Three Cultures of Management: The Key to Organizational Learning in the 21st Century Edgar H. Schein MIT Sloan School of Management Table of Contents Abstract 1. The Predicament: Organizations Don't Learn, Innovations Don't Last or Diffuse 2. The Concept of Culture and Occupational Communities 3. Cultures and Sub-Cultures 4. Three Cultures of Management 5. The Engineering Culture 6. The Executive Culture 7. Dysfunctional Interactions Among the Three Cultures 8. Implications 9. The Dilemma of 21st Century Learning References Table 1 The purpose of this article is to provide some possible explanations for the failure of organizational innovations to occur in the first place, or to survive and proliferate. In other words, why do organizations fail to learn how to learn and therefore remain competitively marginal. The typical explanations revolve around vague concepts of "resistance to change," or "human nature," or failures of "leadership." I will propose a more fundamental reason for such learning failures deriving from the fact that in every organization there exist among its sub-cultures three particular cultures, two of which have their roots outside the organization and are therefore more fundamentally entrenched in their particular sets of assumptions. Every organization develops an internal culture based on its operational success, what I will call the "operator culture." But every organization also has in its various functions the designers and technocrats who drive the core technologies of the organization. I will call this the "engineering culture" and note that their fundamental reference group is their world wide occupational community. Every organization also has its executive management, the CEO and his or her immediate subordinates, what I will call the "executive culture." CEO's because of the nature of their jobs and the structure of the capital markets also constitute a world wide occupational community in the sense that they have common problems that are unique to the CEO role. These three cultures are often out of alignment with each other, and it is this lack of alignment that causes the failures of organizational learning as the below examples will show. This will raise thi question of whether we have misconceived the initial problem by focusing on organizational learning when, in fact, it is the executive and engineering communities that must begin their own learning process if 21st century challenges are to be met. [Back to the Table of Contents] The Predicament: Organizations Don't Learn, Innovations Don't Last or DiffuseThe ability to create new organizational forms and processes, to innovate both in the technical and organizational arenas, is crucial to remaining competitive in an increasingly turbulent world. But this kind of organizational learning requires not only the invention of new forms, but their adoption and their diffusion to the other relevant parts of the organization and to other organizations in a given industry. Organizations still have not learned how to manage that process. The examples of successful organizational learning we have seen either tend to be short-run adaptive learning, doing better at what we are already doing, or, if they are genuine innovations, they tend to be isolated and eventually subverted and abandoned. Example 1. A new product development team in a large auto company worked with the MIT Organizational Learning Center to develop their capacity for learning. By using various techniques derived from "action science" (Argyris, et al, 1985), systems dynamics (Senge, 1990), and organization development (Beckhard & Harris, 1987) high levels of openness between hierarchical levels and increased communication and trust among members of the teams were created. This openness and trust permitted team members to reveal engineering design problems as they arose instead of waiting until they had solutions, as prior traditions in this company had dictated (OLC Learning History, 1995). Early identification of those problems was crucial in order to avoid later interactive effects that would require costly and complex redesigns. For example, changing the chassis design might increase weight which might require a different tire design which, in turn, might cause more internal noise, and so on. By revealing such problems early, the team could view the whole car more systemically and re-design could therefore be speeded up. However, the pile-up of early problems caused higher management to make a false attribution. They considered the team to be "out of control" and ordered it to get itself back under control. The team realized that higher management did not understand the value of early problem identification and continued to use its new learning, assuming that the ultimate results would speak for themselves. The team was able to complete the design well ahead of schedule and with considerably lower costs, but, contrary to expectations, higher management never understood the reasons for these notable results nor gave the team credit for having learned a new way of solving problems. Instead, higher management gave itself credit for having gotten the team "under control." The team was not considered to be particularly innovative and was disbanded. Several of its members and leaders were subsequently encouraged to take early retirement as part of a general downsizing program in the company. Example 2. An insurance company decided to move toward the paperless office (Roth, 1993). Top management hired a key manager to implement the new system, mandated a schedule, and provided whatever resources the manager needed to accomplish the task. In order to use the new system, employees had to learn a complex new set of computer routines to replace their familiar work with paper. Because the company was also under financial pressure, it had instituted a number of productivity programs that caused line managers to be very insistent that all the daily work continue to be performed even while the learning of the new system was supposed to take place. The new manager was equally insistent that the system be implemented on schedule causing employees to short-circuit certain routines, to learn only the rudiments of the new system, and even to misrepresent the degree to which they were now working without paper. The new manager, based on partial and incorrect information, declared that the system was implemented "on schedule" and was given public