision, proposes initiatives (projects) and then seeks resources for establishing them. The main parts of the induced process, however, concern the selection and retention. For the selection, top management applies administrative procedures (such as strategic planning) to define Current productmarket environment Emerging productmarket environments Induced strategy Autonomous strategy Exploiting existing opportunities Creating new opportunities Exploiting most viable autonomous initiatives Long-term adaptation of induced strategy Changed productmarket environment Changed induced strategy Strategic context determination 46 Cognitive and strategic inertia Barr, Stimpert & Huff (1992) analyze longitudinal data from a matched pair of U.S. railroad companies during the decades after World War II, in which the entire U.S. railroad industry was undergoing a significant decline – private cars, and trucks for transportation of goods, were taking over the transportation market. These external dynamics of course hit both companies, which displayed a number of similarities from the outset, such as geographical area, size of company etc. Still, the two companies responded very differently to the changes in the external environment, and their different courses of action led them to different destinies – the one (Rock Island) sought bankruptcy in the mid 1970s and ceased to exist. The other (C&NW) was viable as the article was published in 1992. Based on archival data (annual reports) the authors thoroughly identify and analyze the mental models of the top management of the two corporations. They link the different courses of action to differences in the development of top management’s mental models: “…the leaders of the C&NW not only recognize changes in their environment, they also gradually change their mental models…” - “At the Rock Island, in contrast, changes in beliefs and action are not undertaken until the railroad is near bankruptcy.” In other words, C&NW successfully adapted their mental models to changes in the market conditions, whereas this learning did not take place at Rock Island, which therefore ended with having a business understanding out of sync with market reality. Using vector symbols, this analysis is illustrated in figure II-7. Figure II-7. A graphical model of the railroad case by Barr, Stimpert & Huff (1992). Mental models of Rock Island Market conditions before World War II Sync Declining market after World War II Out of sync Market transformation Mental models of C&NW Sync New mental models of C&NW Sync Transformation of mental models 47 Tripsas & Gavetti’s (2000) case study of Polaroid’s struggle with the challenges of digital photography similarly deals with the problem of adapting the top management cognition to changes in the external environment – in case, the new technology of digital imaging. The new technology per se was not a challenge for Polaroid – their digital cameras got top scores in comparative product evaluations. Polaroid’s challenge mimicked the one of Rock Island – they did not realize that the changes in external reality required them to rethink their business. Polaroid had built their success on a business model, which they themselves compared to Gillette’s razor-blade model: They sold high quality cameras at relative low prize to retain the profits from the consumer goods needed for making pictures (i.e. the films). They wrongly believed that digital cameras would not change the consumers’ wish for paper prints of the images; so top management stuck to the same business model, which had paved the way for Polaroid’s historical success. This cognitive inertia led to the decline of Polaroid. Another example of cognitive inertia and resulting strategic inertia is found in Christensen & Bower’s (1996) study of the disk-drive industry, which presents the dilemma of incumbent firms between serving existing customers with sustaining innovation versus addressing new markets associated with innovations based on disruptive technologies. Disruptive technologies are defined as: “Technologies which disrupt an established trajectory of performance improvement, or redefine what performance means” (ibid, page 202). These technologies tend to be valued more in remote or emerging markets than in established markets (ibid, p. 203). Thus, the managerial challenge is caused by external factors (new technologies) – however, I am here more interested in the internal dimension of the challenge. The dilemma lies in the resource allocation for innovation – in the end, the well-known existing customers will be prioritized over the uncertain emerging new customers segments, resulting in lack of ability to change the innovation strategy. Interestingly, Christensen & Bower (1996) describe how the incumbent firms often experiment with the disruptive technologies internally in the form of technical prototypes etc., but top management disregards these projects in their selection for the innovation strategy, because the existing customers respond negatively to the prototypes. In Burgelman’s (2002) wording, the strategic context determination is blocked by a coevolutionary lock-in, which makes the companies unable to exploit the potential of the autonomous initiatives. Christensen & Bower (1996) state that when the incumbent firms later are forced to take the disruptive technologies up again for product development, it is often too late – consequently, entrant firms are more successful in building businesses based on the disruptive technologies. Christensen & Bower’s theory is modeled in figure II-8. The model shows how the cognitive inertia of top management, by bonds to the existing customer needs, hinders the exploitation of the new technologies. First when entrant companies have success in establishing product-markets based on the new technologies, this external change will force a shift in the cognition at top management level, who will now open for utilization of the new prototypes, which had been put on shelf. 48 Figure II-8. A model of Christensen & Bower’s (1996) theory on incumbent firms’ failure with regards to integrating disruptive technologies in their innovation strategies. Dominant logic Prahalad & Bettis (1986) analyze the mechanisms which lock top management in their thinking, as illustrated in the previous case studies. They describe how managers develop a dominant general management logic based on experiences from the main business of the firm. “A dominant general management logic is defined as the way in which managers conceptualize the business and make critical resource allocation decisions” (ibid, p. 490). “Dominant logic, as we have defined it here, is a mind set or a world view or conceptualization of the business and the administrative tools to accomplish goals and make decisions in that business” (ibid, p. 491). This ‘dominant logic’ summarizes the ‘success formula’ of the dominant management coalition and causes managers to perceive problems and solutions as framed by past experiences. Prahalad & Bettis (1986) refer to the concept of operant conditioning; meaning that behavior, which is reinforced, will occur more frequently in the future. Thus, the dominant logic evolves in a positive cycle of reinforcement of ‘doing the right things’ in the past. “This reinforcement results in their [managers’] focusing effort on behaviors that led to success” (ibid, p. 491-492). The learning cycle which leads to establishment of the dominant logic is shown in figure II-9. EXTERNAL environment INTERNAL environment 1) Prototypes are made of disruptive architecture 2) Negative responses from existing customers 3) Prototypes are put on shelf because of pessimistic sales forecast 4) New firms emerge – trial & error approach 5) Disruptive products step by step outperform existing products 6) Established firms take their prototypes down from the shelf – too late 7) Entrant firms win the new market Here, cognitive inertia is in play! Here, the management cognition changes... 49 Figure II-9, from Prahalad & Bettis (1986), p. 491, showing the learning cycle behind the establishment of the dominant logic of a firm. Bettis and Prahalad (1995) elaborate the concept of dominant logic. “We have come to view the dominant logic as an information filter…Organizational attention is focused on data deemed relevant by the dominant logic. Other data are largely ignored” (p. 7). The direction of organizational attention towards certain issues (deselecting other) is a central element in a behavioral analysis of strategy, cf. Ocasio (1997). Thus, Bettis & Prahalad (1995) place the dominant logic “as a fundamental aspect of organizational intelligence” (p. 7). Their key point is that dominant logic is crucial for an organization’s adaptability. “Interestingly, it [dominant logic] provides a set of heuristics that simplify and speed decision making. This inherently results in ‘adaptive ability’, so long as changes in the underlying logic are not necessary” (ibid, p. 11, my emphasis). As indicated by the previous studies, the bond to past experiences has a ‘toxic side effect’, if the management team wrongly applies the established dominant logic to situations, where changes in the environment have made the learned behaviors inappropriate. Since the dominant logic is shaped via positive reinforcement of successful actions, it can be difficult to alter or unlearn the logic, once established: “Interestingly, the more successful organizations have been, the more difficult unlearning becomes” (Prahalad & Bettis, 1986, p. 498). Therefore, the dominant logic can cement as a ‘deep structure’ (Gersick, 1991), like the scientific paradigm in Kuhn’s (1964) analysis of scientific revolutions. This understanding of the role of dominant logic is reflected in Bettis & Prahalad (1995). They describe the dominant logic as a ‘local optimum’ or equilibrium, which it requires substantial efforts to escape (Bettis & Prahalad, 1995, p. 12-13). Small fluctuations from the dominant logic will end by returning to the original equilibrium – it takes a far move to allow a firm to establish a new equilibrium = a new logic. See figure II-10. 50 Figure II-10 – taken from Bettis & Prahalad, 1995. It takes a far move to escape the established equilibrium of the current dominant logic. Hence, dominant logic may hinder strategic renewal. 51 Burgelman (1983, 1988) uses the term ‘concept of strategy’ in a way that resembles the concept of dominant logic (Burgelman 1983, p. 68 – my emphasis):  “The concept of strategy provides a more or less explicit, and more or less shared, frame of reference or "paradigm" (Duncan & Weiss, 1979; Jelinek, 1979) concerning the bases of the firm's past success. Not unlike the sociological notion of a paradigm (Kuhn, 1970; Masterman, 1970), it provides guidance for further strategic action in line with it. At the same time, it crystallizes the attitudinal and social factors that were selected together with the cognitive, substantive factors underlying the past success. As such, it also is likely to prescribe, often implicitly and tacitly, attitudes and managerial styles and an ideology deemed necessary for the prolongation of the firm's success. Autonomous strategic behavior, identified here as the major source of strategic renewal, thus is likely to encounter nonrational obstacles in its efforts to convince top management that changes in corporate strategy are necessary”. As can be seen, the overlap to the concept of ‘dominant logic’ is huge, and even the ‘gravity’ (Figure II- 10) is described. If I should point to differences, then the dominant logic is a purely cognitive phenomenon; an information filter for making sense of the whole environment of the business; whereas Burgelman’s ‘concept of strategy’ is more closely linked to the strategy. Furthermore, in Burgelman (1988) the change in a firm’s ‘concept of strategy’ is the result of the initial strategy making process, concluding the emergent state. The establishment of a dominant logic would require longer reinforcement cycles within the induced strategy process, perhaps even linked to the eventual stage of ‘co-evolutionary lock-in’ of the strategy (Burgelman, 2002). I have here chosen to apply the concept of dominant logic, both because of the opening towards the literature on management cognition, and because it not always is crystal clear, when Burgelman’s (1983, 1988) ‘concept of strategy’ refers to a deep cognitive structure underlying the firm’s strategy making (the ‘paradigm’) or to the ‘strategy concept’ (the strategy as the individual theory or plan, building on the underlying paradigm). The concept of ‘dominant logic’ is here easier to decode; it always terms the underlying ‘mental models’, the ‘paradigm’ beneath the strategy making. How the dominant logic might be loosened up Based on a recent case study, Kaplan (2008) links cognitive frames and strategy making under conditions of uncertainty into a theory of ‘framing contests’. Kaplan (2008) defines frames as “the means by which managers make sense of ambiguous information from their environments” (p. 729). She describes strategy making as a ‘framing contest’ between different groups within the organization, each promoting their cognitive frame “to make their frame resonate and mobilize action in their favor” (ibid, p. 730). In her ethnographic case study of a single company (a disguised manufacturer of communication technologies), she observed such ‘framing practices’: “By attempting to establish the legitimacy of a frame or of themselves as claimsmakers or by realigning the frames in play, actors sought to push the strategic choice in the direction of their own frames and 52 interests... If framing practices were successful, this process produced a predominant frame” (ibid, p. 736). By proposing this concept of framing contests, the view on management cognition is lifted above the perspective of cognitive inertia – the contest between different cognitive frames becomes a driver of strategic renewal. “I find that frames influence strategic choices, not in a deterministic fashion, but rather in one mediated by organizational framing contests. This model opens up the black boxes of politics in cognitive models and of cognition in political models of strategy making by showing that frames are both constraints and resources for actors acting purposefully to shape strategic choices” (ibid, p. 745, my emphasis). Kaplan (2008) herself makes the link to Prahalad & Bettis (1986) by stating “The framing contests model sheds light on the organizational processes by which dominant logics emerge and change” (p. 746). To conclude: where dominant logic is formed through reinforcement cycles, starting from the initial framing of strategy, the dominant logic may change as result of framing contests between managers – at least, under conditions of uncertainty. Conclusions on management cognition and strategy In conclusion, some key findings from the theories on management cognition could be:  The link between top management’s cognitive structures and strategic inertia is well documented. However, most of the underlying case studies are on technological discontinuities and environmental changes not controlled by the focal firm, setting top management in a reactive role (Barr, Stimpert & Huff, 1992; Christensen & Bower, 1996; Tripsas & Gavetti, 2000). The entrepreneurial role of top management cognition is underresearched.  The initial identity and framing of strategy form the basis of organizational learning cycles, which have an enduring effect via establishment of the dominant logic. One study analyzed how the dominant logic might undergo change processes: Kaplan (2008) analyses ‘framing contests’ as an ecosystem of cognitive frames, which compete for becoming dominant. This theory therefore points towards a more dynamic model of managerial cognition. Integrative competencies The evolution of innovation strategy at Novo Nordisk to large extent is a case of integration of complementary products (medical devices) in a drug-based, pharmaceutical company. Some theoretical concepts could facilitate the understanding of this perspective. In a context of Open Innovation, Christensen (2006) describes two trends in today’s business environment:  Growing technological complexity, which makes deep technical competencies less important than so-called ‘background competencies’ for utilizing emerging areas of knowledge 53  Increasing vertical disintegration of the industries, where nobody can have full control and ownership of the entire value chain of their business. These trends force large companies into a role as system integrators or ‘innovation architects’, and this role again requires integrative competencies for synthesizing various knowledge resources into applications. The foundation of integrative competencies are the complementary assets (Teece, 1986), which typically characterize larger, established companies as compared to technology-specialized startup companies. Complementary assets include customer linkages, established distribution channels, production know-how for economies of scale etc. The role of complementary assets is to amplify the value of the core assets. Teece (1986) describes the core of an innovation as the “technical knowledge about how to do things better than the existing state of the art” (p. 288). Around the core technological know-how, a range of complementary assets are needed for commercializing the basic technical invention. Interestingly, Teece states: “In some cases, as when the innovation is systemic, the complementary assets may be other parts of the system. For instance; computer hardware typically requires specialized software, both for the operating system, as well as for the applications”(ibid, p. 288). Going back to Christensen (2006): integrative competencies, which utilize complementary assets, enable large firms to orchestrate a portfolio of technologies and transform these into product offerings. This whole perspective of complementary assets and system integration is very relevant for the case study of the medical device activities at Novo Nordisk, where the medical devices most of the time have been perceived as complementary products compared to the pharmaceutical drugs in the product innovation strategy. Still, in the present form, the theories lack the longitudinal perspective of the dynamics between core and complementary assets for innovation strategy in a single company – i.e. what happens with the balance between core and complementary over time? 54 A synthesized model of strategy evolution This section provides a synthesized theoretical model of how a strategy comes into being, is sustained and eventually changed, based on the research presented earlier in this chapter. Mintzberg (2007) describes combined deliberate and emergent processes of strategy formation: “If deliberate strategy is about control, then emergent strategy is about learning …almost every sensible real-life strategy process combines emergent learning with deliberate control” (p. 5; my emphasis). This quotation underpins that the conceptual dichotomy between deliberate and emergent strategy is exactly conceptual; in real life, these processes are closely interrelated. Mintzberg’s graphical model (see figure II-1) may therefore be misleading, since it shows the deliberate and emergent processes as two separate sources of strategy. In reality they develop together, as analyzed in Burgelman (1988). In fact, the dichotomy may simply consist of different hierarchical standpoints: what is deliberate seen from one hierarchical level is emergent seen from the next level (cf. figure II-2). However, there is empirical evidence for varying levels of significance of respectively emergent and deliberate processes:  In Burgelman’s case studies of internal corporate venture units (1988) and of Intel (1991) – and in Gavetti & Rivkin’s (2007) case study of the internet portal Lycos – the emergent (experience-based) learning seems to dominate the strategy making when it comes to strategic renewal.  However, in Noda & Bower’s (1996) case study of two telecom operators’ different approaches to the upcoming mobile telephony, the initial strategic framing (deliberate strategy formulation) seemed to determine both the scope and the outcome of succeeding experiences in the market. Based on the above reflections on the conceptual dichotomy, the learning or reinforcement cycle (figure II-4) has been applied as the basic model of strategy formation over Mintzberg’s two separate streams (figure II-1); because the interrelatedness of deliberate and emergent processes is visible. Still, it should be noted that the two sides of the learning cycle might have shifting emphasis in different stages of development. In the following, different evolutions of strategy are applied by applying the generic model of strategic learning (figure II-4), laid out in four scenarios: 1) the emergent state; 2) strategic lock-in; 3) induced strategic adaptation; 4) induced strategic reconfiguration. Scenario 1: the emergent state In the emergent state, the ‘theory’ is most often established via retroactive rationalization of emergent experiments (Burgelman 1988). Figure II-11 models this initial formation of a strategy – i.e. strategy making before institutionalization, referring to Burgelman (1988). 55 Figure II-11. The initial formation of strategy as a learning cycle between deliberate (forward-looking) and emergent (backward-looking) processes. The depicted learning cycle represents strategy making before institutionalization; i.e. in the emergent state (cf. Burgelman 1988). In this phase, the experience-based learning seems to dominate, according to the presented literature. As an empirical example, Burgelman (1988) builds his theory on an investigation of strategy making in internal corporate venture units. When a strategy first has been institutionalized, the strategy making changes character and becomes an induced process: planning before action (Burgelman, 1988). However, strategic learning still takes place. In that sense, Mintzberg’s (2007) statement “If deliberate strategy is about control, then emergent strategy is about learning” is unfortunate; it’s not an either-or between deliberate strategy and learning. Burgelman & Grove (1996) state: “The fact is, corporate strategy is realized by performing a series of such strategic actions, and not via strategic planning. Strategic plans are abstract, far away, and give managers a lot of chances to reconsider as they go along” (p. 21). This implies that when the induced strategy is implemented, practical experiences provide feedback to the strategy making process in the form of positive or negative reinforcement. Scenario 2: strategic lock-in Let’s look at the scenario of positive reinforcement. This occurs, when the central strategic actions are successful. Consequently, the core of the strategy is sustained, although it will continuously be adjusted based on feedback from practice. Such feedback includes the perceived distinctive competencies (Burgelman, 1991) and the perceived critical tasks for success (Prahalad & Bettis, 1986). Following Prahalad & Bettis (1986), repetitive positive reinforcement can eventually result in the establishment of a dominant logic, which becomes hard to change. The dominant logic is the cognitive aspect of ‘co-evolutionary lock-in’ of strategy, using Burgelman’s (2002) terminology. In case of lock-in, strategic renewal is suppressed, and the strategic activities are focused on exploitation of established capabilities and positions. The dominant logic retains the lock-in by directing management attention narrowly towards the core business – everything else is perceived as distraction (cf. Burgelman, 2002). See figure II-12. An empirical example of this learning cycle is Initial strategy formation Framing Strategy formulation Emergent practice Experiments Explorative search Experience Forward -looking vision Backwardlooking rationalization 62 organization of Novo Nordisk, in order to analyze the local management cognition. Together, there are thus three empirical investigations and three data sets: 1) the historical development of the medical devices activities 1980-2010, which forms the overarching narrative; 2) the development in device innovation project portfolio 1980-2010; 3) the two interview rounds 2007 and 2010 amongst device R&D managers. The specific research methods for each of these investigations are described in separate sections below. Special research setting Since I am myself employed at the case organization, the research method includes participant observation and can be characterized as ‘self-ethnography’ (Alveson, 2003) and ‘engaged scholarship’ (Van de Ven, 2007). ‘Self-ethnography’ is defined by Alveson (2003) as “a study and a text in which the researcher-author describes a cultural setting to which s/he has a ‘natural access’, is an active participant, more or less on equal terms with other participants” (p. 174). Where a conventional researcher has the challenge to ‘break in’ to experience the focal organization from the inside, in self-ethnography the challenge is the opposite; the researcher “struggles in order to create sufficient distance in order to get perspective on lived reality” (ibid, p. 176). In the present case, the struggle especially was about seeing the corporate perspective of the history. In the beginning, I found myself again and again taking a standpoint from a pure device area perspective. The corporate perspective and the understanding of the external environment were then built up gradually. Limitations due to business confidentiality As a consequence of the specific research setting, Novo Nordisk A/S had to approve the case study before publication, in order to avoid disclosure of confidential information. As result, all information about the years 2009 and 2010 has been left out from the findings. A further consequence of the confidentiality issue is that names and roles of individual managers are not disclosed – apart from some of the early founders. For the same reason, the names of external consultants are not disclosed. Similarly, the names of specific organizational departments are not disclosed – since these names often label the concrete projects, which the departments host. 63 1. Research method for the longitudinal case study This research project takes a historical perspective to gain insight into the evolution of the strategy, and has the overall format of a longitudinal case study. Qualitative data were used to identify and describe the strategic phases which the Novo Nordisk medical device innovation activities undertook from the start around 1980 until yearend 2010. Quantitative data were then included for analyzing the evolution in product innovation activities, described in the next section. Data collection, data analysis and conceptualization have interacted in an iterative process of interviewing, mapping data along a time axis, comparing with archival documents, categorizing and analyzing data, describing the historical phases and the development of the product innovation portfolio, addressing open questions in conversations with managers (informal interviews), new formal interviews to validate the findings, adapting the analysis etc. A retrospective graph of the overall research process is shown in figure III-1. Figure III-1. Diagram of the overall research process. Data collection The data consist of public annual reports; a design case study published in 1993 (Freeze, 1993); internal documents especially on strategy; internal project portfolio lists and project documentation; a comprehensive internal report of the entire history of the Novo Nordisk device activities, made by a former device production manager in 2006; 43 semi-structured, recorded interviews with current and previous managers from the device area as well as at corporate level; and hundreds of informal day-to-day conversations with current managers and employees with historical experience from the device area. This daily access to data was enabled by the fact that the author since yearend 2001 has been employed at the case organization, serving as an internal consultant within innovation management and innovation processes. The prioritization of data has been, in order of significance: interviews – internal documents – public documents. The interviews were semi-structured and lasted from 10 to 123 minutes each (mean 58 External environment evolution Research topic Qualitative research Quantitative research: Corporate level evolution Interviews Internal documents Public documents Project data Classification Statistics Analysis Device area evolution 64 min.). They were conducted in Danish language from June 2007 through August 2011. All interviews but one were recorded and 26 were transcribed. Detailed notes were taken in all cases but two. See a list of the interviews in table III-1. Date Position Duration Recorded Notes Transcribed Interview used for: 2007.06.28 Department manager 2:03 Yes Yes Yes Mindset analysis 2007 + historical case study 2007.06.29 Department manager 1:36 Yes Yes Yes Mindset analysis 2007 + historical case study 2007.07.02 Department manager 1:19 Yes Yes Yes Mindset analysis 2007 + historical case study 2007.07.03 Device VP 1:21 Yes Yes Yes Mindset analysis 2007 + historical case study 2007.07.04 Device VP 0:54 Yes Yes Yes Mindset analysis 2007 + historical case study 2007.07.05 Senior project manager 0:49 Yes Yes Yes Mindset analysis 2007 + historical case study 2007.07.11 Department manager 1:15 Yes Yes Yes Mindset analysis 2007 + historical case study 2007.07.12 Department manager 0:52 Yes Yes Yes Mindset analysis 2007 + historical case study 2007.07.16 Project director 1:08 Yes Yes Yes Mindset analysis 2007 + historical case study 2007.08.08 Device SVP 0:59 Yes Yes Yes Mindset analysis 2007 + historical case study 2008.04.15 Strategy manager 1:00 No Yes No Historical case study 2008.05.05 Former device manager 1:13 Yes Yes No Historical case study 2008.05.09 Device VP 1:06 Yes Yes Yes Historical case study 2008.05.26 Device manager 1:13 Yes Yes No Historical case study 2008.06.23 Project director 1:01 Yes Yes No Historical case study 2008.07.01 Device VP 1:16 Yes No No Historical case study 2008.08.11 Production manager 1:02 Yes Yes Yes Historical case study 2008.08.26 Department manager 0:42 Yes Yes No Historical case study 2009.01.23 Device manager 0:45 Yes No No Historical case study 2009.01.27 Quality manager 1:01 Yes Yes Yes Historical case study 2009.03.23 Device SVP 0:32 Partly Yes Yes Historical case study 2009.03.30 Device VP 0:10 No Yes Only notes Historical case study 2009.08.14 Quality engineer 0:53 Yes Yes No Historical case study 2009.09.18 Development engineer 1:12 Yes Yes No Historical case study 2009.10.22 Strategy manager 0:46 Yes Yes No Historical case study 2009.11.18 Former executive 0:51 Yes Yes No Historical case study 2009.12.03 CSO 0:35 Yes Yes No Historical case study 2009.12.08 Former device manager 1:34 Yes Yes No Historical case study 2009.12.17 Former device manager 1:37 Yes Yes No Historical case study 2010.06.30 Strategy manager 0:27 Yes Yes No Historical case study 2010.07.07 Strategy manager 0:34 Yes Yes No Historical case study 2010.08.06 Device VP 0:57 Yes Yes Yes Mindset analysis 2010 2010.09.09 Department manager 0:48 Yes Yes Yes Mindset analysis 2010 2010.08.11 Device VP 0:46 Yes Yes Yes Mindset analysis 2010 2010.08.12 Department manager 0:55 Yes Yes Yes Mindset analysis 2010 2010.08.20 Department manager 0:45 Yes Yes Yes Mindset analysis 2010 2010.08.23 Department manager 0:46 Yes Yes Yes Mindset analysis 2010 2010.08.25 Department manager 0:44 Yes Yes Yes Mindset analysis 2010 2010.09.08 Device VP 0:42 Yes Yes Yes Mindset analysis 2010 2010.09.09 Device VP 0:35 Yes Yes Yes Mindset analysis 2010 2010.09.15 Senior project manager 0:44 Yes Yes Yes Mindset analysis 2010 2011.04.07 Former CEO 1:11 Yes Yes Yes Historical case study 2011.08.17 CEO 1:00 Yes Yes Yes Historical case study TOTAL / average 43 interviews 29 informants 0:58 All but one All but two 26 Table III-1. List of interviews. (The 20 shorter follow-up interviews for the mindset analysis 2007-10 have not been included in the list). 3 informants were executives. 4 were at the next level (SVPs). 65 Definitions In order to identify the strategic phases and the linked transitions, two indicators were defined:  Strategic changes are defined as changes in: 1) overall vision for or purpose of the device area; 2) targeted customer segments and value proposition; 3) field of activities; 4) source of revenue. Changes in these parameters are identified via statements in interviews and archival data.  Structural changes are defined as organizational restructurings, in which the entire device innovation area at Novo Nordisk is moved around within the Novo Nordisk organization; internal restructurings within the device area are disregarded. Structural changes are identified via the organizational charts. Data analysis For the analysis of the historical data, I followed the “steps in research on strategy formation” as described by Mintzberg (2007, p. 380-390). The main steps are (based on Mintzberg, 2007): 1. Basic data: collect data to develop chronologies of decisions and actions, trends and events, and results – across all key strategy areas as well as aspects of the environment. 2. Determination of strategy patterns: map each track of events on a common timescale, if possible as visualized graphs under each other. Determine vertically the concurrent changes and identify and label the strategic periods. – See a list of the mapped tracks below. 3. Analysis of each major period: investigate intensively each period of the strategy, including drivers that shaped it, the underlying causes of changes in strategy and interrelationships. Conduct systematic theoretical analysis of each period of change in strategy by use of the chosen theoretical framework. – The outcome of this step forms the basis of Chapter 4. 4. Theory building: Extract the core of each period and its drivers. Interpret, brainstorm, make hypotheses and extract conceptual insights, for each period and for the overall pattern in the whole study. – The outcome of this step forms the basis of Chapter 5. In step 2, I mapped in total 13 tracks distributed at 3 main levels – see table III-2 below. 66 ------------------------------------------------------------------------------------------------------------------------------- 1. External environment:  Management dogmas (i.e. the prevailing concepts of management practice)  Pharma industry trends  Medical device inventions for diabetes  Impact of Type 2 diabetes  Insulin inventions  Pressure from competitors 2. Novo Nordisk corporate level:  Corporate management cognition and identity  Corporate strategy and events  Product tracks (drugs) 3. Novo Nordisk medical device level:  Product tracks (devices)  Device strategy  Device R&D organization  Portfolio of ongoing device innovation projects -------------------------------------------------------------------------------------------------------------------------------- Table III-2. The tracks mapped for the “Determination of strategy patterns” – step 2 in Mintzberg’s (2007) methodology (p. 381-383). The construction of a narrative In the method described by Mintzberg (2007), the phases of the historical development are set via identification of concurrent events at the different tracks in the map of patterns (see Table III-2): “in scanning vertically for concurrent changes in a number of important strategies, delineate overall periods for the whole study and label them” (p. 381). Even it sounds straightforward, it is not. The determination of phases totally depends on the perspective of interpretation. It took me a couple of years to arrive at the phase model applied here; and I have seen many other classifications of the historical epochs of Novo Nordisk’s device activities. Van de Ven (2007) writes:  “Thus, as we move from surface observations towards a process theory, we move from description to explanation. Explanation requires a story, and stories can be understood as process theories (Pentland 1999). In narrative theory the story is an abstract conceptual model; it identifies the generative mechanisms at work. At a minimum this story must describe a progression or sequence of events. In narrative theory, however, the ‘story’ includes a great deal more than just event sequence. In particular, a process theory should hold the following features in the story…” (p. 223) 67 Van de Ven then continues by describing five elements, respectively: 1) sequence in time; 2) focal actors; 3) narrative voice; 4) frame of reference; and 5) other indicators. Especially element 4 deserves to be illuminated here. Van de Ven states: “Narratives carry meaning and cultural value because they encode, implicitly or explicitly, standards against which actions of the characters can be judged” (p. 224). He concludes the description by saying “These five steps in theory building are easier said than done” (p. 224). My frame of reference was the content and nature of the innovation strategy for the device activities at Novo Nordisk A/S. Key questions were which sort of product-markets the innovations were targeting (e.g. was glucose monitoring products included in the strategy?); the degree of systemic and patient-centered approach to the innovation activities; and the definition of either the pharmaceutical drugs or the medical devices as most central. Only through such story, the narrative provides meaning and explanation. A concrete example could be the phase division around the merger of Novo and Nordisk in 1989. I have seen narratives setting the discriminating year as 1989 (the merger was announced in the beginning of 1989 and the implementation started later that year) or 1990 (where the organization was in place). I have chosen 1988, since the visionary strategy about patient-centered homecare was launched in the beginning of 1988, organized in the new Medical Systems Division within Nordisk – although this was one year before the merger. The reason for my choice is that the strategic intent and the organizational setup was continued after the merger – the much smaller device activities of Novo were integrated into the strategy and organization of MSD. Thus, data in itself does not provide the narrative; you need an underlying story. It is also clear from this description that only a qualitative in-depth study can provide the data needed for establishing such narrative. It is the voice of many different informants, inside and outside the case organization, which accumulates not only into a sequence of events, but into an explanation of the change processes at the case organization. Validation of findings My mapping and interpretation of data was continuously, from 2008 on, validated in the way that I presented my findings to different audiences at the device area of Novo Nordisk at many occasions – such as large seminars for employees, management team meetings etc. This gave me the opportunity to capture comments from many sources, which helped the further data capture and interpretation. Based on this iterative validation, I am absolutely sure that my findings are robust, concerning the overall case narrative. 68 2. Research method for the analysis of the device innovation project portfolio The current research project builds on the understanding that strategies belongs to a mental realm of theory or reasoning as well as to a physical realm of action; and that strategy making evolves in learning or reinforcement cycles between these two realms. Accordingly, I investigated whether this relationship between strategy in theory and strategy in practice was traceable in empirical reality: are changes in innovation strategy and changes in the actual innovation projects linked? Or do these realms live their own, separate lives? The product innovation activities were mapped, quantitatively and qualitatively. This was established by a combination of 1) objective data from internal documents regarding year of project start-up, project closure or product launch, and 2) inter-subjective assessments of the nature of each project. For the latter, each project was classified with regards to Burgelman’s (1991; 2002) concepts of induced and autonomous strategic initiatives. The inter-subjective assessment was established via interviews with managers and employees, as well as by the author himself; in the latter case based on project documents, informal conversations with project managers or firsthand experience. See the qualitative categories and empirical definitions in Chapter 4, in the section about the project portfolio. A good starting point was an overview of the historical device development pipeline, made by two device managers in 2003. These data were compared to two internal lists of previous and current development projects, the one including one-page descriptions of each project. The information was then matched with a comprehensive historical report about the Novo Nordisk device activities, made by a former device production manager in 2006. The remaining ‘holes’ in corporate memory were partly solved by visits in the archives and conversations with former project managers. To validate the findings, all device ‘veterans’ were invited to a meeting, where they were confronted with the results. This meeting resulted in even more details and identification of several historical development projects, which had not been discovered yet. All data were compiled in a spreadsheet for statistics. The research resulted in the identification of 233 official development projects. These were first classified to sort out a) product development projects, e.g. injection devices; b) explorative research projects regarding technologies and materials, internally called ‘technology projects’; c) development 69 of accessories and packaging material, e.g. new generations of needles for the insulin pens; and d) process improvement projects like lean management initiatives etc. As result, 102 product development projects were registered; 95 could be classified and analyzed (with regards to the data sample for 1980-2008; not including data from 2009 and 2010). Only these 95 projects were included in the further quantitative analysis 5 . The analysis was made by use of a traditional spreadsheet, compiling objective data (year of project start-up, duration, launch/closure) and the subjective classification of each project. Statistics on the product innovation portfolio were made for the number per year of: product development projects initiated; development projects in parallel; new product launches; project closures; in all cases distributed per category. Data validation The qualitative findings were continuously presented to informants and verified through their comments. Also the project data behind the quantitative research were tested by presenting these to knowledgeable project managers. Their comments and additional information brought the analysis further and filled out many gaps, as for example identification of not yet registered projects or classification of projects, which had not been classified until then. 5 The restriction to the 95 product development projects was made because it showed to be very difficult to gather detailed information about the historical projects; in many cases the only information left was the name of the project. Therefore, it would have been impossible to establish the basic information about all 233 projects. 70 3. Research method for the mindset analysis 2007 The purpose of this empirical investigation was to identify and analyze the ‘dominant logic’ (Prahalad & Bettis, 1986) of the device innovation area with regards to the management mindset on innovation – here termed ‘dominant innovation logic’. Prahalad & Bettis (1986) in their implications for further research request ‘rigor approaches’ for identifying the dominant logic of a firm, beyond simple conversations with managers. The current empirical investigation builds such method for identifying the dominant logic of an organizational unit. Therefore, the research method in itself can be said to hold a novel research contribution at methodological level. Consequently, I have included a detailed description of the entire method in Appendix A. For reasons of business confidentiality, all information after 2008 had to be excluded. Hence, the comparative analysis of the two investigations in 2007 and 2010 could not be included. In this section, I therefore only describe the overall research method for the mindset analysis in 2007 (for details, see Appendix A). The findings from the 2007 analysis are presented in a separate section of Chapter 4. Sample and data The unit of analysis was the device innovation area of Novo Nordisk A/S. A sample of 10 managers was selected as informants. These 10 were chosen in order to represent the organizational functions as broadly as possible. First of all, the VPs of the three functional areas were chosen, plus a VP from a governance unit, who was a former VP from the device area. The other 6 were department managers. The interviews were conducted as qualitative, semi-structured in-depth interviews throughout the summer of 2007. See total list of interviews, table III-1. – All interviews were recorded and transcribed. The interviews were based on lists of questions, which mainly included the themes a) industry identification; b) value proposition; c) core capabilities; d) product innovation portfolio; e) innovation barriers and enhancers. See table III-3 for the lists of core questions. 71  What is the value, we produce for our customers? (we = device R&D)  Which are our most essential capabilities?  Could we be an independent company?  Which industry would we then become a part of?  What are the drivers of this industry?  Who are the jokers within this industry?  Could you mention 3 highlights from the history of device innovation at NN?  Are we still creating history?  What would it take to ensure that we have success in 2015?  What is holding back and what is advancing our device innovation?  Our current development projects: How innovative are they?  Which are the most crucial factors, when deciding if an idea or a project is either supported or stopped?  What are the external blocks and dilutors? – and the internal?  What are our fears/concerns/doubts that prevent us from giving our best?  Do we tend to stick to what we already know? – Why (not)? (Where have we settled in our comfort zones?)  What’s the biggest learning we have made at device R&D?  If time: Actant modeling (subject, goal, provider, receiver, supporter, opponent)  If we were a car, which car would it be? Table III-3. The list of key questions for the interviews in 2007. Summaries of the interviews were made in a format resembling ‘cognitive maps’ (Eden, 1988; Fiol & Huff, 1992; Barr, Stimpert & Huff, 1992; Bougon, 1992), containing condensed statements clustered in themes. Short follow-up interviews were conducted to get approval from the informants of the summaries. The follow-up interviews were not recorded, but the corrections by the informants were noted and approved at the interviews. These follow-up interviews typically lasted 15-30 minutes. Thereafter, the data analysis was based on the summaries as data input. The summaries contained 680 condensed statements in total, corresponding to 0.9 condensed statement per interview minute. Data analysis The condensed statements were taken from the individual cognitive maps into a spreadsheet, reformulated in 107 yet more generic terms, termed ‘constructs’, so that each reformulated construct could cover several informants’ condensed statements. At the same time, the language was changed from Danish to English. For the analysis, the constructs were clustered in three main spheres, Foundation-Innovation-Future (FIF), which basically formed a timeline of past, present and future. The FIF structure emerged out of the data analysis. Within each sphere, the constructs were clustered in groups. See table III-4 with the generic FIF structure. 78 Novo Nordisk in brief Novo Nordisk A/S came to existence in 1989 as a merger of two former rivals, Novo Industry and Nordisk Gentofte, both Danish insulin manufacturers established in respectively 1925 and 1923. Novo Nordisk today defines itself as a leader in the diabetes care market, mainly active within the insulin business, producing the insulin drugs as well as the injection systems for the drug delivery. Novo Nordisk also offers an OAD (Oral Anti Diabetics) product for tablet treatment, NovoNorm® 6 , but this product holds a minor share of both the total tablet market and of Novo Nordisk’s revenue. By yearend 2010 7 , Novo Nordisk had a global market share of the insulin market of 51% (measured in volume) and was by far the world’s largest insulin manufacturer. The company employed 30.000 people worldwide by yearend 2010. Medicine for diabetes accounted for 75% of Novo Nordisk’ turnover; the other business areas being growth disorder, hemophilia and menopause treatment. Headquarters are in Denmark. Key figures from the 2010 account are inserted below. 6 Prandin in the U.S. 7 2010 has been chosen, because it is the last year of my empirical research. However, due to confidentiality issues, the findings presented in the rest of Chapter 4 end by yearend 2008. 79 The diabetes industry From a societal perspective, far the largest part of the costs for treating diabetes and its comorbidities are the costs for hospitalization and treatment in clinics and at private doctors (healthcare services, so to speak). Novo Nordisk actually owns a diabetes hospital, the Steno Diabetes Center in Copenhagen, which is driven as a public hospital. However, the operation of the hospital is not perceived as business – rather as means of building expertise in diabetes treatment and for building credibility within the community of diabetes experts. The same could be said about many other diabetes related activities, which Novo Nordisk is engaged in, such as the World Diabetes Foundation, the Oxford Health Alliance or various diabetes research initiatives. The only explicit 80 diabetes business of Novo Nordisk A/S is the pharmaceutical drug business, covering insulin, GLP-1 (another blood glucose regulating hormone) and tablet treatment. Besides the institutional care at doctors, clinics and hospitals, the diabetes care industry consists of the following sub industries within pharmaceuticals and medical devices, with estimated turnovers from 2010 8 : - Insulin: 90 BDKK - Insulin pumps (devices and utensils, exclusive insulin): 10 BDKK - GLP-1 (a new drug for Type 2 patients): 5-10 BDKK - OAD tablets: 90 BDKK - Blood glucose monitoring (BGM) devices and test strips: 50 BDKK. At Novo Nordisk there have, as can be seen in the historical case description, been attempts to expand the business into the blood glucose monitoring (BGM) market. Traditionally, the BGM and the insulin market have represented two separated sub industries, because the value chains of the two businesses have little in common, apart from serving the same end users. Thus, no global industry players were persistently active on both the insulin side and the glucose monitoring side in the focal period of this longitudinal case study (1980-2008). In general, the insulin industry is a relatively stable industry, dominated by a few of well established companies, displaying ‘low velocity’ (slow rate of change, long product lifecycles). As an example, the NovoLet® prefilled insulin system was launched in 1989 and had a product lifecycle of no less than 20 years (undergoing incremental improvements). The industry today is dominated by three global players: Novo Nordisk, Eli Lilly and Sanofi. Lilly and Sanofi both have a much broader pharmaceutical product portfolio – Novo Nordisk is the only of the players, which is focused on diabetes (app. 75% of the sales come from diabetes products). The most substantial change in external market conditions within the studied period (1980-2008) was the enormous growth in the market due to the pandemic of Type 2 diabetes beginning in the later decades of 20th century. However, the Type 2 diabetes market is very different from the Type 1 market, in terms of patient population, prescribed treatment and partly also the profile of the healthcare professionals dealing with the patients. Therefore, it took many years for the industry players to re-orientate their businesses towards the Type 2 market. From the discovery of insulin in 1921 and forward, the general development on the product side was a competition in purifying animal insulin. From 1982, manufacturing of human insulin was possible, by using gene modification technology on yeast cells. In the 1990’s, insulin ‘analogues’ were 8 The turnover for 2010 has been estimated by a strategy manager of the NN device area 81 introduced. These are insulin molecules, which have been artificially modified to achieve certain characteristics, such as prolonged effect or very fast effect. Since 1985, where Novo introduced the NovoPen®, injection devices also have been a competitive factor. Thus, in 2004, only 44% of the insulin sold worldwide was sold in traditional vials – the rest being sold in cartridges for devices (durable insulin pens) or in pre-filled devices (prefilled insulin pens). Like the rest of the pharmaceutical industry, the insulin business chases the ‘block buster’ business model; i.e. it’s a game of massive investments in R&D as well as in clinical trials and manufacturing, which are paid back by large volume sales, creating a revenue stream of 1 billion USD per year from each drug as the minimum target. Structure of the rest of the chapter The chapter follows the structure of the 3 empirical studies: 1. First, the case study analyzes the historical development of the strategy for the medical device innovation activities at Novo Nordisk 1980-2008, beginning with a brief summary of the development of the entire period. Thereafter, the period has been divided in 6 phases, divided by major organizational changes. These phases are analyzed one by one, looking for drivers and processes of change in innovation strategy. 2. Following the historical case study, the development in the innovation project portfolio throughout the studied period is analyzed. 3. Thereafter, the dominant management logic within the medical device innovation area in 2007 is analyzed. 82 Part 1: The historical case study Summary The medical device activities began separately in the two companies Novo and Nordisk around 1980. Several factors paved the way: a) the prevailing management dogma of diversification, where devices fit well in as yet a business opportunity; b) developments in diabetes treatment made multiple daily insulin injections necessary for the patient, and continuous infusion from insulin pumps was discussed as the future; c) Eli Lilly developed and launched recombinant human insulin in 1982, and this disruptive technology threatened to make animal insulin obsolete – both Novo and Nordisk were years behind, and hence were looking for other parameters of competition. After some years of positive experiences with medical devices, the two companies merged in 1989 and organized the device activities in a separate division, in parallel to the drug divisions. The ‘honeymoon’ after the merger was a very optimistic and entrepreneurial period, and devices (especially disease monitoring devices) were envisioned to become a substantial business of its own. However, the corporate business portfolio after the merger was far too broad – a focusing process started, also inspired by the new management dogma of ‘core competencies’ (Prahalad & Hamel, 1990). In 1992, a crisis hit the company badly, and this amplified the ongoing process of divesting ‘non-core’ business areas. The device activities were confined to only comprising insulin pens – insulin pumps and monitoring devices were terminated. In 1995, the device area was moved to a Production subunit. Within the limited scope of insulin pens, there was here a blooming activity of incremental product innovation until the next strategy shift in 2001; now devices were again envisioned to create substantial new business, including glucose monitoring and ‘everything the patient needed to control his/her disease’. As a new organizational frame, the device innovation activities were separated from device production and got status as an independent R&D unit within corporate R&D, having the same status as the drug R&D units. This new phase experienced a setback already in 2002, due to a new crisis which made the corporate top management team far more cautious and conservative. Thus, the glucose monitoring projects were closed down in 2004. The strategy, however, was not officially altered until 2005, where the device innovation activities were integrated into a drug research unit. The strategy now was ‘back to basics’, i.e. insulin pens. In this period, the device innovation was modest, both quantitatively and qualitatively. By yearend 2008, the device innovation activities were reorganized again. All in all, there have been two waves of device-based innovation strategy, where devices were foreseen to create a business of its own – 1988-92 and 2001-05 – the rest of the period 1980-2008, devices have been perceived as complementary assets for innovation, which should sustain the drugbased business strategy. See an overview of the phases in table IV-1. 83 Early attempts to device innovation (1980-1987) The period around the merger (1988- 1992) The quality crisis (1992- 1994) Harvesting (1995-2001) The vision of closed loop (2001-2005) Integration into the drug research area (2005-2008) General characteristics Exploration and learning: Search and diversity; gradually organizing and institutionalizing. Entrepreneurialism; organizing for innovation in a new division; strategy building. Extreme optimism. Shock; focus on core competences. Turn-around of device innovation activities. Exploitation: Consolidation, focus, alignment. Blooming product innovation, within a restricted area. Exploration – emphasis on new business creation. Diffusion, diversity. Extreme optimism. Exploitation: Consolidation, focus, integration. Back to basics. Overall vision or purpose Supporting new diabetes treatment forms (so-called basalbolus treatment and pump therapy) The vision of patient-centered homecare, supported by a medical system Getting back on safe track Device activities should support the drug business via market differentiation The visions of closed loop and system integration; ‘onestop-shop to diabetes’ The ‘value added pharmaceutical product’, to secure NN leadership via market differentiation Autonomous / induced strategy Mixed picture – mostly autonomous Autonomous Induced Induced Autonomous Induced Top driven change or bottom-up Top driven Top driven Top driven Top driven Top driven Top driven Role of devices: Core / complementary Novo: Complementary. Nordisk: Core. Core Complementary Complementary Core Complementary Corporate organizational structure Multidivisional (M-form) Multidivisional (M-form) Multidivisional (M-form) Functional (U- form) Functional (U- form) Functional (U- form) Organizational setup of device activities More or less hidden in small departments Own division Subunit under the Diabetes Care Division Subunit under Production (in new corporate, functional organization) Device R&D separated from device production, as one of 3 functional areas at corporate R&D Device R&D integrated into the diabetes drug research unit as 3 VP areas Value proposition for customers / users Enabling more convenient treatment forms for patients (flexibility) Enabling homecare of the patient Convenient injection devices for the patient Convenient devices for the patient, ‘meeting individual needs and lifestyles’ (via segmentation) Better glucose control and convenience for the patient, via (intelligent) closed loop systems Increased Quality of Life via convenient devices for the patient, combined with drug benefits Glucose monitoring included in the development portfolio Yes Yes No No Yes No Envisioned revenue from devices Novo: No Nordisk Gentofte: Yes Yes No No Yes No Global industry players in order of relative market strength Eli Lilly. Novo. Nordisk Gentofte. Hoechst. Eli Lilly. Novo Nordisk (merged in 1989). Hoechst. Eli Lilly. Novo Nordisk. Hoechst. Eli Lilly. Novo Nordisk. Hoechst, which from 1999 became Aventis. Novo Nordisk. Eli Lilly. Aventis, which from 2004 became Sanofi- Aventis. Novo Nordisk. Eli Lilly. Sanofi- Aventis. Pfizer (inhaled insulin). Table IV-1. Overview of the historical epochs of the device activities at Novo Nordisk. The following sections of Chapter 4 describe the phases in detail, one by one. The phases are here defined by structural changes. If the structural change took place in the middle of a year, this year is attached to both periods – for example, DRU was established in the summer of 2005, hence the year 2005 is attached to both the PDS and the DRU period. 84 Phase 0: Maturing for medical devices (1923-1980) The corporate DNA Nordisk Gentofte 9 was founded by the Nobel Prize winner in physiology, August Krogh. His wife, Marie Krogh, practiced as a doctor and researched in human metabolism. Furthermore, she had diabetes (Type 2) herself, and she consequently had a natural interest in diabetes. In 1922, August and Marie Krogh therefore went to Canada to visit the scientists, who had discovered insulin the year before. Being a Nobel Prize winner in physiology 1921, August Krogh easily persuaded the Canadian scientists to give him a license to manufacture insulin in Denmark. So, if we are looking for the roots of the internal corporate identity (Tripsas, 2009), I think it is fair to identify these as the combined scientific and medical standpoint of the founders (Novo being a break-out from Nordisk). This identity had impact of the framing of the business (Kaplan, 2008) and thus laid the ground for the learning cycles which established and reinforced the dominant logic (Prahalad & Bettis, 1986) of Nordisk as a science-based, pharmaceutical company. The impact of the founders’ scientific and medical standpoint on Nordisk’s identity can be recognized as late as the 1970’s in this quotation from the official corporate history book: “At the beginning of the 1970’s, the situation at Nordisk was critical. The company had almost no exports. The reason for this was that the management, led by Hagedorn, did not want increased marketing of the company’s products. What mattered most, the management felt, was research and the treatment of diabetes”. Novo had a more engineering oriented starting point: to build up the production, August Krogh had employed his manager from the mechanical workshop at his university laboratory, Harald Pedersen. Harald’s brother, Thorvald Pedersen, who was a pharmacist, was also hired in. However, due to collaboration problems, the two brothers left Nordisk and founded Novo in 1925. Harald Pedersen was an engineer, and he designed a special metal syringe for injection of the insulin already in 1925. This Novo Syringe was the first customized insulin device and it was sold until World War 2, where supply problems stopped the production. Thus, if the ‘founding fathers’ of a company have impact on the internal identity of the company, then Novo from the outset would have a more production and engineering oriented approach to doing business, combined with higher emphasis on broad 9 Both Novo Industry and Nordisk Gentofte have changed names several times during their history. For the sake of readability, I discard this information and shall from here on refer to the two companies as respectively Novo and Nordisk. 85 commercialization: due to Nordisk’s strong position in the home market, Novo was from the beginning forced to be more international. Industry dynamics Both Novo and Nordisk were active in improving the insulin products. The first problem to be solved was “that the effect of the existing insulin products was too short – a serious disadvantage because patients had to inject themselves with insulin several times a day. Therefore, Nordisk and Novo turned their attention to developing a longer-acting type of insulin” (quotation from “Novo Nordisk history”). Thus, Nordisk launched their long-acting insulin in 1936; Novo followed in 1938. The next big problem was to get rid of the ‘impurities’ left from the basic animal material, since these substances made all patients form antibodies against insulin, and some patients even became allergic to insulin. After long research, Novo launched highly purified insulin in 1973; Nordisk followed in 1974. According to an executive informant, the highly purified insulin meant great competitive advantage, which especially Novo used to capture market shares in Europe. The European insulin market had been very fragmented, with 21 European insulin manufacturers in 1972. These rather small companies could not make the technologically difficult leap to highly purified insulin, and hence competition got concentrated on fewer players, with Novo and Nordisk based in Denmark and Hoechst in Germany. Eli Lilly dominated the American markets. Even though both Nordisk and Novo thus were successful in the competition within the core business, both companies diversified into many other business areas, in accordance with the management trends of their time (e.g. see Ansoff, 1965, on diversification). Nordisk for instance developed growth hormone for treatment of children with growth disturbances in the 1960’s and blood plasma products for treatment of hemophilia in the 1970’s. Novo, on their side, started up production of penicillin in 1947 and began producing industrial enzymes in the 1960's. The enzyme business grew rapidly and forced Novo to become even more international. In the late 1970’s, the enzyme business had become bigger than the insulin business, and the enzyme managers at Novo joked about selling off the insulin business; it didn’t grow so much, anyway. Diagnostic systems One more example of diversification from Novo deserves to be described, since it implied some learning for the top management team with relevance for device innovation. Around 1976-77 Novo acquired a small Danish company, which had developed a so-called Cerebrograph – an electronic equipment for measuring the blood circulation in the brain. Hereby Novo established ‘Novo Diagnostic Systems’ as a business area. However, sales never really took off. An executive informant explains: “This taught me two things: first of all this was a highly specialized field. Secondly, this was an area, where you competed with Siemens, Japanese companies etc., and therefore it was no good that you came with just one product. The other companies competed on modularity: measuring equipment, which ‘talked’ with each other – so it was systemic solutions being sold...the possibility of getting a 86 significant position in the market, as a one-product-company, was small. It was an area where we were not so sharp, and the competition was quite different from insulin: when you develop your diagnostic scanners, the competitors can just take a screwdriver and dismantle it and see how it is made. – [Why did you then acquire this company?] – That was also a wrong decision. But we were looking for opportunities for growth, you know. ...You can say that the reason why we jumped into it was the aspiration for new opportunities – and then an analysis, which was insufficient, due to lack of expertise”. Concluding on the period 1923-1980 The early development shaped the corporate DNA in the form of internal identity (Tripsas, 2009). From the founders, Nordisk inherited an orientation towards science, pharmaceuticals (insulin) and medical (specialist) treatment of diabetes. Novo seems to have inherited a more pragmatic and commercial identity, compared to the extremely science and research grounded Nordisk. These initial identity traits of course affected the management cognition, when top management set direction for the business; the basic identity will always affect the worldview or core beliefs, on which strategic cognition is based. Both companies were successful with their insulin business; still, they went through an era of diversification in search of growth opportunities, which led them into many other business areas. Only in one case, a new business area took over from the original insulin business: the enzyme business of Novo in the 1970’s and 1980’s became bigger than the insulin business and in fact drove the internationalization of the company further. At Novo, the experiences from Novo Diagnostic Systems shaped the cognition at corporate management level about the importance of deepening the competencies within the selected business areas and about developing a full portfolio of complementary products within each area. This learning seems very important for the later entrance to medical devices as part of the product offering within diabetes, since it defined a rationale for complementing the basic insulin offering with medical device products. 87 Phase 1: Early attempts within device innovation (1980- 1988) Insulin industry The introduction of highly purified insulin in the beginning of the 1970’s meant a strong competitive advantage for the four companies, who mastered this complex process technology: Eli Lilly, Novo, Nordisk and Hoechst (mentioned in order of market size). However, the insulin molecules still were animal, derived from the pancreas of cows and pigs. Therefore, research was made in order to find methods to produce an insulin molecule identical to the human. The small biotech company Genentech Inc. was founded 1976 to explore and commercially exploit “a new scientific field called recombinant DNA technology 10 ”. The first application was human insulin, achieved via genetic engineering of yeast cells, which would then produce human insulin molecules as result of the fermentation process. They licensed out the technology to Eli Lilly, who thereby were able to market the first recombinant human insulin in 1982, setting back competitors. Hoechst was hit worst, because they were not even able to experiment with the new technology because of German legislation against gene manipulation techniques. Novo had increased the company’s global market shares since the introduction of highly purified animal insulin. However, the recombinant human insulin represented a truly disruptive technology, which could make highly purified animal insulin obsolete. As long as insulin was manufactured based on animal pancreases, the production volume would be limited by the supply of animal material. At the same time, there was increased awareness about Type 2 diabetes and the possibility of applying insulin treatment at earlier stages, in order to avoid late complications of the disease. In combination with the explosion of Type 2 diabetes, this intensified treatment led to an enormous growth in the global demand for insulin in the subsequent decades, with two-digit growth rates. Without the invention of the recombinant technologies, which made the manufacturing process independent of the supply of animal pancreases, it would simply have been impossible to meet this explosion in demand. In other words, both technologically and market wise, the invention of recombinant insulin had crucial impact. Confronted with the threat from the recombinant technologies, which Novo must have known about at least from 1976 on, Novo made two moves: 10 http://www.gene.com/gene/about/corporate/history/timeline.html accessed 2011-06-05 94 Phase 2: The period around the merger (1988-1992) Insulin industry Ely Lilly remained the biggest insulin manufacturer in the years to come, supported by the introduction of recombinant insulin. In the beginning of the 1990’s, Lilly had at least 75% of the U.S. market 12 and therefore probably did not feel threatened by Novo’s and Nordisk’s launch of insulin pen systems. At least, Lilly did not respond to Novo’s insulin pen; and in fact, insulin pens had a very slow adoption rate in the U.S. Perhaps Lilly had invested so much identity and pride in the invention of recombinant insulin that they insisted in seeing the drug as the market differentiator and therefore disregarded the new insulin pens. But I’m only guessing here. The fact is, however, that Lilly didn’t enter the pen market until 1995. If we disregard Nordisk, who was less successful with their Insuject® pen, the first competitor product to NovoPen® was an insulin device introduced in 1990 by the U.S. company BD (Becton, Dickinson and Company), which was designed to fit the insulin cartridges from Novo. Being a pure device company, BD did not threaten Novo in the core insulin business. An executive informant from Novo said: “What surprised me most, was how long time we were alone on the market. Throughout the 80s, the two market drivers are respectively the pen systems from Novo and the premixed insulin from Nordisk”. Obviously, Eli Lilly must have perceived the market drivers differently. Corporate management cognition Novo’s top management registered some trends within the pharmaceutical industry, as explained by an executive informant: “There always have been waves, where the large pharma companies became interested in the diabetes market; now they wanted to enter. At this time also biotech firms emerged [like Genentech and Zymogenetics], so we had to calculate with biotechnical insulin manufacturers too. Or the scenario could have been that the two of us [Novo and Nordisk] competed in the European market until we both were completely tired out, and then one of the large companies would come and make a bet on one of us…There was no doubt that Novo had a very strong position with the pens and Nordisk with the premixed insulin – so you didn’t need an Einstein to figure out that it would be a good idea to merge the two”. The merger was announced January 1989. At that time, Novo was 4-6 times bigger than Nordisk, measured on all key parameters. Therefore, it became crucial for top 12 Source: http://www.fundinguniverse.com/company-histories/Eli-Lilly-and-Company-Company-History.html accessed 2011-06-06 95 management to avoid the feeling of Novo taking over Nordisk. Consequently, each functional area was analyzed one by one and compared between the two, and the most competent unit would continue. This way, the reorganization after the merger was completed around September 1989 and resulted in appointment of more managers with a background from Nordisk compared to Novo. This signal was perceived as crucial for building a common corporate identity. Similarly, both companies’ CEO’s continued jointly– however, the CEO of Nordisk had cancer and died 1990. The merger resulted in managerial optimism – as explained by an executive informant: “By the merger, we suddenly get in new product areas, the palette is much larger, and everybody is excited – now we can join in with the large companies and so on…” Medical device level In Nordisk, top management had followed a vision of making medical devices the ‘third business leg, besides two existing pharmaceutical divisions, Diabetes and Biopharmaceuticals (the latter taking care of other diseases than diabetes). Consequently, Medical Systems Division (MSD) was established in 1988. The overall vision was patient-centered homecare, supported by a medical system, comprising the drug, the delivery system and the monitoring system. See figure IV-5. Figure IV-5. The Nordisk vision of a medical system, from around 1988. “Monitoring” in diabetes refers to glucose monitoring devices. “Drug delivery” refers to injection and infusion systems: insulin pens and pumps. As a genuine division, MSD comprised the usual business functions: Marketing, R&D, Production, Quality, Regulatory, Logistics etc. The revenue stream for the business unit was envisioned to come from the sales of insulin pumps as well as utensils/accessories for the pumps (infuser sets etc.), to be supplemented later with revenue from selling monitoring devices and utensils for these. According to an informant, the monitoring devices were seen as the future revenue generator for MSD – injection devices were seen as less important (remember that the insulin pens were usually given away as free ‘samples’ and would therefore not create revenue by themselves). 96 The CEO of Nordisk, assisted by another member of the executive team, played a major role in the establishment of MSD. The activities grew rapidly, and by the merger in 1989, MSD employed 119 persons. “It was a dynamic period, something happened all the time – it was full steam ahead all the way through”, as one informant puts it. At the merger, Nordisk was recognized by Novo as having the strongest organization for medical devices. Consequently, MSD was continued as the organizational frame of the joined device activities with the status as one of five divisions in the Health Care Group of the merged company (see organization charts from before and after the merger in respectively figure IV-6a and IV-6b). MSD continued to cover the usual business unit functions. Figure IV-6a. The MSD organization before the merger in 1989. Biopharm = non-diabetes related. Figure IV-6b. The Novo Nordisk organization after the merger in 1989, with MSD as a division of Health Care Group. The time after the merger has been described by informants in the device area as very optimistic and entrepreneurial, opening for a lot of innovation projects. Furthermore, the innovation projects established before the merger were continued. The most significant of these projects was Novo’s development of the prefilled insulin pen, called NovoLet®, which was launched the same year as the merger (1989). NovoLet® changed the business model: the drug and the device until then had been separate parts, of which only the drug generated revenue for Novo Nordisk. With the new NovoLet®, the device could neither be sold nor sampled (given away for free) apart from the insulin – instead, the drug-device system was sold together at a higher price per insulin unit, compared to insulin sold 97 in traditional vials or in cartridges for durable pen systems. The NovoLet® system therefore introduced a third insulin business paradigm, besides insulin sold in traditional vials and insulin sold in cartridges for durable pen systems (or for pumps). Eli Lilly saw this; they followed with their similar prefilled pen system in 1995. Because of its simplicity 13 (no filling of insulin or shift of insulin cartridge was needed) the new product category was well suited for the broad segment of Type 2 diabetes patients. Hence, the integrated drug-device systems gained success. A 1997 Financial Times report states: “Insulin was the largest single product category [in the total diabetes market in Europe], helped by the success of the more expensive disposable pens, prefilled with insulin, which are rapidly becoming the standard form of insulin therapy in most countries” (Adamczak, 1997, p. 1). NovoLet® was established as a joint venture between Novo and a Danish plastic manufacturer, Pharmaplast. The plastic components were produced by Pharmaplast, and the first part of the assembly process was carried out by the joint venture company, Diabetes Care Products, at Pharmaplast’s site. Thereafter, the products were handed over to Novo, who took care of the final assembly of the drug container (the insulin cartridge). Novo Nordisk then took the strategic move to acquire the joint venture company, Diabetes Care Products, from the partner Pharmaplast. The decision was driven by management recognition of the future importance of this integrated product. As a key informant states: “Marketing saw it as pure packaging. But if you look at it, then we actually make more money on the plastic then on the insulin. Damn it, if we sold all our insulin in vials, we would have no business. What makes the difference? - That we fill it in cartridges and put them into devices. That’s our business. That’s why I’ve fought for making devices part of Novo Nordisk’s core business; both in development and production wise”. Following this logic, Diabetes Care Products was acquired in 1990 after initiative from the MSD top. The manufacturing of prefilled pen systems was integrated into MSD, in parallel to the existing manufacturing of durable pen systems. Since both Novo and Nordisk had their durable insulin pen systems, it was natural to initiate a common development project for a Novo Nordisk insulin pen. The result was NovoPen® 3, launched in 1992. The brand name of NovoPen® was kept, because it had the widest market acceptance, almost as a generic name for the product category. Technically, NovoPen® 3 was a compromise between principles of the pen systems from both companies. After solving some initial technical problems, NovoPen® 3 became robust and gained market acceptance as the most sold durable 13 ”NovoLet” is Danish for NovoEasy 98 insulin pen for more than 15 years. In 2010, it was still sold; but it was being phased out in favor of NovoPen® 4 country by country. Concluding on the evolution of innovation strategy 1988-1992 Although driven by top management, the establishment of medical devices with Medicals Systems Division can be interpreted as an autonomous strategic initiative for two reasons: 1. The idea implied a complementary business model (revenue stream from devices instead as only from the two drug areas). Hence, devices were internally labeled ‘the third business leg’. 2. The vision of homecare opened for entering new product-markets, such as glucose monitoring products, which represented a distinct market served by a separate industry. The aim of MSD was to make a revenue stream of its own based on medical devices, also by selling devices to other companies; devices were to become the core asset of this new business unit. Thereby, the device innovation strategy differed from the established strategy. On the other hand, the medical device business was perceived one ‘business leg’ in a diversified corporation; medical devices were not envisioned to substitute the drug business. The device organization should continue also to support the two pharmaceutical drug divisions with insulin pens etc. for their drugs; so from the drug divisions’ point of view, MSD was seen as a supplier of complementary assets for their drug business. Consequently, the move into medical devices could be seen as an induced strategic initiative, extending the current drug strategy. However, following the explicit definition by Burgelman (2002), the MSD initiative explored new product-market environments, for example insulin pumps and glucose monitoring products, and in conclusion must be defined as an autonomous strategic initiative. If we apply the model of strategic learning (figure I-1), then the entrepreneurial management approach within MSD surely led to trial-and-error learning. A typical example is the learning from the collaboration with Pharmaplast in the joint venture company Diabetes Care Products, which ended with Novo Nordisk taking over the joint venture. However, the entrepreneurial experiments did not alter the vision of patient-centered homecare; only substantiated it. The overall strategic direction was envisioned by Nordisk’ top management team from the outset in 1988 (or probably already in 1987). Consequently, the innovation strategy must be characterized as theory-driven rather than experience-driven. The strategic learning cycle of MSD is modeled in figure IV-7. The vision of homecare led to the exploration of medical devices as a new business, with diverse product innovation experiments. The results were mixed: the pump business was successful, but the monitoring projects never succeeded (only one product was launched; and had to be recalled, due to malfunction). Only in one case, medical devices were sold to another company. 99 Figure IV-7. The strategic learning cycle of MSD, 1988-92. MSD 1988-92 3rd business leg. Vision of homecare via medical systems Exploration & innovation Mixed experiences Search of growth / new business via devices 100 Phase 3: The quality crisis (1992-1994) Insulin industry After the introduction of recombinant human insulin, the next step was the invention of so-called insulin analogues. These are insulin molecules, based on the human insulin molecule, but artificially modified for achieving special effects, such as very rapid metabolic effect or a very long lasting effect. Eli Lilly was engaged in research on insulin analogues, and also Novo Nordisk had done some research, but had given it up, due to potential cancer risks from these drugs. The growing Type 2 diabetes market continuously attracted attention within the insulin industry, since it became more common to prescribe insulin to Type 2 patients. Novo Nordisk’ response, amongst others, was to invest in and push the market towards insulin sold in cartridges for devices or in prefilled devices, and to position these product categories as more user friendly (compared to traditional vial and syringe) and therefore well suited for the broader segment of Type 2 patients. The use of insulin for treatment of Type 2 diabetes was enhanced by the publication of the first of the so-called major outcome studies, namely the public research program ‘Diabetes Control and Complications Trial’ (DCCT), which was a large study on treatment of diabetes patients in the U.S., which concluded that intensified treatment with insulin provided better treatment outcomes. Corporate level – cognition and crisis Top management seems to have been inspired by the new management dogma about focus on core competencies (Prahalad & Hamel, 1990), facilitated by dialogue with external consultants. An executive informant explains: “After the merger, the company suddenly gets bigger, we get more competencies and business areas, and more resources – and the subsequent business strategy is very broad, also broader than the resources and competencies could cover. I guess that’s why the progress was not as immediate as hoped for… a very broad portfolio, much bigger than our resources and management competencies could cope with”. The focusing of the business began already in 1991 when Novo Nordisk divested the veterinary business, and continued thereafter. Let me quote the annual report for 1991 (my emphasis): “In accordance with our strategy of focusing our efforts on our main business areas, we divested Novo Nordisk Diagnostics Ltd, UK.” In the following years, numerous business areas were closed down or sold. These focusing efforts were amplified by the fact that Novo Nordisk ran into problems at both internal and external frontiers. Internally, top management realized that the company was lagging behind the competitors on the drug side; the pipeline of new diabetes product was too weak. A newly appointed management team of the Diabetes Care Division therefore initiated two projects for developing insulin analogues – first a project for fast acting insulin, and later the same year (1993) also a project for long-acting insulin. “Now we also had long-acting insulin on the way, and after many years it began to look a bit brighter. But we were behind – we had simply nothing [in the pipeline 101 before]”, as stated by an executive informant, who added: “We were still frontrunners on insulin pens. We had the NovoLet®, which had full steam ahead, and we were surfing on the NovoPen® wave, and even though NovoPen® 2 never became a success, we still had total device leadership at that time”. Externally, however, other problems occurred. In 1993, the U.S. Food and Drug Administration (FDA) criticized the insulin manufacturing at Novo Nordisk; according to FDA, the manufacturing did not live up to the U.S. standards of Good Manufacturing Practice (GMP) with regards to the sterilization process. The sterilization of the insulin, when filled into vials or cartridges, is one of the core steps in the manufacturing process of insulin. The critique from FDA were received with skepticism from many internal experts; they perceived Novo Nordisk as the world’s leading insulin company, so of course they knew better than FDA how to manufacture insulin – as an executive informant described this skepticism. The perception of FDA’s requirements as not being valid might have deepened the conflict or postponed a proper reaction. Therefore, the conflict with FDA escalated and resulted in Novo Nordisk losing a large part of its insulin sales in the USA throughout 1994, leaving the market open to Eli Lilly. As a culmination, the head of the Health Care Group at Novo Nordisk was forced to leave his position in 1994. In the end, Novo Nordisk had to comply with the requirements from FDA. The crisis meant an internal shock in the organization, since it hit the core of the core of Novo Nordisk’s business and historical identity. As an executive informant says: “It shook the very foundations of the company”. The crisis is internally referred to as ‘the GMP crisis’ and is still a sensitive issue; even if it implied a steep learning curve for the organization by starting a long-lasting program for assuring quality throughout the manufacturing process. In the annual report for 1993, a new corporate vision is announced, listing 7 topics, one of them being Quality; “The name Novo Nordisk must be synonymous with quality”. The crisis is here touched upon indirectly: “In the last few years the US authorities, in particular, have set far more stringent requirements concerning documentation of the many processes in the development and production of pharmaceuticals. As a result, we have greatly intensified our efforts in the quality assurance field. During the year the number of employees working on quality assurance tasks more than doubled” (my emphasis). The annual report for 1994 is closer to admitting the severity of the crisis. The Directors’ report begins with the words “In many ways, 1994 was a difficult year for Novo Nordisk” and few lines later goes on: “a year when the company incurred substantial extra costs in connection with significantly increased quality assurance and documentation activities” and “…the delivery problems that arose mainly in the wake of the company’s effort to fulfill the increasingly stringent quality and documentation requirements of the authorities. The net effect for Novo Nordisk of these delivery problems was a loss of global market share in 1994 of approximately 2 percentage points. While Novo Nordisk lost insulin market share in the US and Southern Europe, it consolidated its position in several other countries…By the end of the year, the [production] plants had resumed normal business” (my emphasis). The crisis amplified the movement for focusing the business on fewer areas. The official book of Novo Nordisk’s history (“Novo Nordisk history”, 2011) mentions a new business strategy: “In 1994, Novo Nordisk implemented a new business strategy to ensure progress in a rapidly changing and ever more competitive world. Focus was on the two core business areas, Health Care and Enzyme Business, while 102 other areas such as Ferrosan (dietary supplements and other over-the-counter products) and Plant Protection (biological plant protection) were divested.” In the annual report for 1994 is stated: “1994 was also the year when Novo Nordisk changed its business strategy. The sharpened business focus provides the basis for and belief in renewed profitable growth…The essence of the new strategy is increased focus on the company’s core businesses”. According to the annual report, the new strategy was launched “towards the end of 1994”. The new strategy also affected the corporate structure. The annual report for 1994 says: “Implementation of the new strategy required a substantial adaptation of the organisational structure. The former divisionalisation of business areas was replaced by a functionalized and process-oriented structure, and a number of staff functions were gathered at a corporate level.” According to my informants, this restructuring was implemented in 1995, implying a ‘reverse Chandlerian’ restructuring from a multidivisional to a functional organizational structure. Such U-form structure (Chandler, 1992) was a logical consequence of the corporate move from diversification towards focus on core business. An executive informant, who was part of the creation of the new strategy in 1994, explains it this way: “The strategy process was amongst others a result of the quality crisis. Corporate management was not sure that we had the right strategic focus. I was asked, together with [name] and some consultants to analyze the long term pharmaceutical strategy…we should lay down the future track in order to become competitive. And today you can see how we since then [1994] have narrowed down and narrowed down, so that we are now [2011] a very diabetes focused company. We regard our core competencies as understanding proteins for treatment of chronic diseases, especially diabetes; the formulation and administration of these, the modification of these – and then to understand and relate to a group of people, who has a chronic disease, and the responsibility which follows…So it was just too broad [before 1994] – we didn’t have the structures and procedures and evaluation models and management competencies to secure the success of all this. Hence there has been a narrowing, a professionalization process, making it more effective… there has been a sharpening of the core competencies both at Novo Nordisk and Novozymes, as I see it – also due to changes in the environment. Other companies began being able to make similar products, which could compete with ours, and we lost market shares”. The same informant concludes on the learning from the quality crisis: “From the quality crisis we learned how important a well developed and robust quality system is, with training and documentation, - that might even become a competitive advantage, which I actually think it has become today”. Medical device level When the focusing process started in 1992, medical devices were disregarded as ‘core business’. It was suggested to give Medical Systems Division status as supplier in the form of an independent company, but this proposal met resistance by the MSD management: “We did not want to be treated like suppliers. At that time, insulin was a generic product. The whole value creation was in the innovation of the pen systems – so we said that we would rather give up our independence and become a part of the diabetes division and have influence there, than just being placed as a supplier”. The 103 company was (since the merger in 1989) organized in two main business units, Health Care Group (pharmaceuticals) and Bioindustrial Group (the enzyme business). Each of the business units was split into several divisions, as result of the fusion of multidivisional setups in both companies. Eventually, the device area became a subunit to the Diabetes Care Division in the reorganized Novo Nordisk in 1992 under the name Medical Systems. See figure IV-8 for an organizational chart. Figure IV-8. Medical Systems placed under the Diabetes Care Division in 1992. As can be seen, MS kept the usual functions of a division; except for a marketing function. The structure is ‘M-form’ = multidivisional (Chandler, 1992). As response to the corporate quality crisis, MS was told from top management to discontinue both the insulin pumps and the monitoring projects; focus should be on insulin pens. An informant comments: “The idea of being able to cover all aspects of treating diabetes was shot down. Medical Systems was only allowed to develop and produce pens. Hereby the strategy of Medical Systems Division was winged.” The informant here refers to the ‘vision of homecare’ as shown in figure IV-5. The pump projects were discontinued, and the existing customer base was handed over to a competitor. For several reasons: top management did not believe that insulin pumps would ever become a big market (at the end there were about 5,000 users of the Nordisk Infuser®), and at the same time these products required a costly service-oriented organizational set-up, including call centers to assist patients when using the complex insulin pumps. “It’s a different type of business than just pouring drugs out to the pharmacies”, as stated by an informant. Furthermore, the customer complaint rates from the complex pumps were high compared to pharmaceutical drugs. Selling drugs via pharmacies only rarely results in customer complaints, so Novo Nordisk was not used to handling a relative high amount of direct complaints from the users. “Novo Nordisk was not geared to this, it’s was alien for the organization”. The development projects for disease monitoring devices were then stalled, partly due to technical problems, but also because of lack of business synergy, despite the obvious link in usage between 110 monitoring. Such physical integration was often desired by patients in market research studies. InDuo® was launched in 2001, but flopped for several reasons, primarily marketing issues. A specific reason was the business model: reusable devices as Innovo® were usually ‘sampled’, i.e. given away for free to create customer loyalty. Innovo®, being an electronic device, was far more expensive than mechanical devices, and this extra cost would have to be held by the local subsidiaries, which would in the first place see this extra cost as a loss of profit. Forced by competition, Novo Nordisk licensed in a development project from Aradigm Corporation in 1998 on inhaled insulin. This project required massive investments, also in the device system. The development was kept at Aradigm in California, and Medical Systems was not directly involved. According to an informant, Medical Systems was still run much like an independent business unit, despite its displacement under Production. Consequently, there was a pressure from the rest of the organization to force Medical Systems into the same (functional) logic as the rest of the firm; the autonomy of Medical Systems was seen as a problem. Thus, in 1996, the production of disposable pens was separated from Medical Systems. At the next restructuring in 2001, device R&D and device production were fully separated. Concluding on the evolution of strategy in the harvesting period (1995-2001) For corporate management it was vital to keep Novo Nordisk as an independent company. In order to secure independence it was perceived necessary to focus on selected areas and build deep competencies within these; "we will compete with the big boys in areas where they cannot compete with us" as expressed by an executive at that time. Michael Porter (1980) would have called this a ‘focus strategy’. The efforts to focus the business led to the corporate reorganization into a functional, U-form structure (Chandler, 1992) in 1995, which was better suited for deepening the competencies within a narrow area. When analyzing the internal drivers, the focusing process thus first and foremost is theory-driven, based on the reasoning just described; however, the external pressure of competition from large pharma companies drives the process as well; management reasoning interprets the external pressure from competition. In the period after the corporate restructuring in 1995, the device area was placed as a functional area under Production. This implied relatively little attention from corporate management, and also relatively high degree of freedom, as long as the overall strategy was not challenged – i.e. focus should be on insulin pens. Some informants have mentioned this relative freedom as a positive side effect – and in fact, the product innovation seems to have bloomed, even though it was within a confined area, mostly unfolding incremental innovation projects (see the analysis of the project portfolio in a separate section of Chapter 4). Such an environment could be seen as ideal for autonomous initiatives in the classic Burgelman (1991) sense; but the actual innovation projects extended the established strategy – for example, supplementary products were developed such as needle inserters, and the devices were customized for different customer segments. There were no attempts to create new business or independent revenue streams based on devices, or to explore 111 new product-markets. In other words, with regards to device innovation strategy, the induced strategy was continued unchanged from the previous period, and the status of devices as complementary assets compared to the drug was never questioned. The strategic learning cycle of the harvesting period (1995-2001) therefore displayed theory-driven search of market differentiation opportunities via product innovation, aimed at specific customer segments. The success with these activities reinforced the strategy of devices as means of market differentiation. See figure IV-11. Figure IV-11. The strategic learning cycle of the harvesting period (1995-2001). Devices as market differentiator Product innovation Success in differentiation via devices Harvesting 1995-2001 Search of differentiation opportunities 112 Phase 5: The vision of closed loop (2001-2005) Insulin industry The reorientation of the industry towards the diabetes Type 2 market continued. Hence, the longacting insulin analogue Lantus®, launched 2000 by Aventis, gradually gained market share as typical insulin for treating Type 2 diabetes, since it was marketed as a ‘once-daily’ insulin, implying a simple and easy-to-understand treatment regime. In 2003, Lantus® had gained a 14.6% market share of the Western insulin market (USA-Europe), mostly based on sales in the U.S. This corresponds to sales revenues for Lantus in 2003 of 569.4 m$ in Europe and U.S.A. (Hamilton, 2004). Also other insulin analogues gained increasing market acceptance. Thus, in 2002 Eli Lilly’s fast-acting analogue, Humalog® (introduced 1996), passed the sales of the previously leading insulin product, the human insulin Humalin® in the U.S./E.U. market (Hamilton, 2004). To address the exploding Type 2 market segment, the diabetes companies aimed for developing new drugs, first of all the hormone GLP-1 (glucagon-like peptide-1), which is well suited for treating early Type 2 diabetes, before the onset of insulin treatment. GLP-1 lowers the blood sugar level in a stable and slow-acting way, without the risk of hypoglycemia, which always is present when treating with insulin. A 2004 report states: “Future growth drivers include the non-invasive insulins [e.g. inhaled insulin]…and the glucagon-like peptide-1 (GLP-1)” (Hamilton, 2004, p. 13). In 2005, Eli Lilly introduced the first GLP-1 drug, Byetta®, licensed in from Amylin Pharmaceuticals. Novo Nordisk was several years behind in the development of their GLP-1. However, Novo Nordisk was still successful in pushing the market towards devices. By 2002, insulin sold in cartridges for durable devices or in prefilled devices had overtaken insulin sold in traditional vials; 52% of the global volume of insulin was now sold in devices (according to internal statistics). Novo Nordisk especially pushed the prefilled insulin systems (NovoLet® and FlexPen®). During the 5-years period 2000-2004, the total market growth in volume of insulin sold was 0% annual growth for insulin in traditional vials, 7% for insulin in cartridges for durable pens systems and 16% for insulin in prefilled pen systems. Novo Nordisk had also success in other ways. Besides the device-based strategy, an increased sales effort in the U.S. resulted in Novo Nordisk these years taking over the position as the world’s largest insulin manufacturer, leaving Eli Lilly as number two. In 2004, Aventis (including the insulin activities of the former Hoechst) was acquired by the French pharmaceutical company Sanofi-Synthélabo, forming Sanofi-Aventis. This merger gave financial power to the third largest insulin manufacturer. These three companies (Novo Nordisk; Eli Lilly; Sanofi-Aventis) now totally dominated the global insulin market. However, a new player had found an entrance: In 2004, the pharmaceutical giant Pfizer licensed in the development project for inhaled insulin from Nektar Therapeutics. Inhaled insulin was foreseen to have great advantages in the Type 2 diabetes market, because it made injections obsolete. 113 Corporate strategy The change of strategy began at a special workshop in 2000 for the team of corporate managers from Operations (production, sales and marketing). The background was discussions in the pharmaceutical industry about diversifying from medical drugs into total healthcare solutions, as well as general attention to potential revolutions in diabetes treatment. The workshop was ignited by input from an external management professor on topics around possible future changes in diabetes treatment: What could change the whole business model? What were the unique strengths of Novo Nordisk? Why didn’t Novo Nordisk set more outrageous ambitions for diabetes control and convenience for the patients? Why didn’t Novo Nordisk play the role as system integrators and then deliver the most convenient products to the patient? – During the next half year, a taskforce driven by managers outside the device area made a detailed plan for a new device organization. After some consideration, the Executive Management Committee approved the plan and released the funding. The new device strategy was based on the so-called ‘closed loop’ vision: A system mimicking the functions of the healthy body by combining an insulin delivery system with a blood glucose monitoring system. This meant that the monitoring of the blood glucose level and the infusion of the needed amounts of insulin could be maintained automatically in a ‘closed loop’ of delivery and feedback. “Closed loop was broadly discussed in the diabetes society – it was like the ‘holy grail’ of diabetes treatment”, as a key informant put it. The ‘closed loop’ might resemble the 1988 vision of ‘homecare’ (see figure IV-5). However, the big difference is the continuous monitoring of blood glucose and infusion of insulin, carried out by a more or less automatic system. Thus, to construct ‘closed loop’, the devices had to move from mechanical products to ‘intelligent products’, enabled by electronics and software solutions. Detailed business plans were created for the needed development projects, and the emphasis was on business creation, since the strategy again (like in the MSD phase) opened for entering the market for glucose monitoring. “Yes, we wanted to develop total systems. Everything, the patient needed to control his/her blood glucose level. We wanted to be a ‘one-stop-shop’ to diabetes”. A Financial Times report from 1997 might shed light on the background behind these visions (my emphasis): “The trend has been towards greater product sophistication: urine testing has largely been superseded by blood monitoring, and insulin syringes are rapidly giving way to user-friendly injector pens. … In the more distant future, hi-tech monitoring systems could transform implantable infusion pumps into a functioning artificial pancreas”. “However, the main opportunities in the near future lie in the exploitation of new technologies, particularly for noninvasive glucose monitoring, especially continuous monitoring; and in more rapidly absorbed types of insulin, probably as inhaled insulin.” “The market leaders of the future are likely to come from alliances already formed between the producers of insulin delivery products and the manufacturers of monitoring systems.” “A number of alliances have recently been formed between manufacturers of insulin, insulin delivery systems, and monitoring systems…These strategic moves are expected to lead to a new generation of diabetes 114 products based on noninvasive monitoring technology and improved, probably in vivo, insulin delivery systems.” (Adamczak 1997, p. 3-4). The systemic strategy, including glucose monitoring products, was envisioned to multiply the revenue from every diabetes patient, a so-called ‘value upgrade’, which in the strategy documents was related to the market evolution created by the introduction of insulin pens. One of the key persons involved comments: “Novo Nordisk could get a bigger share of the pie via diversification into other parts of the diabetes market. And the executive committee saw that if anybody could do this, it was us. Also the board of directors approved that we should be more expansive”. The needed competencies for developing new glucose monitoring devices and for the system integration were envisioned to be partly sourced in from external partners, partly built up via resource upgrades. Also acquisitions of specialized firms were considered – but in each case given up. An informant, who assisted the creation of the new strategy, lists three main factors as driving the plans: 1. The vision of medical devices seen as ‘the third business leg’ (stemming all the way back from Nordisk in the 1980’s and the MSD period) had not been forgotten. 2. CGM: Continuous Glucose Monitoring, was introduced in 1999. Instead of single blood glucose measurements, a device could continuously measure the glucose level in the subcutaneous tissue, thereby enabling the patient to see the glucose level at any time, as easily as seeing the time at a watch (in principle). 3. The insulin pump market now grew rapidly (around 40% p.a.), driven by new types of pumps being introduced. The pumps represent the other side of the ‘closed loop’: the continuous insulin infusion. The responsiveness from corporate top management to the new visions was influenced by optimism following the establishment of Novo Nordisk as an independent pharmaceutical company in 2000, where the enzyme business was demerged. Yet another factor was the fear of being left behind by competitors, in case they succeeded in exploiting the new technologies: “If they succeeded, then we could risk getting reduced into suppliers of this muddy liquid, which was inside the pumps and devices, which were controlled by others”. The top management team, coming from the Health Care part of Novo Nordisk, was new in the role as a corporate management team. The new team saw the logic in the proposed strategy plans for lifting the device research and development activities out of Production; “time was against having the R&D of Medical Systems placed in Production”, as an informant says. Hence, device production and device R&D were separated, and the device R&D activities were moved from Production to the corporate R&D function. The two existing R&D units, the drug areas Discovery and Development, respectively responsible for early drug research and late drug formulation, were thereby supplemented with a new device unit, called Protein Delivery Systems (PDS). See organizational chart in figure IV-12. 115 Figure IV-12. The device R&D unit, Protein Delivery Systems, placed in corporate Research & Development from 2001, in parallel to the two drug units, Discovery and Development. A drug formulation department from the drug Development unit was transferred to PDS to ensure better system integration. PDS was publicly announced in August 2001, and the plan was to double up of the device related R&D activities. A part of the expansion was due to formal start up of a technology research unit for devices – such research previously had been embedded in the product development projects. Protein Delivery Systems was presented with these words in the annual report for 2001: “Novo Nordisk has committed itself to leadership in drug delivery. One primary research area is Protein Delivery Systems (PDS), and a new unit has been set up to develop innovative methods of administrating insulin and other therapeutic proteins. In practical terms, PDS combines Novo Nordisk’s expertise in diabetes protein formulation and insulin injection systems, for developing convenient and fully-integrated protein delivery systems for patients in several therapeutic areas. The unit focuses on injections devices, advanced continuous infusion systems and glucose monitoring systems.” In the same annual report, there’s a 3-page long article on “Leadership in insulin delivery devices”. For the new corporate management team, the creation of PDS was also a way of opening a window for making medical devices a business of its own. I let an executive informant explain: “When we had carried out the split of the company in 2000, and we had experienced the boost in motivation it gave to make a sharper, better defined profile and strategy for the company…with some vision one could imagine the company from 2000 develop into a device company, a diabetes company and a biopharm company. It’s three different business areas, three different types of innovation and risk – therefore one could imagine that it would attract different types of investors…With PDS there was an ambition that we could sell our device knowledge and make it commercial to other companies. We believed that we were world champions within devices, so there ought to be somebody who could make use of this. One should also not exclude that we would actually learn something from working together with partners”. However, in 2002 the new corporate executive team was shaken by a crisis. The sales revenue for the first quarter of 2002 did not live up to the expectations – it even was necessary to announce a profit warning and adjust the budgets for the whole of 2002. An executive informant comments: “This implied a breach of confidence. The causes were partly internal – there were things, which we could do better, our internal control and forecasting systems probably were too poor – but it was also the global economy; it was just after 9-11 in 2001, which had provoked a sudden depression”. Furthermore, in the summer of 2002, a research project for tablet treatment had to be terminated, due to cancer risk. The annual report for 2002 begins with the words: “The year 2002 has been a very challenging time for 116 Novo Nordisk. On 10 April 2002 we announced that due to unexpected factors, full-year performance was not likely to meet our previous guidance”, and states on the same page “On 22 July 2002 we suspended the phase 3 trials of ragaglitazar (NN622), a promising dual-acting insulin sensitiser. This was done based on urine bladder tumour findings in one mouse and a number of rats.” The confrontation with these problems of course had implications for the new corporate management team. An executive informant comments: “I think we got a rap on the knuckles as new management team in 2002. It made us swear that we should show them that we, damn it, could manage – so there was not much internal wrangling or fighting, we simply wanted to survive and show that we could do it. Therefore, we have had a high degree of continuity and internal trust within the team”. One of the effects of the events in 2002 was the retreat from development of tablet treatment. An executive informant explains: “We decided to concentrate on what we were good at, namely proteins. So no more synthetic chemistry and small molecules… It was partly a reaction to the NN622 project, which was stalled in 2002. We had a large portfolio of these projects; therefore it took some years to withdraw. But it showed that there was increasing distrust in this whole class of drugs, so we decided to stop”. According to the same informant, the combination of financial problems and drug research problems in 2002 turned the corporate management team into a more conservative or internally oriented mode of management: “In the following period, we strengthened our control and forecasting systems, and the whole management team focused on basic issues to ensure that we had a good understanding of our markets and where we were going. That was perhaps also a reason behind that it took so long time, before we took the strategic decision to terminate the research within small molecules. Remember, at the same time the dollar exchange rate decreased dramatically, so we needed to increase our productivity, because many of our expenses are paid in Danish currency. So this was principally a period with internal focus in order to become better at controlling the business and increase our productivity because of the weak US dollar…We overcame both the breach of confidence and the productivity crisis”. In such management climate, risk willingness of course decreases, and this development hit back on the ambitious visions behind PDS, comprising complex ‘closed loop’ systems and business creation based on devices. I will let an executive informant explain the change in management view: “We saw that there were problems within devices in getting new versions [of existing products] ready, and on top of that we were engaged in much more complicated devices, which we from desk analysis thought would be fantastic, including ideas of closed loop and all this. We made this integration of Innov®o and BGM, it was called InDuo®, but we were not able to do it in a way, which was accepted by the market. The market turned towards more simple and convenient solutions – injections once daily, disposable devices etc. And in the end, we also moved in that direction, because we said ‘enough with all these complicated things’. We decided that we should not serve people outside our own business as long as we had problems with developing our own devices, which were one of the competitive advantages of Novo Nordisk. So as result, PDS was folded back again and we said ‘this is an internal part of Novo Nordisk and we don’t sell devices to others’”. This development in top management cognition implied a meltdown of PDS in the years 2004-2005. 117 Medical device level With PDS, Novo Nordisk again aimed at entering the blood glucose monitoring market. This time the entry ticket was new continuous glucose monitoring technologies – that was, measuring the glucose level via a three-day sensor, which was inserted under the skin, instead of the single measurements via finger pricking. These technologies were premature, and the development had highly explorative character. R&D for inhaled insulin was still carried out by the partner Aradigm Corporation in the U.S., but from 2004 Novo Nordisk took full ownership of the project by acquiring Aradigm Corporation. The project remained in the U.S., organizationally anchored in Novo Nordisk at Product Supply (i.e. Production), but PDS began to play a more active role in the development than the device area had previously done. This so-called AERx project continued to require massive investments. According to some informants, the PDS strategy left little attention to the need for lifecycle management of existing products; that is maintaining the existing products with continuous improvements (often production-wise) or developing new versions with incremental improvements. One informant says “There were skeptical people saying that we would risk losing our leading position in our core area. I guess this partly also happened. For a time we did not develop ‘engines’ for our injection pens. I guess it gave some sort of set-back. On the other hand, top management was not willing to increase the frequency of introducing new pen generations”. An executive informant puts it: “We kind of diluted the classical virtues – our resources within the classical disciplines – in such way that we lost momentum. And at the same time we placed our resources in high risk areas; it was too much ‘blue sky’ – we were too optimistic, we believed the solutions were just around the corner.” A development engineer comments: “We were happy to become a part of R&D; that’s where we belong. But the closed loop visions – well, it’s easy to write down such buzzwords…”. The latter statement represents a critical attitude towards the new strategies, which was widespread in the device area; remember that the strategy plans for PDS were developed outside Medical Systems. A device manager expresses it: “The establishment of PDS was not our choice [from Medical Systems]. It was driven from the top and by consultants with business plans for ambitious product areas – this was driving it, rather than knowledge on technologies. They were very little attentive to allocating resources for maintaining our injection systems. It was business plans all over”. Besides this internal skepticism, the ambitious strategies ran into some other challenges, namely the 2002 crisis (see above). One of my informants comments: “It was an annus horribilis”, and continues: “Then we were in a chaotic situation with regards to financing our R&D activities – what did we really want? – which path should we follow? – with regards to drugs as well as devices. The whole trend was like: perhaps we should move back to basics, meaning only bet on insulin and GLP-1 on the drug side and injection devices on the device side…” This ‘annus horribilis’ had implications for the strategies of PDS. Initially, it was planned to acquire specialized companies within both continuous glucose monitoring and continuous insulin infusion in order to build on their competencies and technologies. The chaotic situation meant a setback in the 118 willingness to invest in such companies. Consequently, the only available way forward was in-house development. However, the complexity and risk of the projects had been underestimated, according to an informant; the technological challenges were much harder than envisioned. Hence, the continuous glucose monitoring project was given up in 2004, which was a de facto strategy change, as seen by a device manager. Furthermore, the drug formulation unit, which had been integrated into PDS, was transferred back to the drug development area to ensure the development of a GLP-1 drug, leaving PDS back with the device competencies solely. The skepticism towards the ‘closed loop’ vision increased and the visionary strategy behind PDS slowly was falling apart: “the dream about the role as system integrator died”, as stated by an informant. “The PDS visions were like romantic dreams”, as another informant stated it. This informant also points to the existence of different power bases behind respectively the initiation and the closure of PDS: The closed loop strategy behind PDS was driven by Operations (production, sales and marketing), who did not focus on the risk of such radical innovation projects – “they did not assess the scientific reality”, as he puts it. According to this informant, the withdrawal from the closed loop strategy was driven by R&D. On top of these internal factors, which gradually made the PDS strategy fall apart, the competition in the traditional insulin device market began to threaten Novo Nordisk’s leading position in devices. As result, reorganization was carried out in the summer of 2005. Concluding on the evolution of the PDS strategy (2001-2005) The second ‘revolutionary’ period (after the MSD period 1988-92) was initiated as result of a reasoning process, which can be traced back to a particular management seminar at an international business school in the summer of 2000. The strategy was then worked out via desk analysis during the next half year, driven by a management team outside the device area, supported by external consultants. The aim was to create advanced ‘closed loop’ systems and to include more elements of the value chain of diabetes care into a ‘one-stop-shop to diabetes’, just like in the MSD period. Hence, this top management driven, theory-based process in content was a truly autonomous strategic initiative; because it explored product-markets outside the existing strategy (cf. Definitions). The vision was again to create a business of its own, based on the competencies within medical devices, which eventually could be demerged as a separate company. Thus, medical devices changed role from complementary to core assets for innovation. Some inputs from the external environment played a role. Other pharmaceutical companies had tried integrating more elements of the value chain the years before (e.g. Lilly buying a distributor company), so vertical integration was seen as a potential trend, which might alter the market. Also the emergence of Continuous Glucose Monitoring in 1999 was seen as a potential threat, in case other companies would be able to develop integrated solutions, which would reduce Novo Nordisk’ role in the value chain to supplier of ‘muddy liquid’. In conclusion, the transition into the PDS strategy was partly ignited by external factors, such as the potential future competitive landscape, in case the competitors succeeded with their initiatives. However, the external drivers were again mediated by management interpretation; top management reacted to the perceived potential risk. 119 The strategic learning cycle of PDS is a straight-forward theory-driven search for technologies, based on the visions of ‘closed loop’ and ‘one-stop-shop to diabetes’, resulting in explorative innovation activities, which never succeeded; hence the reinforcement from the experiences was negative. See figure IV-13. Figure IV-13. The strategic learning cycle of PDS. PDS 2001-05 Closed loop. ‘One-stop- shop’ Exploration & development of advanced systems Negative experiences with advanced systems Negative reinforcement Search for technologies 126 launch of the first durable insulin pen (NovoPen®) in 1985 and the first prefilled insulin pen (NovoLet®) in 1989 all projects within these two categories were autonomous. Once these product categories were established, one could expect the relative number of autonomous projects to gradually decline; until the strategy was altered next time, most activities would aim for developing improved versions of the already established product categories (staying within the strategic frame). If – and only if – the strategy for the device innovation was changed and opened for exploration of new product-markets, one could expect to see changes in the nature of the development portfolio in the form of increased autonomous activity. As known from the historical case study in the previous sections, there was only one such strategic change after the initial institutionalization of the device strategy in 1988, which opened for entering new product-markets: namely the PDS strategy, formulated in 2000, implemented in 2001. The other strategic changes turned the strategy back to already established positions and capabilities within durable and disposable insulin pens. Hence, one could expect two peaks in autonomous activity: the early phase, up to the institutionalization in 1988, eventually including the MSD phase, and then again in connection with the establishment of PDS. The rest of the periods would be expected to be colored mostly by induced activities. Basic measurements The structural changes were set as follows: 1. 1988: Establishment of the Medical Systems Division. 2. 1992: Moving the device area to the Diabetes Care Division, as Medial Systems. 3. 1995: Moving Medical Systems to Production. 4. 2001: Establishing Protein Delivery Systems (PDS), thereby transferring the entire Medical Systems R&D to corporate R&D. 5. 2005: Moving the device R&D activities to Diabetes Research Unit. The strategic changes were set as follows: 1. 1987: The strategy for making devices a separate business area, building on the vision of homecare centered on the patient (since MSD was established in the beginning of 1988, the strategy must have been formed during 1987). 2. 1992: Termination of the homecare strategy, hereunder the disease monitoring projects and the insulin pumps – the new strategy only included insulin pens. 3. 2000: The formulation of the ‘closed loop’ strategy, leading to the establishment of PDS in 2001 (delayed only for financial reasons). 4. 2005: Official termination of the closed loop strategy. Focus now on the ‘value-added product’ as the integrated device-drug system. The restructuring in 1995 is not included in the strategic changes, because the strategy for devices was not altered. The total portfolio of device development projects (not differentiated in autonomous and induced) is shown as the number of ongoing development projects per year in figure IV-16. The number of ‘ongoing projects’ includes all projects in operation that year: some were launched as new products, 127 others were stalled during the year; and some were still ongoing at yearend. Overall, the number of ongoing projects increases throughout the studied period, which is partly to be explained with increased resource input. Interestingly, the development displays recurrent waves, following the structural changes: the number of development projects increases after each restructuring to reach a peak 2 years after (for the MS period 1995-2001 also displaying a second peak 4 years after), and then declines. This pattern shows also in the case, where the restructuring is not accompanied with strategic changes (in 1995). I first interpreted this pattern as the effect of a motivational boost after each restructuring: each reframing ignites entrepreneurial activity; after some years, the ‘frame’ gets worn, and in order to release a new period of growth, a new reframing takes place. This way, the repetitive cycles of expansion and contraction (the ‘pulse beat’) form an overall growth in the project portfolio. Perhaps the metaphor of a hermit crab could be used: it finds a shell for protection and lives in it until the shell gets too constraining; then it seeks a bigger shell and drops the old. When presenting this interpretation of figure IV-16 to a device manager, he replied: ‘this gives a wrong picture of what happens; it is like as if the activity level decreases in the second part of each period. Rather, we focus the activities on a fewer number of projects’. His statement clearly points to the weakness of using the number of projects as a measure: the resource input is not visible. His explanation implies that each restructuring opens a ‘divergent’ period, meaning that the latitude of the activities expand and the possibilities of the new organizational frame are explored – then a ‘convergent’ phase begins, where the portfolio of projects is narrowed down in order to exploit the most viable projects. If this explanation is valid, we should be able to track a pattern of two peaks within each period: the project initiation curve should peak in the beginning of each period, and the project termination curve should peak in the middle or end of each period. I have analyzed the data and indeed, this interpretation was supported: the number of project initiations peaked either by the year of reorganization or the year after. The number of project terminations peaked either in the middle or the end of each period. It therefore seems as if organizational restructurings have an impact on the activities, displayed as lifecycles of exploration and exploitation, reflected in the total number of innovation projects – regardless of whether or not the strategy is changed. 128 Figure IV-16. The total number of ongoing device development projects, year by year. Dotted vertical lines: Strategy changes. Full vertical lines: Structural changes. ‘Ongoing projects’ include all projects in operation that year – some were launched as new products, others were terminated during the year; and some were still ongoing at yearend. The development of the nature of the innovation project portfolio The nature of the project portfolio can be analyzed by using different sets of metrics. One way is to count how many ongoing projects that are in operation simultaneously, year by year, split in the two categories induced and autonomous. This is shown in figure IV-17. 129 Figure IV-17. The portfolio of ongoing projects, split between autonomous and induced. Dotted vertical lines: Strategy changes. Full vertical lines: Structural changes. The graph indicates a cyclic pattern, which is analyzed in the subsequent text. To see the link between strategy/structure and the project portfolio more clearly, each structural/strategic phase can be characterized by the relative distribution of projects in respectively induced and autonomous initiatives. Table IV-2 and IV-3 show this distribution for respectively the structural and the strategic periods. Structural epoch Early 1980-87 MSD 1988-91 Crisis 1992-94 MS 1995-00 PDS 2001-04 DRU 2005-08 Autonomous projects in average, per year 2.7 6.7 5.0 4.7 10.7 11.2 Induced projects in average, per year 1.5 7.0 6.7 9.5 5.5 8.5 Total average, per year 4.2 13.7 11.7 14.2 16.2 19.7 % autonomous per period 65 49 43 33 66 57 % induced per period 35 51 57 67 34 43 Table IV-2. Relative distribution of ongoing projects for each structural epoch. Portfolio of ongoing projects (= number of projects in operation per year) 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 0 1 2 3 4 5 6 7 8 9 10 Year Induced Autonomous 130 Strategy epoch Early 1980-86 Homecare 1987-91 Basics 1992-99 Closed loop 2000-04 Value add 2005-08 Autonomous projects in average, per year 3.0 5.6 4.6 9.8 11.2 Induced projects in average, per year 1.0 6.6 8.5 6.2 8.5 Total average, per year 4.0 12.2 13.1 16.0 19.7 % autonomous per period 75 46 35 61 57 % induced per period 25 54 65 39 43 Table IV-3. Relative distribution of ongoing projects for each strategic epoch. These figures demonstrate that structural and/or strategic changes are in fact correlated with the distribution of development projects between autonomous and induced activities. In both tables, the percentage of autonomous projects starts at a high level (65% for structure and 75% for strategy), where after it decreases as expected until the ‘closed loop’ strategy results in establishment of PDS, where it jumps to a new height (66% for structure and 61% for strategy), to decrease after the termination of PDS. The distance between the lowest and highest percentage of autonomous projects is respectively a factor 2.0 for structural evolution (from 33 to 66%) and 2.1 for strategic evolution (from 35 to 75%). This evolution of the portfolio is graphically shown in figures IV-18 and IV-19. Figure IV-18. The relative distribution between ongoing autonomous and induced projects during the shifting structural epochs. Ongoing projects per structural period 0% 20% 40% 60% 80% 100% Early 80-87 MSD 88-91 MS1 92-94 MS2 95-00 PDS 01-04 DRU 05-08 Induced strategy Autonomous strategy Epoch - starting by year of restructuring 131 Figure IV-19. The relative distribution between ongoing autonomous and induced projects during the shifting strategic epochs. The above statistics were based on the average number of ongoing projects per year. However, one could argue that ongoing projects not necessarily reflect the current strategy; projects are not so easy to kill, when they first have a business case, technological feasibility and a handful of missionaries to fight for their ‘child’ – thus, a project might often survive in spite of a changed strategy. Therefore, you only see a gradual decrease in autonomous activity over several phases, after each peak created respectively by the early attempts and by the PDS strategy. Some of the autonomous projects ‘hang out’ after the bar had closed, so to speak. Following this logic, the impact of strategy on the product innovation portfolio is best obtained by the number and nature of the projects initiated, year by year. This is shown below in figure IV-20. Average ongoing projects per strategic period 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Early, 80-86 Homecare, 87-91 Basics, 92-99 Closed loop, 01-04 Value add, 05-08 Strategy period, by year of strategy formulation Induced strategy Autonomous strategy 132 Figure IV-20 The number of initiated projects per year, split between autonomous and induced. Dotted vertical lines: Strategy changes. Full vertical lines: Structural changes. The evolution is analyzed in the following. Furthermore, similar calculations as above have been made concerning the relative percentage of autonomous projects among the population of projects initiated year by year. Table IV-4 and IV-5 below display the results. Structural epoch Early 1980-87 MSD 1988-91 Crisis 1992-94 MS 1995-00 PDS 2001-04 DRU 2005-08 Autonomous projects initiated in total 6 9 2 9 12 7 Induced projects initiated in total 5 9 8 16 4 8 Total initiated projects per period 11 18 10 25 16 15 % autonomous projects initiated per period 55 50 20 36 75 47 % induced projects initiated per period 45 50 80 64 25 53 Table IV-4. Relative distribution of projects initiated for each structural epoch. Number of projects initiated per year 0 2 4 6 8 10 12 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 0 1 2 3 4 5 6 7 8 Year Induced Autonomous 133 Strategic epoch Early 1980-86 Homecare 1987-91 Basics 1992-99 Closed loop 2000-04 Value add 2005-08 Autonomous projects initiated in total 6 9 10 13 7 Induced projects initiated in total 4 10 22 6 8 Total initiated projects per period 10 19 32 19 15 % autonomous projects initiated per period 60 47 31 68 47 % induced projects initiated per period 40 53 69 32 53 Table IV-5. Relative distribution of projects initiated for each strategic epoch. The observation that the initiated projects might provide a clearer ‘portrait’ of a strategic/structural period than the portfolio of ongoing projects is confirmed in the sense that the variance between the values is higher for the initiated projects. The pattern is the same: Starting with a high degree of autonomous activity, then decreasing until a jump in connection with the formation of PDS, thereafter decreasing again. The structural periods display the period of crisis in 1992-94 as a significant low of only 20% autonomous activity, whereas PDS jumps to an all-time high of 75% (a variance of a factor 3.75). The strategic periods have same pattern, but vary less; from 31 to 68% (factor 2.2). This difference in variance can be explained: the structure was changed in 1995 (to a new functional corporate structure) after having overcome the quality crisis, whereas the strategy for devices was kept focused on basic insulin devices all the way from 1992 to 2001. Since the strategy period thus covers a longer time span, the variance in portfolio gets leveled. The patterns are visually depicted in figures IV-21 and IV-22. 134 Figure IV-21. The relative distribution between initiated projects in autonomous and induced during the shifting structural epochs. Figure IV-22. The relative distribution between initiated projects in autonomous and induced during the shifting strategic epochs. Initiated projects per structural epoch 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Epoch - starting by year of restructuring Early 80-87 MSD 88-91 MS1 92-94 MS2 95-00 PDS 01-04 DRU 05-08 Induced strategy Autonomous strategy Initiated projects per strategy period 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Induced strategy Autonomous strategy Early, 80-86 Homecare, 87-91 Basics, 92-99 Closed loop, 01-04 Value add, 05-10 Strategy period, as per year of formulation 135 Conclusion on the development of the project portfolio The composition of the project portfolio developed as expected: The ratio of autonomous projects was high at the entrance to the medical device activities, and then decreased until a peak in connection with the launch of the new autonomous PDS strategy in 2001. After termination of the PDS strategy, the ratio of autonomous projects decreased again. So, the product innovation activities (strategy in ‘practice’) and the formulated strategic intent (strategy in ‘theory’) were actually linked. When analyzing the interrelatedness between strategy in ‘theory’ and strategy in ‘practice’, we can do it both ways:  Going out from practice (i.e. project portfolio): For the total portfolio of ongoing projects (i.e. the total number of projects in operation in a given year), the strategic epochs provide higher variance in the balance between autonomous and induced projects than the structural epochs. For initiated projects, the structural epochs show higher variance than the strategic epochs; this can partly be explained by shorter lifecycles of organizational structures compared to the strategy lifecycles.  Going out from ‘theory’, as expressed either in strategic or structural epochs: The initiated projects provide higher variance in the balance between autonomous and induced projects than the total number of ongoing projects, for both strategic and structural epochs. This stronger link between strategic intent and project initiation (compared to the total ‘stock’ of projects) confirms the basic idea that new strategic visions imply new activities. We can also conclude that it is in fact possible to establish an assessment of the balance between autonomous and induced activities using the composition of the project portfolio as indicator: the nature of the project portfolio, classified in autonomous and induced projects, depends on the nature of the innovation strategy. Autonomous strategies indeed cause a significant higher ratio of autonomous innovation projects. An interesting observation is the cyclic development: that each organizational restructuring implies a period of exploration, where many new projects are started up, until a peak in project closures marks a period of exploitation, in which the number of project initiations decreases (because the portfolio is focused), until the next restructuring starts a new cycle. These cycles in the number of initiated projects seem independent of whether or not the strategy is changed. 142 The overall pattern in the evolution of innovation strategy If we synthesize the phases of development of device innovation strategy as analyzed in the previous chapter, the development can be defined as five overall periods, divided by the transformations of device innovation strategy (see empirical definition in the methodology chapter), namely: 1. The early attempts with medical devices at the two companies, Nordisk and Novo (1980- 88), where Nordisk started with insulin pumps and Novo with insulin pens. These were entrepreneurial activities, which still lacked a formal strategy. Thus, in Burgelman’s (1988) terminology, the period represents the emergent state; before institutionalization of strategy. 2. The strategic vision of patient-centered homecare, based on a medical system, anchored in the new Medical Systems Division (MSD). MSD was established in 1988 as a business unit, which was envisioned to establish devices as ‘the third business leg’. Devices should create revenue streams of their own – partly by selling devices such as insulin pumps, partly by manufacturing devices for other pharmaceutical companies. The activities comprised drug delivery (pens and pumps) as well as disease monitoring devices (e.g. glucose monitoring). The MSD strategy was terminated in 1992. 3. The focusing period, in which insulin pumps were sold off and disease monitoring projects were stalled – focus was directed explicitly at insulin pens, and devices were to support the drug business. Devices were not envisioned to create revenue streams of their own. This phase was partly evoked by the quality crisis in 1992-94, and partly by Novo Nordisk being behind competitors with regards to the insulin pipeline. Thereafter, as result of the corporate restructuring into a functional organization, the device activities in 1995 were transferred to Production, keeping the name Medical Systems (MS), without changes in strategy – focus was still on insulin pens only. This organizational setup was kept until mid 2001. 4. In 2000, the new strategic vision for closed loop and ‘one-stop-shop’ to diabetes was conceptualized. The implementation was postponed until 2001, where Protein Delivery Systems (PDS) was established as the organizational frame for the new holistic device innovation strategy. Thereby device R&D was separated from device production, which remained under the corporate Production unit. The strategy included both insulin delivery and continuous glucose monitoring and envisioned substantial revenue streams based on devices, including complex and expensive ‘closed loop’ systems. 5. A crisis already in 2002 undermined the risk willingness needed for the ambitious visions of PDS, and in 2004 the continuous glucose monitoring project was terminated; but the strategy was not officially changed until 2005, where the innovation strategy returned to a ‘back to basics’ mode, i.e. focus should be on insulin pens, like in the MS period. The device innovation activities were integrated into a drug research area, Diabetes Research Unit (DRU). By year-end 2008, DRU was terminated. 143 The strategic learning cycle of each strategic period was analyzed in the previous chapter. These cycles are depicted in figure V-1 below (as in Chapter 4, the focusing period of MS has been split in two sub-phases: the crisis 1992-94 and the ‘harvesting’ period 1995-2001). Figure V-1. The strategic learning cycles, phase by phase, as extracted in Chapter 4. The strategic learning cycles in figure V-1 were useful in Chapter 4 for analyzing the strategy-making processes of each phase. They also provide evidence for the ‘theory’-driven scenario of strategic reconfiguration, as shown in figure II-14. In the current case study, the strategic changes implied by respectively MSD and PDS did not begin with negative reinforcement of the existing strategy; rather, Devices as market differentiator Product innovation Success in differentiation via devices Harvesting 1995-2001 Search of differentiation opportunities MSD 1988-92 3rd business leg. Vision of homecare via medical systems Exploration & innovation Mixed experiences Search of growth / new business via devices Failure of advanced systems  ‘Back to basics’ Exploitation of classic skills Partly successful Search of classic skills DRU 2005-08 PDS 2001-05 Closed loop. One-stop- shop Exploration & development of advanced systems Negative experiences with advanced systems Negative reinforcement Search for technologies The crisis 1992-94 Core competencies Closure of pumps and monitoring projects Success in finding core competencies Positive reinforcement Search for core/ non-core competencies Early attempts 1980-88 NG: Artificial pancreas. Novo: Devices as part of ‘care’ NG: pumps. Novo: Pens Succes Positive reinforcement Search for technologies and means of differentiation 144 these learning cycles were driven by forward-looking search for exploration. Such vision-driven strategic search does not resemble the normal behavioral understanding of strategy change as result of bottom-up processes. However, these individual learning cycles are perhaps less useful for identifying patterns across the phases. To do so, the device innovation strategies can be classified as falling in two distinct modes: A. Autonomous, device-based strategy, including new business creation and a systemic or integrated approach to innovation. These strategies are ‘autonomous’, because they aim for creating new product-markets (see Definitions in Chapter 1). B. Induced, drug-based strategy, focused on devices as complementary to the drugs. The result of this distinction can be seen in figure V-2. Figure V-2. A model of the development of device innovation strategy at Novo Nordisk. The ‘early attempts’ have a dotted outline, since these activities represent the ‘emergent state’ (Burgelman, 1988) – before institutionalization of strategy. The model of the development, shown in figure V-2, of course leaves out many nuances of the strategic phases; that is the sacrifice you make in search of clarity from models. That being said, the cyclic nature of the development is striking. Generally spoken, one can identify the balance between autonomous and induced strategic activities in an a company at a given time – the analysis of the innovation project portfolio in Chapter 4 does so, and Burgelman & Grove (2007) do it for Intel Corporation at different moments of the company’s lifetime. Such analyses reflect a balance between autonomous and induced strategy ‘in space’ at a given moment. But only the longitudinal lens can provide the image of the balance between autonomous and induced strategy over time, as shown in figure V-2. It is apparent that the two ‘revolutionary’ periods (MSD 1988-1992 and PDS 2001-05) aim for innovation at a more systemic level than the normal periods, which focus on the physical product itself. As put by a PDS manager: “Novo Nordisk could get a bigger share of the pie via diversification into other parts of the diabetes market. And the executive committee saw that if anybody could do this, Mode A: • Autonomous strategy; new business creation • Device-based strategy including glucose monitoring • Systemic or integrated innovation for patient care Early attempts PDS 1980-88 1988-92 1992-2001 MSD 2001-05 2005-08 MS Mode B: • Induced strategy; product innovation of injection devices only (mostly pens) • Drug-based strategy • Devices complementary to drugs Vision of homecare Vision of closed loop Focus on pens Focus on pens Pumps and pens DRU 145 it was us”… “We wanted to be a ‘one-stop-shop’ to diabetes”. In such view, the physical product is one element in or component of the innovation. The contrast between system and component has been expressed in theories on ‘architectural innovation’. This concept has been proposed at two levels of analysis. Abernathy & Clark (1985) describe ‘architectural innovation’ as follows: “Innovation of this sort defines the basic configuration of product and process...it lays down the architecture of the industry” (p. 7, my emphasis). “Using new concepts in technology to forge new market linkages is the essence of architectural innovation.” (p. 10, my emphasis). In a later paper, Henderson & Clark (1990) describe architectural innovation differently: “We define innovations that change the way in which the components of a product are linked together, while leaving the core concepts (and thus the basic knowledge underlying the components) untouched, as “architectural” innovation” (p. 10, my emphasis). In Abernathy & Clark (1985) ‘the system’ includes market linkages, whereas ‘the system’ in Henderson & Clark (1990) is limited to the physical product itself. (For instance, they use an electric room air fan to illustrate ‘the system’). The two meanings of the concept are reflected in the case study at hand: the two ‘revolutionary’ periods (mode A in figure V-2) aimed for reconfiguration of market linkages, corresponding to the former definition of ‘architectural innovation’; the ‘normal’ periods (mode B in figure V-2) only aimed for reconfiguration of the product itself, referring to the latter definition. The challenges for corporate management seem to occur when the architecture of the industry is in play, i.e. when more elements of the value chain are included. In contrast, the architecture of the product as a system could be reconfigured throughout the entire period without disturbing the corporate strategy, so long as the product was defined as a system for drug injection. The case of medical devices at Novo Nordisk to a large extent is a case of integration of a new set of complementary assets for innovation. Nordisk had a vision of making medical devices a ‘third business leg’ (beside the two drug businesses: diabetes and biopharm), so here the ambition was to make medical device competencies a core asset for innovation. Novo, in contrast, had no such visions, when they launched NovoPen®; the medical devices were seen as purely complementary assets. The start-up of the medical device activities, however, was like letting in a Trojan horse – soon, the medical device activities strived for independence and for becoming a business of its own. Both autonomous strategic waves, respectively MSD 1988-92 and PDS 2001-05, made the attempt to establish medical devices as the ‘third business leg’. These cycles of innovation strategy show that the distinction between core and complementary assets for innovation is dynamic and negotiable, not static. The potential implications of the two ‘revolutionary’ strategies were profound, had they been realized; the consequences would have reached far beyond the device area alone. Why did corporate management at all let the medical device activities try to escape the role as complementary enhancers of the pharmaceutical drug? Well, perhaps it was exactly because the medical devices were seen as complementary assets that the corporate ‘immune system’ was not provoked – had medical devices from the outset been seen as a new business area, the corporate filtering mechanisms would perhaps have been more alert. The ‘camouflage’ as complementary assets made the experimental 146 strategies slip through the filter – since the strategic experiments were not perceived as a threat to ‘core business’. A veteran device manager said: ““Why we succeeded with the only radical innovation ever in Novo Nordisk – the NovoPen® – ‘well, it sounds like a good idea, and it doesn’t really cost anything, and some 7,000 patients might benefit from it, why not try it’. So, in reality it was because we started in such small scale and with so small consequences for the firm that it didn’t really matter… it was so manageable and easy to oversee that it just slipped through. Management stalls if it something is so big that you risk the whole enterprise”. This context of integration of complementary assets differs from Burgelman’s research. Burgelman (1991) described the intra-organizational ecology of established and emerging businesses: in case, the established memory business and the emerging microprocessor business at Intel Corporation. The case of Novo Nordisk it not about two competing businesses, but about integration of some new complementary assets for innovation, which then repeatedly are utilized for visions about establishment of new business areas. Even though the two case contexts differ, the integration of new complementary assets for innovation at Novo Nordisk seems to generate managerial challenges just as serious as the integration and portfolio management of new business ventures, as demonstrated in the case of Intel. Still, the challenges are different. Where the management of alternative businesses requires portfolio management competencies, then the management of dynamics between core and complementary assets for innovation requires integrative competencies (Christensen, 2006). It’s interesting here that throughout the entire period of the case study at hand, devices and especially utensils like injection needles have in fact have created revenue streams, which – seen in a Danish context – are substantial. However, within the normal (drug-centered) strategic periods, this income is not seen as business – it is not accounted for separately and there’s not made a profit and loss account for devices 16 . In other words, the actual value creation has existed ever since the introduction of insulin devices, understood as revenue streams; but these revenue streams only became explicit elements of the strategy within the autonomous periods, where the income from the medical devices and utensils for these was part of the strategic framing. This fact elucidates the role of cognitive frames in strategy. The development of innovation strategy, as shown in the model of figure V-2, is the result of change processes, evoked by different drivers of change. We shall now turn towards these drivers; i.e. the events and processes of change behind the strategic transformations. 16 At least, if such accounts exist, they are not communicated in the organization. 147 The drivers of change Based on the analysis in Chapter 4, table V-1 synthesizes the events and processes behind change in innovation strategy, transformation by transformation, divided into external and internal drivers (‘drivers’ understood here as events and processes). External drivers are split in a) Industry or market trends b) Technological discontinuities c) Competition d) Externally evoked crises. Internal drivers are divided into four dualities: e) The origin of the strategy: stemming from forward-looking cognition or back-ward looking experience. f) The ‘gravity’ towards the established strategy: does the strategic initiative stay within the established strategy = induced strategy, or does the strategic initiative explore new productmarkets, outside the established strategy = autonomous strategy? g) The integration of devices: are medical devices seen as core assets for innovation, to create a business of its own – or as complementary assets for innovation, to support or enhance the sales of the pharmaceutical drug? h) The actor: is the change process driven by top management or bottom-up, i.e. organizationdriven? Like the model in figure V-2, the scheme in table V-1 does not capture all details, which were analyzed in the previous chapter, but provides an overview. For each strategic transformation, I have emphasized the most salient drivers of change. 148 Drivers of change of innovation strategy – transformation by transformation Strategic phase (year of start) 1) Early attempts with medical devices 2) MSD: vision of homecare (1987-88) 3) MS: focus on injection pens (1992) 4) PDS: vision of closed loop (2000-2001) 5) DRU: back to basics (2005) Nordisk (1980) Novo (1981) Crisis (1992) Harvesting (1995) External drivers Industry or market trends Diabetes specialists discuss artificial pancreas and pumps Vertical integration tried out (e.g. Lilly buys a distributor) Cancer risk by tablet treatment  renewed focus on insulin Technological discontinuities Recombinant insulin (1982) Continuous Glucose Monitoring (1999) Competition Eli Lilly 1982: recombinant insulin (in development from 1976) Competitors ahead in insulin pipeline Pharma giants interested in diabetes Fierce competition on insulin pens Externally evoked crises FDA critique  GMP crisis Financial shake 2002 + failure of tablet project Internal drivers Origin Cognitive search (forwardlooking) Seeing pumps as opportunity Broader goal of ‘care’ + analyzing pumps as opportunity The vision of patientcentered homecare Need to focus on ‘core competencies’ Wish for staying independent as company  corporate focus strategy Desk analysis  vision of closed loop and ‘one-stop- shop’ to diabetes Experiencebased search (backwardlooking) (Learning from first pump – building up inhouse competences) Learning from Diagnostics + (Trying out pens – impact not foreseen) Building on positive experiences with pens and pumps. Learning on the way. Mixed results of MSD strategy Success in finding core competencies Learning from mistakes (problems with advanced devices + delay in classic insulin pens) Gravity towards established strategy Autonomous (exploration outside) Pumps as new business = revenue generator Pens announced as a new paradigm New business unit for devices as a ‘third business leg’ Creating new business – ‘value upgrade’ of market Induced (expanding within) (Pumps expand insulin business) Pens as market differentiator Focus on classic strategy – devices seen as market differentiator Continued Classic strategy: ‘back to basics’ Integration of devices Devices core assets Pumps as new product Devices core, as a ‘third business leg’ Devices core for new business / ‘value upgrade’ Devices complementary assets Pumps expand and support insulin business Pens as a marketing tool Complementary assets, serving drug divisions with injection devices Devices only seen as support to drugs Continued Devices seen as ‘valueadding’ Actor Top management Top (CEO) Top (marketing executive) Top (NG CEO) Top (executive committee) Top (Operations) Top (executive committee) Organization (bottom-up) Local MSD management Relative freedom (outside radar of top management) Table V-1. Synthesized overview of the drivers (i.e. events and processes) of change of device innovation strategy at Novo Nordisk, transformation by transformation. Based on the analysis in chapter 4, the key drivers are identified and marked here as emphasized and underlined. The pattern which emerges is:  The first entrepreneurial initiatives with medical devices (early attempts, 1980-88) were driven by external events: the discussion amongst diabetes experts on making an artificial pancreas, and the market introduction by Eli Lilly of recombinant insulin.  The MSD strategy implemented 1988 was driven by internal processes: top management cognition (the vision of patient-centered homecare) and autonomous strategy.  The retreat from the MSD strategy in 1992 was driven by external events: FDA’s criticism leading to the quality crisis, and competitors being ahead of Novo Nordisk with regards to the insulin pipeline.  The next transformation, implemented with PDS in 2001, was driven by internal processes: top management cognition (the vision of closed loop) and autonomous strategy.  The gradual withdrawal from the PDS strategy, ending in 2005, was driven by external events: the financial shake in 2002 combined with increased competition on insulin pens. Interestingly, the pattern in this case study clearly shows that strategic ‘revolutions’, which differ from the established strategy and from the underlying dominant logic (Prahalad & Bettis, 1986), are result of a process where top management cognition creates autonomous strategy. In contrast, the retreat from these ‘mode A’ periods (see figure V-2) is defensive; caused by externally evoked crises. In conclusion, the formula extracted from this case study goes: Top management cognition drives the strategic renewal; external events drive the strategic retreats. I have modeled the overall evolution in figure V-3 below. Figure V- 3. Model of the evolution of innovation strategy of medical devices at Novo Nordisk, displaying the most salient external and internal drivers of change and their resulting innovation strategies. Device level innovation strategy Early attempts 1980-88 1988-92 1992-2001 MSD 2001-05 2005-08 MS Vision of homecare Vision of closed loop Focus on insulin pens Focus on insulin pens Insulin pumps and pens Corporate level cognition and strategy formulation Events in the external environment Diabetes specialists discuss artificial pancreas and insulin pumps. Eli Lilly introduces recombinant insulin (1982). The vision of patient-centered homecare. Devices as ‘3rd business leg’. Setting up new division for devices. Competitors ahead in pipeline of insulin analogues. FDA’s critique  GMP crisis. Interpretation: Focus on core competencies. Stopping insulin pumps and disease monitoring device projects. The vision of closed loop and ‘one-stop- shop’ to diabetes. Separating device R&D from device production. Fierce competition on insulin pens. Financial shake 2002 + failure of tablet project. Vertical integration tried out (e.g. Lilly buys a distributor). Continuous Glucose Monitoring (introduced 1999). Interpretation: Devices = new opportunities. Setting up development activities. Autonomous Induced Market success with first devices (Nordisk: insulin pumps – Novo: insulin pens). 3rd generation insulin device PDS DRU Interpretation: ‘Back to basics’ – away with small molecules and complicated devices. Device R&D integrated into drug R&D. Copenhagen Business School, DBA project No. 888-31483. Arne Stjernholm Madsen, © 2012. 150 Discussing theory in the light of the case study The role of management cognition The development in the present case study displays a cyclic pattern, unfolding two waves of institutionalized, yet autonomous innovation strategy based on devices (the ‘mode A’ waves in figure V-2: MSD 1988-92 and PDS 2001-05). In both cases these autonomous waves lasted about 4 years, after which the innovation strategy was turned ‘back to basics’, i.e. the traditional drug-centered strategy. The two autonomous waves were first and foremost initiated as result of top management cognition and forward-looking search, and thereby the case differs from the widespread behavioral view on top management cognition seen as a filtering mechanism, which unfolds cognitive inertia and hinders strategic renewal. In the case of Novo Nordisk, top management driven cognitive search had a clearly entrepreneurial role by initiating the two autonomous waves – until the crises occurred, where the filtering mechanisms came into play. The autonomous strategy waves thereby counteracted the forces of ‘co-evolutionary lock-in’ (Burgelman, 2002). In what sense did the autonomous strategies of MSD and PDS then differ from the established strategy? Here, we must look all the way back to the initial identity of the company in the 1920’s, rooted in science and medical treatment of diabetes with the newly discovered insulin. In general, the dominant management logic (Prahalad & Bettis, 1986) of a company is built by positive reinforcement of successful activities. In the case of Novo Nordisk, this positive reinforcement led to the logic of a pharmaceutical company, based on manufacturing of pharmaceutical drugs, primarily insulin. In the 1970’s, Novo changed the overall espoused goal from world leader in insulin manufacturing to world leader in diabetes care. The term “diabetes care” covers treatment services, disease monitoring and many other things beyond the pharmaceutical drug itself – the drug is just one component of the value chain of care. Hence, the goal of being a leader in “diabetes care” should in theory open the business for many more areas than insulin manufacturing. However, it does not seem as if the corporate dominant logic ever took the shift in system level from insulin manufacturing to diabetes care; at least, the only business attempts to realize the higher system level vision of ‘diabetes care’ were exactly the two device-based innovation strategies of MSD and PDS. These strategies contradicted the pharmaceutical dominant logic by not setting the drug in the center of the business. Consequently, these strategies were fragile – and at each externally evoked crisis, the corporate strategy bent back to the safe ‘fetal position’, focused on the pharmaceutical drug itself. The dominant logic thus seems deeply rooted in the internal identity (Tripsas, 2009) of the firm. Even in 2007, where I conducted the interviews at device management level, the extracted dominant logic was pharmaceutical; not device-based or ‘diabetes care’-oriented. The pharmaceutical business logic and internal identity seem to form both the starting point and the ‘safe harbor’ of the company, in case strategic experiments fail. This observation reflects the equilibrium theory of the dominant logic, formulated in Bettis & Prahalad (1995). If we apply their analysis as shown in figure II-10, neither of the two autonomous strategic waves at Novo Nordisk (respectively the MSD and the PDS Copenhagen Business School, DBA project No. 888-31483. Arne Stjernholm Madsen, © 2012. 151 strategy) achieved momentum enough to escape the gravity of the dominant logic before the two crises started the process of retreat. In both cases, the strategic visions were never fully implemented; partly because the needed product systems could not be developed within the available timeframe of four years. When external crises then reduced corporate risk willingness, the inherent uncertainties of these explorative ventures made corporate management withdraw from the autonomous, device-based strategies. One should also consider the observation from Prahalad & Bettis (1986) that the more successful a company has been, the more difficult it is to unlearn the dominant logic (p. 498). Novo Nordisk had indeed been very successful in the insulin industry, and consequently it was more difficult to change the prevailing paradigm of thinking. Imagine what it would have taken to make corporate management implement for instance the PDS strategy corporate-wide. Corporate management would have had to reformulate the strategy towards a device-based or systemic business strategy, based on complex treatment solutions, in which the drug was seen as one of many components in the total product offering. Such dramatic move would have required an immense success of the PDS strategy, implying revenue streams at least of the size of the revenue from the drugs. In other words, the envisioned ‘value upgrade’ of the market should first have proven to be viable. One can compare to Intel Corporation, where the official strategy was not changed from memory to microprocessor business until the microprocessor business had climbed to app. 75% of the corporate sales (Burgelman, 1991). This strategic inertia at Intel was partly a result of an internal identity, which was closely tied up with the memory business (Tripsas, 2009). If we return to Novo Nordisk, you should remember that an important element of the autonomous, devicebased strategies of MSD and PDS was the inclusion of glucose monitoring products. Even though glucose monitoring devices are needed in diabetes care, these devices do not directly support the sales of insulin, and they were seen as alien to the corporate ‘DNA’. For example, one veteran device manager told me that an executive manager had commented the inclusion of glucose monitoring devices back in the MSD period with the words: “take that crab out of the portfolio!”. This resistance is caused by the history: glucose monitoring products were not a part of the original pharmaceutical identity and had not been part of the learning cycles, which created the corporate dominant logic. Hence, strategies building on inclusion of glucose monitoring products required a ‘far move’, as described by Bettis & Prahalad (1995), in order to succeed in escaping the established dominant logic (cf. figure II-10). The reader might here make the objection to my analysis that the destiny of the two autonomous strategic periods (MSD and PDS) was instead result of lacking organizational plasticity, which characterizes mature companies, according to Gavetti & Rivkin (2007). However, the organization of Novo Nordisk did in fact adapt to the new strategies: new organizational structures were built, substantial new competencies were acquired and integrated, and new innovation projects were set up. So the ‘structural context determination’ (Burgelman, 1991) was in place; and still the new strategies failed, because the corporate dominant logic pulled the strategy back to the safe harbor, whenever external events threatened the ‘core’ business.