In the hierarchical business organization of the past it was not uncommon to see average spans of 1 to 10 or even less. That is, one manager supervised ten employees on average. In the 1980s corporate leaders flattened many organizational structures causing average spans to move closer to 1 to 100. That was made possible primarily by the development of inexpensive information technology. As information technogy was developed capable of easing many middle manager tasks -- tasks like collecting, manipulating and presenting operational information -- upper managers found they could hire fewer middle managers to do more work managing more subordinates for less money.
The current shift to self-directed cross-functional teams and other forms of non-hierarchical structures, have made the concept of span of control less salient.
Theories about the optimum span of control go back to V.A. Graicunas. In 1933 he used assumptions about mental capacity and attention span to develop a set of practical heuristics. L.F. Urwick (1956) developed a theory based on geographical dispersion and the need for face-to-face meetings. In spite of numerous attempts since then, no convincing theories have been presented. This is because the optimum span of control depends on numerous variables including organizational structure, available technology, the functions being performed, and the competencies of the manager as well as staff.
However, exercising control over activities performed by sub-ordinates and monitoring their communication, the nodes at the upper hierarchical levels would be suffering from information overload, since all communication to other branches of the organizational structure would be routed through them. In addition, a larger number of sub-ordinates also requires supervisors to monitor a high number of interactions below their own level, i.e. that information overload and span-of-control are positively correlated.
Graicunas (Gulick and Urwick, 1937) distinguished three types of interactions–direct single relationships, cross-relationships, and direct group relationships–each of them contributing to the total amount of interactions within the organization. According to Graicunas, the number of possible interactions can be computed in the following way. Let n be the number of subordinates reporting to a supervisor. Then, the number of relationships of direct single type the supervisor could possibly engage into is
The number of interactions between subordinates (cross relationships) he has to monitor is
n (n - 1),
and the number of direct group relationships is
n (2n/2 - 1).
The sum of these three types of interactions is the number of potential relationships of a supervisor. Graicunas showed with these formulas, that each additional subordinate increases the number of potential interactions significantly. It appears natural, that no organization can afford to maintain a control structure of a dimension being required for implementing a scalar chain under the unity-of-command condition. Therefore, other mechanisms had to be found for dealing with the dilemma of maintaining managerial control, while keeping cost and time at a reasonable level, thus making the span of control a critical figure for the organization. Consequently, for a long time, finding the optimum span of control has been a major challenge to organization design. As Mackenzie (1978, p 121) describes it:
”One could argue that with larger spans, the costs of supervision would tend to be reduced, because a smaller percentage of the members of the organization are supervisors. On the other hand, if the span of control is too large, the supervisor may not have the capacity to supervise effectively such large numbers of immediate subordinates. Thus, there is a possible trade-off to be made in an attempt to balance these possibly opposing tendencies.”
Fayol proposed that subordinate employees should be allowed to communicate directly with each other, given that their superiors had agreed upon this procedure. This principle became known under the name of Fayol's bridge.
The use of Fayol’s bridge resulted in a number of other aspects needing to be taken into consideration. In order to put this system to work, Taylor’s functional foremanship has to be abandoned, and unity of command needs to be established. At the same time, decision power is distributed to individuals on lower levels in the organization, and only decisions that exceed the pre-defined decision scope of an employee are referred upwards. This, in turn, strengthens the co-equality of authority and responsibility. Since a Fayol-bridge is not limited to a certain functional area within the organization, but can span over functional boundaries, e.g. from purchasing to manufacturing, it can be considered as a first attempt to create a horizontal integration of related activities under a certain level of self-management, an early business process.
Mackenzie and others (Massie 1965, Pugh et al., 1972) also noted that there is no generally applicable optimum span of control. There are instead several factors influencing the balance between the desired level of control, and the manageability of the organization.
Firstly, it depends on the capabilities of the organizational members, managers and workers. It was assumed, that no manager would be capable of supervising more than 5-6 direct subordinates. However, this conclusion built on the assumption, that the superior must actively monitor the work of all subordinates. Later on, this statement was diversified, and Davis (1951) divided managerial work into two categories, one requiring the attention to physical work, the other one requiring mental activity. Depending on the type of supervision, a span of 3-8 subordinates for managers at higher levels was considered adequate, while first-level supervisors, i.e. those supervising shop floor personnel could have up to 30 subordinates.
The neoclassical theorists have developed a different solution. They assumed that a considerable amount of decisions could be delegated to organizational members at lower organizational levels. This solution would be equivalent to the application of Fayol's bridge combined with the principle of employee initiative that he proposed. As a result, the need for supervision would be reduced from direct control to exception handling. According to this assumption, they considered the opportunity of having access to a supervising manager would be sufficient to satisfy the need for control in standard situations. Peter Drucker (1954) refers to this principle as the span of managerial responsibility.