In a world of fierce competition it is not easy for companies to find the right skills and expertise in order to differentiate themselves and focus on their core competencies. The fierce competition has made companies to create cooperation's and alliances in order to stay competitive (Langfield-Smith and Smith 2003). However, starting cooperation's and alliances also involves risks. The risk could be manifested in not having sufficient information sharing or the lack of trust between the companies. When looking at evidence about these collaborations, there is a high failure rate in such arrangements (Langfield-Smith and Smith). On the other hand, if two partner companies manage to find themselves in a situation where they trust each other the risk failure can be reduced.
Control can be used to make sure that the outcome is as it was required and that the people involved are doing what they are supposed to. Control can be conceptualized as an evaluation process, which is based on the monitoring and evaluation of behavior or of output (Ouchi 1977). Control can be maintained through control mechanisms. Control mechanisms that can be divided into three different categorize; behavior-, i.e. input, or output control (Ouchi) and trust based control (Langfield and Smith). This three control mechanisms shows which different variables companies put more effort to in order to control someone else. These variables can consist of time, cost, information sharing or actions of people involved in collaboration.
AstraZeneca is a pharmaceutical company that has its bases in Sweden, Denmark and England. In this branch, product and medicines is developed all the time and companies need to find ways to differentiate themselves from the others. For AstraZeneca, discovering, opening and exploring new doors that can lead to new means of conduct is important, especially for its long-term survival (Caesarius 2008). AstraZeneca find this opportunity with a project called LinkMedica.
LinkMedica was a child of its time and the purpose with this project was to establish knowledge exchange relationship between AstraZeneca and the end user (Caesarius 2008). Now that AstraZeneca had an idea on what to do in order to create a new product that would make the like for asthmatics easier. Razorfish became the company that would help AstraZeneca develop the IT-platform for LinkMedica. The reason that Razorfish was chosen as the developer of the system was because of their expertise in IT-system solutions and the fact that they had helped other big companies with their IT-systems before.
In this project there where many people involved in it with different ideas, believes and assumptions but it was the relationship with Razorfish that would play a vital role in the development of LinkMedica. Razorfish was the only partner involved in the project that AstraZeneca outsourced something to and by doing this AstraZeneca's control over the collaboration could be an important issue to evaluate. This is because the project did not provide the results that AstraZeneca had hoped for. One way to evaluate why LinkMedica was terminated shortly after its launch is to look at the control that AstraZeneca had on the product development. This paper is therefore going to focus on how AstraZeneca used control mechanism on Razorfish during the pre-production and production stage of the development of LinkMedica. This leads out to our purpose of this paper; How did control mechanism changed from pre-production to production stage during the development of LinkMedica between AstraZeneca and Razorfish?
Throughout the paper, we will draw empirical evidences from the Caesarius (2008) and case articles to connect it with the theoretical framework. We are going to use Ouchi (1977), Ouchi and Maguire (1975) and Langfield and Smith (2003) to build a theoretical framework for analyzing and discussion.
1.2 Limitation
Control mechanism varies according to the structure of the firm. There is control mechanism for within firm control and inter-firm control. We will mainly focus on inter-firm control as the Empirical part of our paper consists of two independent firms: AstraZeneca and Razorfish.
2. Theory
In the theory part we will present a number of theories presented by different authors including Caesarius (2011) in class lecture.
2.1 Control System
Control system in organization is an important topic in the study of management. Control system is necessary to control the performance and thus we can say that it's a "mechanism" for controlling the performance. Control systems have been conceptualized and categorized in various ways: formal versus informal controls, behavior versus outcome controls; mechanistic versus organic controls; bureaucratic versus clan controls (Langfield and Smith, 2003). Control system can be categorized as formal and informal. Formal control consists of behavioral control and output control (Caesarius, 2011). Whereas informal control consists of trust-based control, social control or clan based control.
Behavioral control is conducted either directly or indirectly. Direct behavioral control is commonly used for new employees and could for instance be training or applied on complex tasks. Indirect behavioral control comes later in the process through employees who looks after each other in order to make sure that none of them is lagging behind (Caesarius, 2011) The other part of formal control is output control that consists of financial and non-financial control. Example of financial control can be annual reports, goals or targets and measurements whereas non-financial control can be values which are difficult to measure by numbers and that are important to the company (Caesarius, 2011).
Behavior control is most appropriate to use when the activities and the process is well known for the manager and the subordinates. Behavior control is usually used by small organizations and when there is a lack of goal coherence within an organization. Output control on the other hand is used when there is non-personal surveillance. Large and complex corporations use this form of control when the controlling party understands mean-ends and when there is known goal coherence (Caesarius, 2011).
According to Ouchi (1977) control system consists of two parts. These are;
A set of conditions that govern the form of control to be used. And
The control system itself.
Conditions are prerequisite for the first part of the Ouchi's control system. These conditions might provide a boundary or limit to the work or project. Management can takes initiatives if performance falls outside the acceptable boundaries. Whereas the second part of the Ouchi's control system consists primarily of a process for monitoring and evaluating performance which means control is also possible through monitoring and evaluation of performance (i.e. outcome of a project). The second one is comparatively more complicated to apply in an organization than the first one.
According to Ouchi and Maguire (1975), studies have distinguished behavior control mainly based on direct personal surveillance and output control mainly based on the measurement of outputs. Thus, we can say control-systems are divided into either monitoring outcomes, or monitoring behaviors. On the other hand, combination of these two is also possible. For example, sales teams are often controlled through a mix of behavior and outcome based control. In many organizations, sales people get a basic salary plus bonuses based on the total sales they make. Thus, which control system to apply depends on the type of organization and type of work. Cost benefit analysis (CBA) can be a good way to take decision regarding which system to choose. Say for example, if the relative cost of measuring outcome is less than measuring behavior and the benefits from output control is also higher, only then an organization might go for output control rather than behavior control. It is to note that, behavior control can be done either directly or indirectly. This control system is appropriate as a mechanism for within-firm control system.
There is advantage and disadvantage associated with both behavior control and output control. For example, Turcotte (1974: 72) suggested that the use of output control reduced the need for direct planning and management for a wide variety of contingencies (Ouchi and Maguire, 1975). Output control reduces complexity in a greater extent. Many organizations prefer output control rather than behavior control as it is easier and requires fewer resources.
It is argued that both output control and behavior control serve the same purpose and are freely substitutable which seems compelling. Thompson (1967: 86) made it clear, however, that such substitution could take place only under a very restrictive set of conditions: when goals were crystallized and means-ends relations completely understood; If goals were not agreed upon, then output control could not legitimately be used. If means-end relations were not understood, then behavior control was not functional, since the appropriate behavior was not known (Ouchi and Maguire, 1975).
On the other hand, clan or social controls are present in all organizations to varying degrees (Ouchi, 1979) (Langfield and Smith 2003). Social controls may develop from shared norms, values and beliefs, and may rely on the internalization of goals, which leads to organizationally desired behaviors. These controls cannot be designed explicitly, but can be shaped and influenced by activities such as frequent interactions, meetings, negotiation of disputes, codes of conduct, senior management attitudes and style, and rituals. There is some evidence that social controls are also relevant in inter-ï¬rm relationships (Gietzmann, 1996; Nooteboom et al., 1997) (Langfield and Smith, 2003).
Ouchi (1977) highlighted the relationships between the mode of control, and the information characteristics of the task: the degree of output measurability and task programmability (Langfield and Smith, 2003). Outcome controls were said to be suitable in situations of high output measurability and low task programmability, whereas behavior controls suit situations of low output measurability and high task programmability. Where both output measurability and task programmability are high, either behavior or outcome controls may be used. Social controls emerge when both output measurability and task programmability are low (Langfield and Smith, 2003).
2.2 Transaction Cost Economics (TCE) and Control
A common framework for viewing the choice of governance structures in inter-ï¬rm relationships is transaction cost economics (Langfield and Smith, 2003). TCE is based on the notion that ï¬rms choose efficient organizational forms or governance structures based on transactional issues, such as ï¬rm-speciï¬c investments, and external or internal uncertainty (Ittner et al., 1999) (Langfield and Smith, 2003). Thus, we can say which control system to adopt is dependent on the transaction cost of each of the control system.
Asset specificity is important in TCE as it is assumed that a high level of asset speciï¬city creates dependency between the parties in a relationship, increases switching costs and leads to difficult governance situations (Langfield and Smith, 2003). If the activities of two firms (or two departments) are highly asset specific to each other then the transaction cost will be low.
Governance structures can be characterized as one of three forms: markets, hybrids (including strategic alliances) and hierarchies (Williamson, 1991) (Langfield and Smith, 2003). Markets, hybrids and hierarchies rely on different control mechanisms to enable successful contracting. Markets rely on free competition to ensure control, and hierarchies rely on authority, whereas hybrid form of control generally entails long-term contracts. Hybrid governance structures are long-term contractual relations that preserve autonomy by providing added transaction-speciï¬c safeguards (Williamson, 1996, p. 378) (Langfield and Smith, 2003).
Van der Meer-Kooistra and Vosselman (2000) developed a model of control in inter-ï¬rm relationships that integrates TCE concepts and trust (Langfield and Smith, 2003). They identiï¬ed three management control patterns relevant to outsourcing relationships: the market based pattern, bureaucracy based pattern and trust based pattern (Langfield and Smith, 2003). Van der Meer-Kooistra and Vosselman (2000) provide a more complete analysis by considering not only the transaction characteristics, but the transaction environment, the characteristic of the parties to the transaction, and the role of trust in achieving control (Langfield and Smith, 2003).
The market based pattern suits transactions characterized by high task programmability, high measurability of output, low asset speciï¬city and high task repetition (Langfield and Smith, 2003). In the market based pattern many suppliers will compete for the contract, and market prices will be directly linked to the quality of the output of the outsourcer's activities. Detailed contracts are not required and the possibility of returning to the market for competing bids provides discipline for the current outsourcer to provide an efficient and effective output. In the face of effective market mechanisms, no speciï¬c control instruments are needed to manage the relationship (Langfield and Smith, 2003).
The bureaucracy based pattern, suits transactions that have high task programmability, high output measurability, moderate asset speciï¬city and low to medium repetitiveness. The transaction environment has relatively low uncertainty and the future is fairly predictable. Controls will be prescriptive and include detailed rules of behavior and rigid performance targets (Langfield and Smith, 2003).
The trust based pattern of Van der Meer-Kooistra and Vosselman (2000) is characterized by low levels of task programmability, low levels of output measurability and high asset speciï¬city. Transactions are not highly repetitive. The environment is highly uncertain and risky, so trust becomes the dominant mechanism for achieving control in this form of relationship, and this mitigates the risks associated with high asset speciï¬city. The initial selection of the outsourcer is based on perceptions of competence trust, contractual trust and goodwill trust, which arise through friendships, former contractual relationships and reputation. Initially, contracts are merely broad frameworks, which then develop further over time. A series of control devices, such as personal consultations and intense communications, are put in place to develop competence trust and goodwill trust. In addition, the institutional environment can stimulate the development of competence trust and contractual trust, through certiï¬cation of the ï¬rm's activities and legal regulations. Goodwill trust becomes the solution to overcoming information asymmetry, and regular personal contacts and an attitude of commitment can lead to its development. Control systems, in general, are more informal under these forms of relationships, and often take the form of social controls. Consistent with Ouchi (1979), behavior controls are not suited in these situations of low task programmability, and when controls are formalized they tend to emphasize outcome controls, and these develop over time through the sharing of private information and the alignment of the parties' performance expectations. Trust is necessary to achieve control, as activities and output cannot be measured with any certainty (Langfield and Smith, 2003).
2.3 Key variables
This model shows how the different control mechanisms from the theories will be used in the paper. There will be three control mechanism that will be used in the analysis: behavior-, output and trust based control. These three are together the base for identifying variables appropriate for the problem.
Input control: in order to se how AstraZeneca choose to have Razorfish as their system developer and how people involved in the project act against AstraZeneca.
Output control will be used to look at how AstraZeneca controlled the time put on developing the system, how AstraZeneca controlled cost limits or if they had any, how the task was divided and if the outcome was to their satisfaction.
Trust based control will be used to look at how the information sharing was between AstraZeneca and Razorfish. Did they have trustful information sharing or when they had discovered something, did they keep it for themselves?
3. Empery
AstraZeneca used a combination of internal and external group to develop LinkMedica. Caesarius (2008) stated, "LinkMedica was the result of a complex collaboration between a plethora of actors belonging to a vast area of communities; all with different professional backgrounds and personal drivers; with different beliefs and assumptions, and consequently, with different wills as well". Primary internal group consisted London based LinkMedica corporate project group which had overall responsibility for the project (Caesarius, 2008). Then there was Global e-business unit situated in New York (USA), Local project group of two piloting countries UK and Denmark. There were a number of external group also which includes Canadian e- health group, Consultancy agency (i.e: McKinsey), Creative agency (i.e: Razorfish), technical consultancy firms, information providers and advisor. Controls of both of these internal and external groups were a huge task for AstraZeneca.
3.1 Pre-production stage
The pre-production stage can be seen as a time when AstraZeneca bounce ideas back and forth constantly between different actors that was involved in the project. The aim with the project in this stage was to " combine the best of technology, medical research, and health care management to assist both patients and health care professionals in their effort to optimize care" (Caesarious 2008:p.154). This stage involved a lot of people in it that helped AstraZeneca to develop the best outcome of everyone's ideas. Razorfish was later introduced to the project when AstraZeneca needed someone that would make the system for the aims they had.
3.1.1 Selection
AstraZeneca had chosen to employ Razorfish as system developer of the LinkMedica. Caesarius (2008), stated "the company soon became a global actor in what was dubbed the digital solutions industry helping firms to leverage digital technologies across platforms and networks". Though the company was mainly a system developer but it was able to differentiate itself from others by playing two roles. It was a mixture between a creative agency dealing with business development and brand strategy operations and a system developer designing, producing, implementing, and integrating system (Caesarius, 2008). Razorfish strategy was to expand aggressively. Razorfish was present in both UK and Sweden where AstraZeneca also had operation specifically dedicated to the development of LinkMedica. Razorfish was interested to develop the (LinkMedica) system for AstraZeneca. AstraZeneca corporate project group in London selected Razorfish as a system developer where a number of factors supported the selection which includes Razorfish's credentials, their impressive list of well known international clients such as Shell Oil, British Aerospace, and Nissan and the movie star image of its founders (Caesarius, 2008).
3.2 Production stage
The production stage was a time when ideas were going to take actions and thoughts become reality. In this stage Razorfish showed AstraZeneca their thought about whom the system should be done and where it could be launched first. "There were a lot of explanations and so on but….. perhaps they had a good underlying idea but I could not get it. We did not accept it immediately, it took some while for us to do so" said corporate project manager (Caesarious 2008:p.185). It was here they tested the project on patients to see how people would react on the system. Communication was a high factor here between AstraZeneca and Razorfish and it turned out to be some big communications problems between them later on in the production stage.
3.2.1 The Blueprint
Razorfish started the project by creating a blueprint for how LinkMedica would be conducted which they described in a presentation for AstraZeneca. Razorfish purpose of the presentation was to define the brand space that is the operational space within which LinkMedica would create value to the involved parties. The purpose of the underlying document was to function as a common understanding among the project members and how to fulfill it. Once Razorfish started to work for AstraZeneca they made a number of recommendations. At first Razorfish declared four different reasons why AstraZeneca should pursue this knowledge management endeavor; First, LinkMedica as an effort to create an additional marketing channel to current and potential future costumers, second, LinkMedica would improve future development of more effective treatment by capturing anonymous patient data that would reside within the system. Third, LinkMedica was also to support the objectives of AstraZeneca's soon to be launched asthma product called Symbicort. Fourth, LinkMedica was to provide AstraZeneca a possibility to differentiate themselves from competitors by providing the best user experience in asthma management (Caesarius, 2008).
In the second phase, Razorfish moved to suggest that e-Asthma should be piloted in two countries: the UK and Denmark. It was also suggested by Razorfish that, AstraZeneca should not "over associate" the e-Asthma system with product launch like Symbicort, which was a new asthma product that AstraZeneca wanted to promote using LinkMedica project. In addition to above mention suggestions there was a lot of suggestions made my Razorfish.
3.3 The Voyage
The journey was not smooth for Razorfish. Soon after they [Razorfish London office and AstraZeneca corporate project group in London] started to work together they were not happy as a partner specially the corporate project group in London. As stated in Caesarius (2008), "Razorfish London office was supposed to collaborate with others on the project but suddenly something happened and AstraZeneca no longer wanted to work with Razorfish's London office". The situation was so bad that they [corporate project group] once used the term "kick out" to describe the action against Razorfish London office. However, coincidently some of the people from Razorfish Stockholm office were involved in the project and AstraZeneca wanted to continue with Razorfish Stockholm office as they [AstraZeneca] were happy with the performance of Stockholm office. Later, one person from Razorfish Stockholm office became the account director and was leading system developer team behind LinkMedica (Caesarius, 2008). AstraZeneca wanted to continue work with Razorfish, as they had no other option. As Razorfish account director once acknowledged: "it would cost time and money for the corporate project team [of AstraZeneca] if they were to establish relationships with another creative agency and time was not exactly something that the firm had in abundance". Due to limitation of time AstraZeneca started to work with Razorfish even without detailed strategies or plans (Caesarius, 2008).
3.2 Control Mechanisms
AstraZeneca's control of Razorfish consists two parts. In the first part they had set a number of conditions for Razorfish and one of them was the deadline for the project. "With a pending launch date of March 2001, our time line relatively aggressive. We would like to begin collaborative brand development work with a creative agency by November 15, and would there for invite responses……..given the tight deadlines and our desire to participate actively in the brand development process, we are seeking a better understanding of agency credentials, past experiences and preliminary ideas, without detailed strategies or plans" (Caesarius 2008; p183).
AstraZeneca wanted the project to be launched as soon as possible. Razorfish had problems to create the system to the deadlines because of the coordination problem in what the project was about. " The specification list was not detailed enough and that was one major reason why this project became so problematic. We were set to construct but we did not know what and how. There were no clear instructions… There were no coordination at all," said the information designer at the system developer (Caesarious 2008:p.204). Razorfish needed to keep contact with the clients to reach clarity all the time. As the system developers they wanted to make everything as AstraZeneca wanted and this took a lot of time to do. Time became a major constraint. During the production stage everything took longer time than expected. Razorfish were experiencing major problem coordinating the development with two market companies at the same time. The communication between AstraZeneca and Razorfish took longer time than usual. At one point of time Razorfish expected that communication time will be reduced especially when the project management and responsibility were transferred to the Lund office of AstraZeneca. But in reality it became longer and even more complex (Caesarius 2008, p. 205).
Second part was the control system itself. In the second part AstraZeneca controlled Razorfish with monitoring and evaluation of performance. For example, when AstraZeneca was not happy with Razorfish London office they decided not to continue with Razorfish London office. Something else AstraZeneca controlled Razorfish with was their performance with what they though should be included in the project, i.e the blueprint they made. This was a way for Razorfish to show their expertise in how the project should work, but AstraZeneca refused some of the ideas presented there and keep going on with their former ideas.
4. Analysis
4.1 Pre-production
In the case of AstraZeneca, both output measurability and task programmability was low in the pre-production part of LinkMedica. That was because there was no clearly defined goal more than that AstraZeneca wanted to develop their knowledge and to connect patients with doctors by creating an IT-system. There were no time constraint in the pre-production part and nobody knew exactly who came up with the idea from the very beginning.
More focus was put on the input control on this stage, which was manifested in deciding which supplier AstraZeneca would use to be the system developer and by selecting consultancies and advisors. The pre-production part was very much based on creativity and idea development it wouldn't be sufficient to use strict output control because that would the way we see it affect creativity negatively. Razorfish was chosen based on their credentials and their fast expansion impressed many. But, the story within the AstraZeneca was a bit different. There was some problem inside the AstraZeneca corporate project team who didn't have any kind of knowledge regarding system development. Caesarius (2008, p. 205) stated, "people that really didn't know anything about system design [for instance] participated just because they were part of the [corporate] project group, it was lot of that and it did not work". From our perspective, the input control mechanisms that AstraZeneca used when deciding to give Razorfish the responsibilities for the knowledge development maybe wasn't the best since Razorfish had never before worked within the pharmaceutical branch and had never been involved in such a big project like LinkMedica.
Razorfish was nearly thrown out of the project at the very beginning but AstraZeneca changed their mind after having done restructurings and had seen some of the work Razorfish had done. By questioning Razorfish this way, Razorfish knew that they had to do a good work and perform at the very best throughout the project otherwise they would be kicked out.
AstraZeneca told Razorfish that they wanted them to be responsible for the development of LinkMedica. After the first arguing about Razorfish competences it seemed like AstraZeneca trusted Razorfish very much since they gave Razorfish the opportunity to created the blueprint of the project. However, Razorfish wanted to use LinkMedica as marketing channel not only for patients and doctors but also for the Symicort something which AstraZeneca earlier had rejected officially. Razorfish and AstraZeneca made a compromise so that symbicort would be launched only indirectly with LinkMedica. This indicates that AstraZeneca had a great amount of trust for Razorfish since they changed their mind after Razorfish tough was best. This also indicates that the information sharing at this stage was good since Razorfish would share all their ideas even if they weren't aligned with AstraZeneca's. Moreover by prodiving a blueprint for AstraZenca, it became easier for AstraZeneca to control that Razorfish would fulfill what they had promise. It was also a very good way of making sure that everybody knew were the project was going and the information sharing was therefore god from Razorfish side. On the other hand AstraZeneca could have prodivded Razorfish with more information about the patients and the doctors and the results from the consultant firm. (I need a source in the empery on this)
4.2 Production
The journey was not smooth for Razorfish "I cannot honestly say that Astra Zeneca always knew exactly what they were looking for. There were a lot of wills I think; a lot of ideas…….there were lot of effort from all other group". Moreover many involved from Razorfish complained about being given a detailed specification list of what AstraZeneca wanted for the project. This made it troublesome for Razorfish but also for AstraZeneca how should you control a supplier if you don't give that supplier complete specification of instruction of what you want.
Though AstraZeneca used a combination of formal and informal control while controlling Razorfish where Informal control was mostly in use. We can see that social control emerged as both output measurability and task programmability was low in the system development work. It was difficult to for AstraZeneca to measure output or to divide the task of development as the system development work was a qualitative task rather than quantitative.
Although, AstraZeneca used output control mechanisms in order to put pressure on Razorfish. (I need a source on this in the empery)This could be seen in deadlines and focus on to spending too much money. This may not have been the most proper control mechanism to use here. That is since time and cost constraints can best be used when there is a known goal and the processes is widely known among the subordinates or the supplier. Furthermore, any problem may be keep among Razorfish since they don't want to give AstraZeneca the impression that they weren't capable of creating the system. Therefore we argue that information sharing may have been negatively affected and that the transparency between AstraZenca and Razorfish could have been better. AstraZenca for instance had problems in describing what they wanted and Razorfish didn't want to seem like they didn't understand what AstraZeneca wanted because AstraZeneca was the biggest Razorfish had ever had. Many times when Razorfish asked questions AstraZeneca wasn't able to provide them with an answer.
The goal wasn't known more than the blueprint which was developed by Razorfish. However, some of the production parts AstraZeneca was very much involved in themselves such as decided which color the Linkmedica should be in(I need a source on this in empery). This indicates that the AstraZeneca didn't have so much trust for Razorfish and therefore want to use direct control in order to make sure that LinkMedica would be a good product.
Social controls developed over time basically from shared norms, values and beliefs the both parties had. Social controls also developed through the internalization of goals. However, the controls was not designed explicitly rather it was shaped and influenced by activities such as frequent interactions, meetings, communication and negotiation of disputes. For example, at the very beginning of project there was dispute among AstraZeneca and Razorfish which was solved through negation. We therefore believed that the social control increased steadily throughout the Linkmedica project between Razorfish and AstraZeneca. However, the trust wasn't straightforward that is because AstraZenca was a bit schizophrenic and change their approach towards Razorfish throughout the project. In their control over Razorfish in some part they gave Razorfish a lot trust such as designing the blueprint and in other parts very little for instance with the color of LinkMedica.
During the development of LinkMedica a prototype was tested in Canada and France where both of them failed. French claimed it had too little focus on the patient and they wanted its parts to revolve around this target audience and Canadians also had similar concern (Caesarius, 2008). The system developers were confident even after that and argued a globally branded solution would work even the local [Canada and France] systems didn't work. AstraZeneca trusted Razorfish even though the Canadians and the French didn't which shows that AstraZeneca was very confident about the LinkMedica product.
5. Solution
*Don't use a tight time and cost framework if you want to create a product which is supposed to be revolutionary.
6. Conclusion
The control mechanisms used in the pre-production part differed from the ones that were used in the production part. From input control during the pre-production to strict output control used varied.
