Managing technology intelligence processes in situations of
radical technological change
Technology and Innovation Management, Swiss Federal Institute of Technology Zurich (ETHZ), Kreuzplatz 5, 8032 Zurich, Switzerland
Received 2 August 2006; received in revised form 2 October 2006; accepted 3 October 2006
For established companies, radical technological change is not only a challenge, but it also constitutes a major source of failure. By establishing effective technology intelligence processes, companies may react to radical trends in time which is a prerequisite for coping with technological change. Therefore, this study analyzes the technology intelligence processes in 25 multinational companies in the pharmaceutical, telecommunications equipment and automobile industries in the context of radical technological change. In the three industries, the technologies combinatorial chemistry, dense wavelength division multiplexing (DWDM) and fuel cell are used as settings to analyze these processes on the technology level against the background of the company-level perspective. By applying this complex view, which allows to take into account interactions between different organizational mechanisms and between different hierarchical levels inside a firm, three types of organizing technology intelligence processes can be identified: the participatory, the hybrid and the hierarchical technology intelligence process. The organization of the technology intelligence process according to the three types is influenced by the corporate culture and the decision-making style of the companies. Furthermore, industry differences are identified which may be explained by different rates of radical technological change in the industries. This study suggests that more complex and differentiated views on radical technological change, on corporate technology intelligence processes and on the variety of organizational structures involved in these processes are required.
© 2006 Elsevier Inc. All rights reserved.
Keywords:Technology intelligence; Radical innovations; Disruptive technologies; Technological discontinuities
⁎Tel.: +41 49 721 942 6897.
0040-1625/$ - see front matter © 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.techfore.2006.10.001
The management of technological change in companies is of high interest to both theory and practice
. Incremental technological change introduces relatively minor changes on the existing products, exploits the potential of the established design and often reinforces the dominance of established firms. Radical technological change, in contrast, requires a new set of engineering and scientific principles and can open up new applications and market segments which may challenge the position of established firms
[2–4]. Many studies show that the insufficient reaction of established companies to radical technological change can lead to their demise [2,5–10]. As main reasons for the limited learning capability of established companies, researchers have identified insufficient information on technological trends and managerial incompetence[11–13].
In order to identify technological discontinuities in an early stage and to increase the effectiveness of technological decision-making, many authors[14–17]have called for a more systematic observation of technological trends already in the early 1970s. In the literature, many terms are used for this process of systematic acquisition, assessment and communication of information on technological trends in order to detect opportunities and threats in a timely manner. The expressions range from technology monitoring, technology assessment and technology forecasting to technology intelligence. In this paper, the term technology intelligence is used because it has been increasingly used in recent years by both researchers and practitioners.
Besides the great importance that is attached to technology intelligence, existing research is contradictory on how this process should be coordinated in situations of radical technological change. Therefore, the present study seeks to deepen our understanding of technology intelligence by analyzing such processes in the context of radical technological change. Regarding theory, the study derives further insights into the reasons for the failure of companies. Regarding practice, it identifies different forms of organizing the technology intelligence process which might help to reduce the probability of organizational failure through managerial action. In Section 2, past research on technology intelligence and on related research fields, such as technological change and informal information flow in R&D, is discussed. In Section 3, the research design is outlined. In Section 4, the results of the empirical study of technology intelligence in the context of radical technologies are presented. Section 5 describes implications for research and management, and Section 6 draws a conclusion.
2. Past research
In prior works, many reasons have been brought forward to explain why some companies master radical technological change and others do not. The most often cited are managerial incompetence, organizational inertia, insufficient financial resources, a rigid organizational culture, and insufficient technology intelligence[7,10–13,18–25]. As many existing studies treat companies as one entity, which reacts too late to relevant trends, the black box of technological information gathering, assessment and decision-making has not really been opened in many cases[8,12,26,27]. Furthermore, a detailed analysis of studies that were carried out in different industries shows that the authors come to different conclusions concerning the ability of companies to master radical technological change.
While Utterback, Tushman and Andersonand Henderson and Clarksee only a low ability of companies to manage radical technological change in their studies in mature industries, Christensen and Rosenbloom  show in their study of the disk-drive industry that companies can manage
fundamental technological change as long as it is perceived of value in their value network. Based on studies in the pharmaceutical industry, Zucker and Darby , Thomke and Kuemmerle , and Rothaermel  also emphasize that many companies have mastered radical technological change. Therefore, industry characteristics, such as the existence of specialized supplier networks, are increasingly regarded as factors which may facilitate the management of radical technological change[22,31–33].
Even though the studies that have been cited above come to different conclusions regarding the ability of firms to manage radical technological change, a systematic technology intelligence process is unanimously considered a major factor in reducing the risk of organizational failure in the face of radical technological change. Prior research on technology intelligence has often focussed on the conceptual description of the different steps of the intelligence process[34,35], which always includes the steps of acquisition, assessment and communication of information. In addition, most authors distinguish between an undirected perspective of technology intelligence, the so-called scanning, and a directed perspective, the so-called monitoring[34,35]. Most empirical studies on technology intelligence[36–44]show that specialized technology intelligence departments exist in many companies but that their output is rarely used adequately. As a consequence, Lichtenthaler[45–47]and Regeremphasize that it is necessary to simultaneously take into account structural, hybrid (project-based) and informal forms of coordination of technology intelligence. Furthermore, Lichtenthaler  gives a detailed overview on the different organizational elements of the three forms ranging from listening posts to full-time scanning specialists. However, there is a lack of knowledge on how these three forms of coordinating technology intelligence processes should interact in order to reduce the probability of failure in the case of radical technological change. Furthermore, it has been shown  that technology intelligence processes tend to be more formalized and participatory in more dynamic industries.
The stream of research into informal information flows in R&D departments, which can be interpreted as an informal type of technology intelligence, often analyzes situations of radical technological change. Unfortunately, it neglects the existence of formalized technology intelligence processes. Rather, the importance of specific individuals in information gathering, such as gatekeepers, is emphasized[49–51]. Furthermore, it is shown that innovations and radical trends often have to be pushed by technology and power promotors or champions against internal opposition in order to produce innovations[52–55]. Besides technology and power promotors, some studies identify process promotors, who translate the initiative of the technology promotor into a language that is understood by top management. As this field of research addresses only informal activities and considers either information gathering or the role of promotors, it covers only parts of the technology intelligence process. Nevertheless, it may be derived that hierarchical levels contribute differently to the technology intelligence process in situations of radical change.
Thus, it can be stated that there is a lack of research on the organization of the technology intelligence process in situations of radical technological change. Current research has not taken a holistic view on the technology intelligence processes in situations of radical technological change which covers all process steps from information acquisition to decision-making. Furthermore, it is not understood how hierarchical levels and the different forms of coordinating the technology intelligence process interact and how they influence the quality of the process and the quality of decision-making. Moreover, industry characteristics seem to influence the quality of the technology intelligence process, which has not been taken sufficiently into account in prior studies. As a consequence, the objective of this cross-industry study is to identify different organizational forms of technology intelligence processes in situations of radical technological change and to analyze their impact on the effectiveness of technology intelligence and of organizational decision-making.
3. Research design
It was the objective of this study to analyze the technology intelligence process in the context of radical technological change. Insight into the antecedents of the ability of companies to cope with radical technological change and into the organization of technology intelligence processes should be derived. Along the process steps, ranging from information acquisition, information analysis, information communication to decision-making and its enforcement, the interaction between structural, hybrid and informal technology intelligence processes should be analyzed. Also the importance and interaction of different hierarchical levels in the technology intelligence process should be studied. Furthermore, industry differences in the technology intelligence process should be explored.
As decisions on innovations are not taken in an ad-hoc style at one moment in time [57–60], the technology intelligence process was conceptualized as a continuous process in the framework for this research (Fig. 1). In the first step, the so-called scanning phase, some employees become aware of a technology and communicate this trend to top management, which decides on its relevance. In the following phase, the technology that has been identified is observed, and possible new trends concerning the technology are detected. This monitoring phase can be run through several times. Each time a new trend is identified, it is communicated to top management, and its relevance is assessed. The scanning and monitoring phases are followed by a decision of top management, which may decide to either build up or not to build up a competence in the particular technology. Along the process, top management may also adjust the resources for this technology.
The case study method[61,62]was used as it permitted a holistic view on the complex research object. In order to analyze both the organization of the technology intelligence process in different phases of the technology life cycle and the interaction between formal, hybrid, and informal coordination forms, a longitudinal two-level research design was chosen [63,64]. The technology intelligence process was
studied on the level of the whole company and on the level of specific technologies. From the view of the company, the overall processes and structures of technology intelligence were studied. On the technology level, the technology intelligence processes of the companies were studied in the context of a specific technology. The study focussed on the technology level against the background of the companies' technology intelligence processes in order to identify relevant context factors. However, only the simultaneous analysis of the micro level of technology intelligence on the technology level and of the macro level of technology intelligence on the corporate level allowed drawing implications for the organization of technology intelligence in general. This meso-research design  also permitted bridging the gap between the macro level research on technological discontinuities and the micro level research on informal technology intelligence on the technology level that has been described above.
Regarding the perspective of the company, specialists from the technology intelligence and technology acquisition units as well as some of their clients in top management, such as the head of research or the CTO, members of middle management and individual researchers were interviewed. Hereby, an in-depth understanding of the technology intelligence process was developed, especially concerning the distribution of tasks between different formal and informal participants in the process. Furthermore, it was possible to shed light on the view of the different hierarchical levels regarding the effectiveness of the technology intelligence process. On the technology level, up to eight persons were interviewed, among them the person who first became aware of the technology but also that person's department head and, if possible, the head of research or the CTO as well as specialists of the technology intelligence unit and the R&D planning department.
In the study, questionnaires were used as a basis for semi-structured interviews. Different questionnaires were developed for the interviews on the technology level and the corporate level. Furthermore, company-internal documents on technology intelligence and planning processes as well as documents on project proposals were studied. Moreover, company-external documents, such as publications and studies about the importance of the technologies and their diffusion, were studied. In addition, patent and publication frequency analyses were performed. Then, case studies were written for each company, and they were reviewed by key informants, who were usually the technology intelligence specialists. Starting from the understanding of technology intelligence as a continuous process, which has been described in the framework above, the writing of the case studies resembled a process of pattern-matching which created new insights[61, p.56]. Finally, the written cases were used for cross-case analyses. According to Yin, multiple case study research tries to generate and replicate findings with theoretical implications and does not try to test hypotheses in a‘sampling’logic.
Altogether, 156 interviews were conducted in 25 Northern American and European technology-intensive companies (Fig. 2). The large number of case studies[61, p. 58] was chosen in order to be able to identify major contingency factors of the technology intelligence process, industry characteristics being potentially one of them. Companies from the pharmaceutical, telecommunications equipment and automotive/ machinery industries were examined with the goal of exploring industry differences in the management of technology intelligence processes. The choice of industries was influenced by the typologies of Kodama
and Gerybadze and Reger. The technologies chosen for the study implied a radical technological change for the three industries. The technologies were sufficiently advanced to be able to evaluate the technology intelligence process but sufficiently new to still have access to the relevant informants. In the pharmaceutical industry, combinatorial chemistry was chosen, in the telecommunications equipment industry the dense wavelength division multiplexing (DWDM) technology was analyzed, and in the automobile industry, the fuel cell was studied. Furthermore, in each of the three industries some selected technologies were chosen to verify the results obtained with the other technologies. These were the bioinformatics and proteomics
technologies in the pharmaceutical industry, voice over IP technologies in the telecommunications equipment industry and the fuel direct injection technology in the automobile industry.
By applying a two level research approach using multiple informants, a complex view on the firms' technology intelligence processes and on the different mechanisms for organizing these processes could be gained, which facilitated detailed assessments of these processes and structures. The quality of the technology intelligence process was evaluated through an analysis of the bottlenecks within the process using several criteria. First, we studied how quickly some employees became aware of external signals, such as publications or activities of competitors. Second, we studied how quickly this information was communicated and integrated into the firm's decision-making processes. Third, we evaluated the quality of the assessments made. The main problem in doing so is that there is no such thing as a‘correct’assessment and decision. Due to the uncertainty of the future, ex-post analyses of the correctness of assessments are only to a limited degree a good criterion for the quality of an assessment. Therefore, we attempted to understand the assessment situation in the past, concerning the global state of R&D and the company-internal preconditions for the size and timing of the investments into the new technology. Also, patent and literature frequency analyses were performed to get an impression of the speed of diffusion of the technology. In addition, our perception of the technology intelligence process was reviewed by the key informant. In order to be able to get sufficient information on the past assessment situation, relatively recent technological change situations were analyzed. Fourth, we investigated, whether the company was able to actually react to an assessed trend and enforce decisions in time, particularly whether necessary resources could be provided. Fifth, the firms' capability to reevaluate decisions, which had been taken earlier, in the case of new trends, was used as an additional criterion for the quality of the technology intelligence process.
4. Research results
4.1. Three types of technology intelligence processes in the companies studied
Starting from an understanding of technology intelligence as a continuous reevaluation process and a research design based on this understanding, three types of technology intelligence processes in the
Fig. 2. The companies studied.
The binational companies Glaxo Wellcome and SmithKline Beecham were grouped as of American origin because the head of R&D and the planning activities are located in the US, where also the interviews took place.
context of radical technological change could be identified: the hierarchical, the participatory, and the hybrid technology intelligence processes. There were major differences between the three types of technology intelligence processes based on the criteria established above. There were hardly any differences concerning the speed with which companies became first aware of the new trend. These differences were negligible compared to those which resulted from the phase after the identification. There were enormous differences concerning the speed with which the identified trends were communicated and integrated into the decision-making processes. The quality of the assessments differed strongly among the three types, also because different groups of employees were involved. Moreover, the three types differed in management's ability to carry out decisions due to the varying availability of resources for radical innovations. Furthermore, the three types differed in their ability to regularly reevaluate decisions that had been taken earlier, which may be necessary due to changes in the environment and possible errors in earlier analyses. In the following, the three types of technology intelligence processes are described.
4.1.1. The hierarchical technology intelligence process
In the hierarchical technology intelligence process, individual researchers became aware of the new technological trend. Due to their prior related knowledge about the technology and the knowledge about an innovation need, they had the necessary absorptive capacity for identifying the trend. The slack resources in the research driven companies that applied the hierarchical technology intelligence process enabled them to test these technologies on the research group level. In this first phase, the research results of relevant publications were often duplicated. Researchers considered this to be necessary because they did not trust external research results and did not want to bother top management with unproved issues. They also feared losing their scientific reputation within the company. A researcher at Novartis emphasized that this scientific culture was deeply embedded within the organization.
When the technology had been tested successfully, the researcher communicated it to the technology intelligence specialists or directly to top management. This was often done solely by informing the head of the research area. This management level only took notice of the procedure without intervening actively. It was part of the innovation culture of these companies that individual researchers contacted top management directly in the case of radically new technologies. At Lucent Technologies, a researcher said that the head of research had once presented an organization chart with himself at the bottom and the researchers at the top, emphasizing that the researchers drove the organization. “He was accessible to everybody, and discussions with him were free of hierarchy.”
In many cases, the researchers had already formulated a project proposal, in which the technology was assessed in detail. If this was not the case, technology intelligence specialists helped the researchers to formulate the project proposal. In any case, the technology intelligence specialists tried to simultaneously assess the technology roughly or engaged external experts for an assessment. The research management board approved the project proposal or rejected it. The board generally trusted the analysis of the researcher but used the analysis of the intelligence experts as a second opinion.Fig. 3gives an overview on this hierarchical technology intelligence process.
In the companies studied, the hierarchical technology intelligence process led to overvaluations and undervaluations of technologies2because neither the researcher nor the technology intelligence specialist
2 There is no such thing as a ‘correct’ assessment, but by comparing the reactions of different companies' systematic
had the competence to assess the technology holistically. The researcher usually only had insight into the scientific importance of the technology, and he or she had to write the project proposal in an optimistic way to get it accepted. In these science-driven companies, the project proposal was also an opportunity for a researcher to receive management attention. Getting one's name linked to a technology or idea was seen as a major motivation for researchers to search for new trends in companies with slack resources.
The specialists in the technology intelligence unit usually had only the knowledge to broadly assess the general strategic importance of a technology. They did not have the knowledge to assess the technology in its application context, e.g. the technological progress required in related application areas. These specialists also lacked the knowledge to be able to challenge the reports of the researchers in detail and were therefore forced to trust the researcher. The support of external experts, which was often sought by the companies in this situation, offered an objective evaluation only in a limited way. External experts could assess the technologies only from a scientific point of view and tended at least in these cases to overly optimistic forecasts. Furthermore, they were not able to determine the strategic relevance of the particular technologies in a company-specific way. Even the heads of research in many companies were not able to evaluate the new technologies in detail. Many of the heads of research stated frankly that they had not been able to challenge the project proposals and the underlying trends in the resource allocation processes and that therefore they had mainly relied on the recommendations of the internal and external experts (Fig. 4). As a consequence, Novartis, for example, invested too early and too much in combinatorial chemistry even though upstream and downstream technologies in the innovation process were not yet available. When the high throughput screening technology was finally available, however, there were not enough resources to develop large scale combinatorial synthesis plants.
The hierarchical technology intelligence process did not only lead to false assessments of technologies but also to a dissatisfaction within the company. In the case of new technologies, existing resources had to be reallocated. If the reallocation was based on a false assessment, middle management did not fully support the implementation of the decision because it did not accept the assessment and felt that it did not get heard sufficiently.
In the process phase after the decision of top management, it was of great importance that systematic monitoring took place. If the decision had been taken to invest in a technology, the monitoring was done by the research group that was created. If top management had rejected the project proposal, the researchers continued to monitor this area. They communicated incremental trends to middle management
which allowed integration of these trends into the decision-making processes. Researchers only addressed top management again if there was a radical new trend in the field. The fundamental decision to spend or not to spend a certain budget on a technology was often not reconsidered regularly by top management because technology planning was too centralized and therefore not sufficiently detailed. However, radical trends, which required an increase or a reduction of the general budget for a technology, were quickly identified by researchers, as in the case of the first identification of the technology. Comparable communication and assessment routines led again to overvaluations and undervaluations of the trend. At Merck, the importance of combinatorial chemistry had been overestimated by the head of research and the researcher at the beginning. Many members of middle management perceived this as a misjudgment and reacted in many cases with opposition, thus hindering strategy implementation. When four years later the size of the budget of combinatorial chemistry was finally adjusted, its actual importance was underestimated and the activities were sharply downsized.
The strength of the hierarchical technology intelligence process was that it led to generally accurate assessments of the direction of technological change and quick decisions to begin research in the correct areas. Due to clearly defined communication routines and the existence of a budget for radical innovations or a comparatively easy reallocation of resources combined with a scientific discourse, few barriers existed in the process. The weaknesses were overvaluations and undervaluations of technological trends
leading to suboptimal timing and size of technological investments. The lack of involvement of middle management in the assessments lead to neglecting the organizational context, such as the availability of complementary technologies and the support in strategy implementation. Even if the assessments of technology intelligence specialists were correct, they only supported the individual learning of the head of research and not organizational learning because they did not include other R&D employees. Decisions that had been taken earlier were only reevaluated in the case of fundamentally new trends or massive internal opposition. As a consequence, overvaluations or undervaluations were often not corrected. 4.1.2. The participatory technology intelligence process
In the participatory technology intelligence process, the technology was identified by R&D employees, and its relevance was tested in an exploratory research project based on slack resources. This test was regarded as important in the companies that applied the participatory technology intelligence process in order not to lose reputation. To avoid inter-departmental politics at a later point of time, the researcher discussed the new trend in detail with the head of his department. After an in-depth test, the technology was communicated in a first step only to large parts of middle management and intensively discussed by different interest groups (Fig. 5). Top management was only informed after intense discussions had already taken place.
In the companies with participatory technology intelligence processes, there were neither communication routines to top management nor planning and resource allocation processes for radical innovations. At Nortel Networks, individual researchers knew about the DWDM technology and its increasing relevance due to new competitors, such as CIENA. They contacted middle management, but there was no research budget for expensive developments. Middle management was rather reserved and not willing to try to influence top management. It rather discussed with other members of middle management. As these colleagues feared a loss of resources for their own competing projects, they opposed to the new trend. As a consequence, individual researchers tried to get access to top management but were told by management's assistant to discuss the new trend in the following year's planning session.
Top management often heard about the radical new trend from middle management rather late. Furthermore, it often received contradictory signals which were distorted due to internal politics. Top management postponed necessary decisions due to expected internal opposition to a reallocation of resources or due to a lack of detailed understanding. In a pharmaceutical company, a researcher said that even the head of research had to create consensus within the research management board in order to get new projects passed. The members of the board were the heads of the different research areas, who were
often directly concerned with a decision because often resources had to be reallocated due to a lack of budget for radical innovations. Each member of middle management who had heard about combinatorial chemistry had tried to influence‘his’member of the research management board in the mean time.
As a consequence, supporters and opponents of the technology ended up in intense conflicts. Finally, a project proposal was written by middle management, or the head of research took up the subject in a highly politicized situation. The head of research used the technology intelligence unit as a neutral consultant and sought the advice of external experts. The output of these two assessments were often documented in a report and broadly disseminated to supporters and opponents of the innovation within the company in order to have an objective foundation for decision-making (Fig. 6).
In this highly political situation, the assessments of the technology intelligence specialists and the external experts were often rejected independent of their quality. Top management took these assessments only partly into consideration in their decision-making. With its decision, top management rather tried to mediate between the supporters and opponents of the trend. If the analysis of the technology intelligence specialists was taken into consideration, the group that was not supported by their assessment regarded the specialists as obstructionists. Consequently, the technology intelligence specialists were no longer supported with information informally gathered by that group.
After top management had taken a decision, a systematic monitoring was of great importance. If the decision had been taken to invest in the technology, the monitoring was done by the research group that
had been created. If top management had rejected the project proposal, the researchers continued to monitor this area. Incremental trends were communicated via middle management and successfully integrated into the decision-making processes. The fundamental decision to spend or not to spend a certain budget on a technology was not reconsidered regularly in most cases because there was no receptivity for radical new trends, which required a reconsideration of earlier resource allocation de-cisions. Missing communication routines and the lack of regular planning and resource allocation processes for radical innovations again prevented a timely reaction as was the case of the initial identification of the new technology.
The strength of the participatory technology intelligence process was the intensive, autonomous discussion among different interest groups within the company. However, these discussions took too long and ended up in intense political conflicts. Middle management or process promotors did not organize rigorous assessments. Due to the lack of communication routines to top management, researchers communicated a trend to middle management, which often filtered it out. Top management often heard about the new trend rather late. In addition, the missing resources for radical innovations further slowed down decision-making and enforcement of decisions. Due to missing planning and resource allocation processes for radical innovations, a regular reevaluation of the taken investments was not made. Rather, intense conflicts emerged again.
4.1.3. The hybrid technology intelligence process
In the hybrid technology intelligence process, the technology was identified by individual re-searchers and was tested in an exploratory research project based on slack resources. With the support of the technology intelligence specialists, the technology was communicated directly to top manage-ment. Only the concerned head of the research area was informed. (Fig. 7). The technology intelligence unit helped to formulate the trend and to communicate it to top management. Top management made a very broad test of relevance and initiated an assessment which was organized by the technology intelligence specialists. In this assessment, all relevant interest groups, including the individual researcher, were involved.
The assessment processes that were focussed on radical innovations included the systematic ex-ploration of future technological needs. Middle management was included in these assessments because it was regarded as being able to assess not only the general strategic importance but also the scientific
relevance of a technology. One top manager stated:“Only middle management can put bench and strategy together.”In doing so, middle management was able to assess the technologies in their context, including the availability of complementary technologies. SmithKline Beecham, for example, had assessment platforms for new technologies, which assessed not only individual projects but compared them also to other technological alternatives as part of a continuous strategic technology planning. As a result, the choice and timing of combinatorial chemistry technologies was comparatively well chosen regarding the internal availability of upstream and downstream technologies in the innovation pipeline, such as high throughput screening.
Technology intelligence specialists served in the hybrid technology intelligence process as a contact partner for the communication of new technological trends and moderated assessment processes by delivering the necessary process and methodological know-how. At DaimlerChrysler, the fuel cell was assessed by using the scenario technique in a very participatory process including not only fuel cell specialists but also the middle managers responsible for the combustion engineering lab. Technology intelligence specialists made control studies and asked advice from external experts. The studies were used as a neutral second information source for top management.
A prerequisite for an efficient hybrid technology intelligence process was to assure that the name of a researcher stayed attached to the new technology. Otherwise, the motivation was lost for the informal bottom up technology intelligence process. For the motivation of middle management to regularly participate in the assessments, it was of major importance that the analyses of middle management were taken into consideration in decision-making. The implementation of the assessments and of the decisions of top management was facilitated through the existence of resource allocation processes which were focussed on radical innovations. In the hybrid technology intelligence approach, the participatory assessments were used as learning and motivation building processes in order to overcome potential opposition. Thus, the assessment became a self-fulfilling prophecy. Fig. 8 gives an overview on the characteristics of the hybrid technology intelligence process.
After top management had made the decision to invest in the technology or not, the technology was regularly reevaluated in the radical planning and resource allocation processes. Similar to the initial identification of the technology, radical new trends were identified by individual employees, communicated to top management via the communication routines and assessed by a team including important members of middle management. Technology intelligence specialists mainly assured that the technology was regularly covered within the planning meetings. A broad monitoring for strategic technology planning was done by specific monitoring teams. At DaimlerChrysler, monitoring teams were created as a part of scenario planning efforts including again a diverse set of employees. The detailed monitoring was carried out by the newly created R&D center“Projekthaus Brennstoffzelle”.
The strength of the hybrid technology intelligence process was the fast integration of the trend into decision-making processes due to clearly defined communication routines to top management, skipping middle management. Another strength was the assessment of trends by the most suited persons leading to a high effectiveness of technological decisions, especially concerning timing and size of investments. The existence of dedicated resource allocation processes for radical innovations led to a relatively easy enforcement of decisions. Furthermore, implementation was improved through the involvement of key personnel in decision-making. As planning and resource allocation processes existed for radical innovations, once taken decisions were regularly reevaluated. Not a real weakness but a difference compared to the hierarchical technology intelligence process was the slower decision-making process due to the integration of key personnel in the assessments.
4.2. The distribution of the three types of technology intelligence processes in the three industries The types of technology intelligence processes were observed with different frequency in the three industries (Fig. 9). In the pharmaceutical industry, hybrid and hierarchical technology intelligence processes were particularly common. In the telecommunications equipment industry, most companies had hierarchical technology intelligence approaches. In the automobile industry, most companies had participatory technology intelligence approaches.
Four pharmaceutical companies had a hierarchical technology intelligence process and were highly science-driven. Pharmaceutical companies have traditionally been science-driven as scientific breakthroughs often lead to an immediate competitive advantage. The six companies with hybrid processes also had hierarchical processes in the past but had changed their approaches due to earlier investments which had not taken sufficiently into account complementary technologies and technological alternatives leading to discontent among middle management. There was only one company with a participatory technology intelligence process. It had spun-off corporate research, cut back the R&D budget and increased controlling efforts.
Five telecommunications equipment companies had a hierarchical technology intelligence process, which led to correct assessments of the technologies, but with a problem in timing the investments. In this industry, the timing of technology investments is of particular importance as there are often many competing technologies
and their diffusion is strongly influenced by standardization processes. Two companies had hybrid technology intelligence processes, which were closely linked to decentralized radical technology planning and resource allocations processes, such as roadmapping and scenario planning. Only one company had a participatory technology intelligence process.
In the automobile companies, high internal barriers for the communication of radical new trends to top management existed despite the fact that many automobile companies had large central research departments and were among the companies with most patent applications in their country. In many companies, corporate research carried out applied research focussing on improving individual components. The top management of many companies learned from journal articles about the fuel cell. There were no clearly established com-munication routines. Four automobile companies had participatory technology intelligence processes, whereas only one company had a hierarchical process and another one followed the hybrid technology intelligence approach.
As it has been shown above, there were considerable differences between the industries concerning the distribution of the three types. One possible explanation is that companies in the more dynamic and science-based telecommunications equipment and pharmaceuticals industries are exposed to radical technological change more often. Accordingly, they might have built a different technology intelligence capability. However, each type of technology intelligence process could be observed at least once in each industry. As a con-sequence, these industry differences cannot be explained only by differences in the rate of radical technological change. In addition, the study on the company level and the technology level allowed identification of two further contingency factors: the company culture and the decision-making style of the company.
The company culture influenced particularly the motivation of individuals to openly and quickly communicate information to top management and how top management appreciated individual efforts. Depending on the existence of such communication routines, a slow or quick integration of trends into the making processes occurred and with it timely or tardy reactions to new trends. The style of decision-making, in contrast, particularly influenced the quality of the assessments and the ability of companies to react to given trends. On the one hand, the decentralization of decision-making and, above all, the involvement of middle management influenced the quality of the technological assessments. On the other hand, the formalization of decision-making, particularly the existence of planning and resource allocation processes for radical innovations, directly influenced the speed of decision-making as it removed barriers, such as necessary resource reallocations. Furthermore, the formalization of planning and resource allocation processes influenced the frequency of reevaluating earlier decisions and thus the speed of decision-making.
As the technology intelligence process was studied longitudinally both on the technology level and on the company level, it could be observed that all companies increasingly formalized and decentralized their technology planning and resource allocation processes for radical innovations during the most recent years. As a consequence, many of these companies have advanced slowly towards a hybrid approach. In the pharmaceutical industry, both Merck and Novartis have recently changed their technology intelligence processes completely to a hybrid approach. In the telecommunications equipment industry, Nortel Networks has improved its resource allocation and technology planning processes, thus creating the preconditions for hybrid technology intelligence processes. Also in the automobile industry, Volkswagen has recently systematized its approaches to monitor and assess trends in power trains and in electronics. As a result, the quality and speed of radical technological decision-making in future situations of technological change is expected to improve in most companies studied.
In this study, differences between companies from three industries in their ability to master radical technological change have been identified. Three types of technology intelligence processes have been distinguished, which are not evenly distributed among the three industries. In addition to industry effects, company culture and the style of decision-making could be identified as the main contingency factors of the three types of technology intelligence processes. In the following, the antecedents of these company-specific and industry-specific differences are discussed in detail, and implications of the results for the research fields of technological change, technology intelligence, promotors and organizational theory as well as for practice are drawn.
Concerningresearch on technological changeit could be shown that a holistic and detailed approach to the study of technological change in companies is necessary. The treatment of companies as one entity is often undertaken in literature[12,21]and explanations of the failure of companies by very abstract terms, such as managerial incompetence and insufficient technology and market information, do not sufficiently take into account the complexity of organizational information gathering and decision-making. The holistic and process-based approach of this study has opened the black box of organizational behavior in the face of radical technological change. It shows that companies can fail in different phases of the process. However, it also shows that there are means to reduce the probability of failure in situations of radical technological change. In all companies, new trends concerning the technologies that have been studied were first identified by individual R&D employees as part of their work, long before dedicated technology intelligence units learned about it. The monitoring of technological trends should therefore be as participatory as possible. A strong bottom-up technology intelligence process requires an innovation culture, the existence of slack resources and recognition mechanisms for employees who have identified interesting new trends or innovations. This result confirms prior research[13,30,68–70].
Concerning the point in time at which the technological trend was first discovered by individual employees, there were comparably small differences between the companies. Differences in the reaction to the trends were rather found in the subsequent process steps. The existence of communication routines to top management proved to be particularly important for a timely integration of new trends in the decision-making processes. Technology intelligence specialists had a key role in translating the trend into the language of top management, in communicating it to top management and in transferring it into participatory assessment processes, such as roadmapping. These communication routines helped avoid the filtering out of radical trends and the emergence of constant conflicts between supporters and
opponents of new technological trends. The importance of such communication routines for the timely integration of new trends into decision-making has not been described in the literature before. It seems to be one of the root causes of the abstract explanations used in literature[11,12], such as organizational inertia or ignorance of top management.
The involvement of middle management in the assessments was of particular importance for the quality of the assessments. Potential overvaluations and undervaluations were reduced due to the ability of middle management to assess in detail not only the general relevance of the technology but also its particular relevance for the company. If trends are only assessed by top management and the individual researchers, errors concerning timing and the amount of resources that are spent for a new technology are frequent. In contrast to past research[11,12]top management did not seem to be‘incompetent’, it just did not have the detailed knowledge of the specific technology and the necessary complementary technologies. Due to time restrictions it should not and cannot be the task of top management to build up this knowledge. Top management has to carefully select people who perform this task. The involvement of middle management also increased the acceptance of the final decisions made by top management and therefore facilitated their implementation. Furthermore, this approach shows that organizational inertia
is not a given fact but may be actively managed.
For the timely reaction of companies to trends, it was of major importance that the assessments and decisions that were taken were linked to the allocation of resources. Successful companies either had a fixed annual budget for radical innovation projects or were able to reallocate resources flexibly due to the power of individual members of top management. In the case of a reallocation of resources, however, only single projects were assessed without comparing them to other investment alternatives. Moreover, organizational resistance was much higher in these companies than in the companies with fixed budgets for radical innovation projects. This lack of availability of necessary resources differed from the explanation‘lack of financial and personnel resources’ of the overall company in literature [10,11]. The companies would have had the necessary resources, they only lacked an appropriate resource allocation process. The lack of such a process might lead to the impression of high organizational inertia. In such situations, however, old and new projects have to be actively championed, which leads to organizational resistance.
Regular reevaluations of earlier assessments and decisions are necessary in order to react to changes in the environment or to correct errors in decisions. This study has shown that such errors may represent overvaluations and undervaluations of a technology's relevance and that both can lead to considerably wrong allocations of resources with equally detrimental effects. This result is in contrast to the existing literature, which mainly discusses undervaluations of technologies in the form of a tardy investment in a technology (e.g.
[9,12,21]). As also overvaluations may occur, research should address the more general question of an optimal size and timing of a technological investment. As the future in general and forecasts in particular are uncertain and as trends are created in a continuous process by forces that are internal and external to the company
[11,71–74], it is difficult to determine the optimal size and timing in advance. For an individual company, the optimal size and timing of an investment in a technology is determined by a continuous reinterpretation of changes in both the environment and company-specific factors which is why the optimal size and timing change over time. Only a regular evaluation of prior decisions and, if necessary, a reallocation of resources can assure that a company will adequately adapt to a changed environment and that previous overvaluations or undervaluations will be corrected. Such adaptation processes are necessary along the whole technology life cycle and not only when a company first invests into a technology, as often emphasized in literature[75,76]. In this study, differences between the industries in the ability of the companies to manage radical technological change have been found. Companies in the mature automobile industry did not manage this
process as well as the companies in the more science-driven pharmaceutical and telecommunications equipment industries. A possible explanation is the different rate of technological change[66,77], which forces companies to accustom themselves to different degrees of radical technological change. The ability to master technological change has to be learned by companies in a slow process because it requires changes in decision-making processes and company culture. However, the fact that in each industry companies with a hybrid technology intelligence approach could be found and that in general companies in all industries were able to adequately cope with the new technologies shows that any company can develop the capability to manage technological change. Moreover, the study, which covered not only the technology level but also the corporate level, has shown that many companies had recently advanced towards a hybrid approach. Basically, the industry differences on the technology level reflect the industry differences which were found by Lichtenthaler
in his analysis of technology intelligence approaches on the corporate level.
The industry differences in the organizational capabilities to manage radical technological change can partly explain the differences between several studies on technological change. Tushman and Anderson
and Utterback studied mainly mature industries with a low rate of radical change. Zucker and Darbyand Thomke and Kuemmerle, in contrast, analyzed the pharmaceutical industry, in which radical change occurs rather continuously. Similarly, Christensen  studied the electronic industry, which also has a high rate of technological change. As the number of companies in each industry was rather limited in this study, the explanation of industry differences given here can only be regarded as a first hypothesis and deserves further research.
Concerningresearch on technology intelligence, this study has demonstrated for situations of radical technological change how different levels of hierarchy and how structural, hybrid (project based) and informal forms of coordination interact in the technology intelligence process. In the companies studied, dedicated technology intelligence specialists became aware of new trends long after the employees conducting research in the respective fields. Therefore, the monitoring of technological trends should only be carried out by dedicated units in the case of exotic technologies which are not covered by the search routines of the employees or if a second opinion is required. Formal technology intelligence should mainly have a coordinating role which supports prior research [35,46,48]. Accordingly, technology intelligence should not be understood as a unit but as a distributed process of organizational intelligence
[46,78–80]. Furthermore, the study has shown that the technology intelligence process cannot be studied in isolation from its organizational context. Gathering information and communicating it to middle or top management is not sufficient but there is also a need of adequate resource allocation processes so that a timely reaction to a new trend is possible. The industry differences that have been found closely reflect the differences identified in prior research[46,48]on corporate technology intelligence processes.
Concerningresearch on promotors, it has been shown that in the companies with a hierarchical or hybrid technology intelligence process the communication routines, which can be regarded as a combination of power promotors, technology promotors and process promotors, represent consciously managed information channels. The researchers have the role of the technology promotor and the members of top management have the role of a power promotor. Technology intelligence specialists have the formal role of a process promotor. The role of the process promotor is considered an informal role in literature, whereas it was part of the formal organization in the companies studied. In these companies, the receptivity to radical trends was high due to well defined communication routines. Moreover, in companies with a hybrid approach to technology intelligence, internal barriers and false assessments were reduced due to participatory planning and resource allocation processes focussed on radical innovations. Technology intelligence specialists had the tasks of bringing together competent employees for the assessments and of moderating this process. The
test of relevance for the company was not done by the process promotors as shown in the literature but by carefully selected employees, mainly from middle management.
In the companies with a hierarchical and a hybrid technology intelligence process, the barriers for the reaction to radical trends seemed to be reduced due to the existing routines for the communication, assessment, decision-making and resource allocation. These formal routines strongly diminished the importance of the traditional informal promotor model. This is in contrast to the existing literature
, which supposes that there are nearly always strong barriers, which have to be overcome, and that projects will nearly always profit from promotors. In the companies with a participatory technology intelligence process, however, there were high barriers due to a lack of communication routines and a lack of resource allocation processes for radical innovations. Nevertheless, there was not an evident power promoter in top management. After intense conflicts between middle management, support from top management rather emerged as a result of initiatives by supporters and opponents to the trend and from general changes in the environment. Therefore, this informal promotor model proved to be less effective than the formalized promotor roles. In the companies with fewer barriers for new technologies, the promotor roles were more pronounced, also because these roles were more formalized. These results are counterintuitive to existing research which claim that promotors are especially frequent if barriers are high. The effectiveness of promotor roles is therefore strongly influenced by the organizational context, such as the existence of resource allocation processes for radical innovations.
Furthermore, as the research stream on promotors normally analyzes only single projects, it has so far not become aware of the disadvantages of promotors, such as the overvaluation or undervaluation of trends, even though the role of opponents of innovations is no longer considered to be solely negative
[81,54]. Future research on promotors should take into consideration the organizational context and should therefore have a multi-project view [82,52] in order to analyze the importance of individual promotors for several projects.
From the point of view of research on organizational theory, the hybrid approach to technology intelligence is a combination of the N-form and M-form of organization of companies as described by Hedlund. The M-form of organization builds on hierarchy and is led by top management. The N-form of organization uses heterarchies. The N-form closely resembles the middle-up-down organization model of Nonaka and Takeuchi. The communication of radical trends to top management via middle management in the participatory type did not prove to be successful in the European and Northern American companies that were analyzed in this study. Also the hierarchical model using a technology promotor and a power promotor did not prove to be successful as it generated systematically false evaluations. The combination of communication routines for radical trends skipping middle management and assessment routines including middle management on the one hand and the communication of incremental trends via middle management on the other provided the companies studied with the ability to manage simultaneously radical and incremental technological change. This approach leads to an ambidextrous organization as claimed necessary by Tushman and O'Reilly.
In this study, the organization of the technology intelligence process has been analyzed in the context of radical technological change. Three approaches to organizing the technology intelligence process could be identified. Starting from a process-based analysis of the information flow, the strengths and
weaknesses of the three approaches could be shown. It has been demonstrated that the analysis of companies as a single entity and the abstract explanation of its demise in situations of technological change by managerial incompetence and organizational inertia do not sufficiently take into consideration the complexity of organizational information acquisition and decision-making. In each of the process phases, different bottlenecks can lead to the failure of companies. Besides the competence of individual managers, the quality of technological decisions depends on a broad and participatory monitoring of technological change, the existence of communication routines to top management, on the assessment processes, and on the resource allocation processes for radical trends and innovations. Moreover, the ability to manage fundamental technological change does not only depend on single decisions at specific moments but also on the ability to regularly reevaluate decisions that have been taken earlier. Overall, the analysis has shown that the ability of companies to monitor and decide on technological trends can be managed more thoroughly than the abstract explanations that have been used in the literature suggest.
Besides the company-specific differences, industry-specific differences have been identified. Companies in the dynamic and more science-driven pharmaceutical and telecommunications equipment industries mastered technological change better than companies in the mature automobile industry. One explanation is that the ability of companies to manage radical change is learned in a slow process. The automobile industry with a lower rate of radical change provides less learning opportunities.
Future research should follow the holistic view on the technology intelligence process that has been adopted in this study and analyze other situations of technological change, such as incremental change, in order to deepen our understanding of the interaction of the different hierarchical levels and of the different forms of coordination in the technology intelligence process. Furthermore, the industry differences identified in this study deserve further research with the same or with different technologies in these industries. As this was a study of large multinationals with specific barriers to information acquisition and decision-making, it seems necessary to perform comparable studies in small and medium sized companies. Appendix A
The technologies that were analyzed, i.e. combinatorial chemistry, DWDM and the fuel cell, implied a radical technological change for the companies studied. Partly, these technologies required the development of new competencies and had the potential to destroy existing competencies. All technologies were comparatively old even though they were described in the press as‘new’to the industry at the time that the study was carried out. The major challenge for the companies studied was the identification of the change of relevance of the technology because technological breakthroughs and changes in customer needs made the technologies become relevant for application. The change of relevance of the technologies followed the typical pattern of their particular industries. The diffusion of the fuel cell does not depend only on its cost but also on industry standardization processes and legislation. The diffusion of DWDM, in contrast, was made possible through standardization processes and shifts in the needs of key clients. After a long phase of exploratory research, DWDM became competitively relevant in a short time. The diffusion of combinatorial chemistry happened very quickly after the technological breakthroughs, which made it broadly applicable to pharmaceutical research. Progress in science was intensely monitored in the science-driven pharmaceutical industry and was applied with only little delay as it was of immediate competitive relevance.
Besides the identification of the change of relevance of the technologies, the companies had to determine the appropriate timing and size of an investment in the technologies through intense monitoring and detailed analyses. The uncertainty about the relevance of the technologies for their own company was
rather high. Due to new trends, initial assessments of the relevance of the technologies had to be revised several times leading to a reallocation of resources. What in an aggregated analysis might appear to be a discrete point of investment in a technology was rather a continuous reevaluation process. Therefore this study did not focus on the identification of a discrete point of investment in a technology but on the analysis of the continuous monitoring, assessment, decision-making. In the following, the development paths of combinatorial chemistry, DWDM and the fuel cell in their respective industries are described. A.1. The development of combinatorial chemistry in the pharmaceutical industry
Since the early 1980s, several fundamentally new technologies have been developed for the identification of potential lead substances. Pharmaceutical companies have traditionally tested a large number of substances on individual targets via random screening in a rather stochastic and time consuming approach. Therefore, pharmaceutical companies have developed large substance libraries through costly synthesis and search in nature.
In the mid 1980s, the screening process could be speeded up enormously by using technologies, such as micro systems technology and robotics. With this highly automated random screening, which is now called high throughput screening, 500,000 substances could be tested per week. Pharmaceutical companies therefore perceived the size and diversity of their libraries as too small. However, it had taken the pharmaceutical companies decades to build libraries of a size of about 1,000,000 substances. In this situation, combinatorial chemistry emerged at the end of the 1980s (Fig. 10)3. Combinatorial chemistry makes it possible to generate up to a million structurally diverse molecules in a short time. Together with the high throughput screening systems, all substances suited for a target could be tested within a short time.
Many traditional pharmaceutical companies perceived combinatorial chemistry as a threat. The competitive advantage of the substance libraries, which had been built up in decades, seemed to fade away. At the same time, combinatorial chemistry made many pharmaceutical companies hope that the innovation process could be highly planned and automated in the future.
Even though combinatorial chemistry was perceived as a suddenly emerging and discontinuous technology, it had already been mentioned in a theoretical paper in 1963. Important research concerning the realization of combinatorial chemistry was done in the early eighties. A quantum leap forward was the use of substance mixtures in the screening process; the so-called split and mix method was published in 1990 and 1991 in Nature and Science. In 1992, a publication in Science showed completely new application fields for combinatorial chemistry in pharmaceutical research. Combinatorial chemistry had so far been regarded by the pharmaceutical companies as a not broadly applicable technology due to its limitation to complex molecules, which are difficult to produce and cannot be administered orally. The applicability also to small molecules seemed to make it all of a sudden a key technology for pharmaceutical research.
Research on combinatorial chemistry was originally driven by science. Not before the mid 1980s, some former academic researchers founded start-up companies, such as Affymax, with the objective to make combinatorial chemistry usable for pharmaceutical research. Even though combinatorial chemistry was seen at that time not applicable to pharmaceutical research in general but rather limited to immunological applications, most pharmaceutical companies started to work with these new approaches with only a small
Due to the science-driven nature of the pharmaceutical industry, research activities were measured in publications based on the science citation index in contrast to patents used in the telecommunications equipment and automobile industries.
time lag. Ciba-Geigy started research on combinatorial chemistry already in 1987. The publications in 1990 and 1991 on the split and mix method and the publication in 1992 on the applicability of combinatorial chemistry to small molecules forced most pharmaceutical companies to reconsider and to increase the size of their existing activities in the field of combinatorial chemistry.
While building up competencies in combinatorial chemistry, most companies realized that it would take much longer than originally hoped to make combinatorial chemistry applicable at large scale. It also became apparent that combinatorial chemistry was particularly useful if there were already certain indications of a lead substance. This finding strongly diminished the euphoria and led to a different view on the potential of combinatorial chemistry. At the end of the 1990s, many pharmaceutical companies therefore reduced their combinatorial chemistry activities. In most pharmaceutical companies combinatorial chemistry is not yet used in more than 10% of all R&D projects.
A.2. The development of DWDM in the telecommunications equipment industry
Optical fibers have been used in telecommunications broadly since the early 1980s. The standard for data transmission on optical fiber is SDH (synchronous digital hierarchy). It is based on data transmission via TDM (time division multiplexing). TDM divides transmission time of a fiber in short intervals, which can be allocated alternately to different users. The shorter the time intervals, the higher is the capacity of the fiber.
In order to increase the band width of networks, there were two possibilities: laying new fiber or increasing the TDM bit rate. As installing new fiber is very expensive, the usual solution was a new TDM multiplicator. Due to technical reasons, this was always a multiplication of factor four. The highest TDM bit rate usable in practice is STM-64 (SDH), which is 10 Gbit/s. The next step would have been 40 Gbit/s. As there are already non-linear optical effects at STM-64, which have to be filtered out costly, 40 Gbit/s were not considered feasible at acceptable costs by many experts.
A new solution for increasing the transmission rate on a fiber is the DWDM technology, whose origins, however, can be traced back to the 1950s. Dense wavelength division multiplexing (DWDM) increases the transmission capacity by transmitting several optical signals in parallel on different wavelengths. In an