Enter Clusters, Exit Hierarchy

Tom Peters

In his exciting new book, Rebirth of the Corporation Harvard Professor D. Quinn Mills boldly proposes a true, sweeping alternative to hierarchy: “the cluster organization.” Mills argues for “dramatic change in the structure of the organization.” Where hierarchy remains, he flatly states, “there cannot be any rethinking of the fundamentals of management.”

Enter the cluster. Mills defines it as “a group of people drawn from different disciplines who work together on a semipermanent basis.” The cluster “handles many administrative functions, thereby divorcing itself from an extensive managerial hierarchy. A cluster develops its own expertise, expresses a strong customer … orientation, pushes decision-making toward the point of action, shares information broadly, and accepts accountability for … results.”

Such largely self-sufficient clusters vary in size from 30 people to 50 people (further subdivided into work teams of five to seven). In complex settings, you may find six varieties of cluster: (1) “a core team” (what used to be called top management); (2) “business units” (clusters that have external customers and conduct business directly with those customers); (3) “staff units” (clusters that have internal customers, but operate in accordance with market dictates); (4) “project teams” (assembled for a specific project); (5) “alliance teams” (joint ventures with outsiders); and (6) “change teams” (created to modify broad aspects of the corporation’s activity).

Four factors are essential if the cluster is to perform well, Mills tells us. First, goals must be clear. “Each person must know and understand the mission of the team with which he or she works,” he writes. Second, “to act on their own initiative, people must have the necessary competence. This requires not only technical specialization, but a grasp of the broader picture as well.” Perpetual training, for instance, must be viewed as an investment rather than a cost. Third, Mills joins the chorus insisting upon freely shared information: “To make the correct choices in local circumstances, people need (not) just local information … but information about the overall setting in which they are acting.” Finally, “people need to know they are trusted; that they will not be unfairly penalized for … failures.”

Mills deftly deals with common objections to clusters. Here’s a sample:

Objection No.1: “Clusters ignore the innate human need for hierarchies.” To the contrary, Mills says that “much of what we perceive as human nature in the workplace is nothing more than a thorough adaptation of individuals to a hierarchical context.” Distinctions among individuals based on ability do exist in clusters: “Natural leaders emerge and play a disproportionate role in the activities of the group. (But) a formally designated hierarchy is not required for leadership to emerge.”

Objection No.2: “It is impossible to motivate people without the prospect of (traditional) promotion.” Mills concedes the importance of promotion. But his research shows that “recognition for job accomplishment and self motivation” are even more important. Tapping these deep impulses will more than offset the absence of hierarchical rungs to climb.

Objection No.3: “Individual performance in a cluster will ‘level down’ to the lowest common denominator.” Leveling down, Mills observes, is more likely to occur in a hierarchy than in a cluster: “Why does any employee try to get away with as little work as possible? Because the game of trying to outsmart the supervisor is more interesting than the work itself.” But there’s no supervisor to outsmart in the cluster! Furthermore, Mills finds, “peers (in clusters), who depend on each other to carry part of the load, are very intolerant of shirkers.”

Objection No.4: “There will be no effective limit on the mistakes made by employees in a cluster because there is no direct supervision.” People on interdependent teams, Mills retorts, are “quick to notice and object to errors” on the part of their teammates. Moreover, individual performance evaluation will continue to exist, weeding out those who “repeatedly make serious errors.”

Objection No.5: “There won’t be any middle managers left, and those of us who are middle managers will lose our jobs.” Paradoxically, Mills suggests, there may be more middle managers in the cluster organization. That is, traditional “middle management tasks”—e.g., starting projects that add value—will become everyone’s job.

Mills supports these controversial notions with solid case studies about the transition to clusters in parts of General Motors and the U.S. Army, as well as at General Electric, Square D, Volvo, SwissAir, and British Petroleum. He doesn’t use Europe’s extraordinary ABB Asea Brown Boveri as a case, but it surely fits his thesis: Chairman Percy Barnevik recently told the Harvard Business Review that he has broken the 240,000-person, heavy-industrial firm into 4,500 profit centers, averaging about 50 people each (including virtually all former staff functionaries). Talk about large-scale validation for the idea of clusters!

(C) 1991 TPG Communications.

All rights reserved.