Organization structure must apply one or both of two principles.
It must whenever possible integrate activities on the principle of federal decentralization, which organizes activities into autonomous product businesses each with its own market and product, and with its own profit and loss responsibility. Where this is not possible it must use functional decentralization, which sets up integrated units with maximum responsibility for a major and distinct stage in the business process.
Federal decentralization and functional decentralization are complementary rather than competitive. Both have to be used in almost all businesses. Federal decentralization is the more effective and more productive of the two. But the genuinely small business does not need it, since it is entirely an â€œautonomous product businessâ€. Nor can federalism be applied to the internal organization of management in every large business; in a railroad, for example, the nature of the business and its process rule it out. And in practically every business there is a point below which federal decentralization is no longer possible, below which there is no â€œautonomous productâ€ around which management can be organized.
Functional decentralization is universally applicable to the organization of management. But it is second choice for any but the small enterprise. It has to be used in all enterprises sooner or later, but the later it can be resorted to, the stronger the organization.
Decentralization whether federal or functional has become so prevalent in industry during the last few years it has become a household word. Its practice goes back at least thirty years. DuPont, General Motors, Sears and General Electric all, started to develop their decentralized organization before 1929.
Yet organization theory has paid little attention to it. General Motors in 1946 was the first to consider decentralization as a distinct principle of organization.
The reason for this lag is that conventional organization theory starts with the function inside a business rather than with the goals of a business and their requirements. It takes the functions for granted if not for God-given; and it sees in the business nothing but a congeries of functions.
Moreover, conventional theory still defines a function as a group of related skills. And it considers this similarity of skills to be both the essence of functionalism and its major virtue. If we look at well-organized functional units, however, we shall find no such â€œbundle of skillsâ€. The typical sales department, for instance, includes selling activities, market research, pricing, market development, customer service, advertising and promotion, product development, often even responsibility for relationships with governmental bodies and trade associations. And the typical manufacturing department covers an equally wide range. No greater diversity of skills, abilities or temperaments could be imagined than that needed in these â€œfunctionalâ€ organization. Indeed, no greater variety exists in the business as a whole. If functionalism were really, as the books say, organization by skill-relationship, the typical sales or manufacturing department would be absurd if not totally unable to function. But they work indeed, they work much better than units organized on similarity of skills because they bring together all the specialized activities needed in one fairly sharply delimited stage of the work. That they require different skills and different temperaments is irrelevant; what matters is that they bring together what is objectively needed for performance. The first principle of good production organization is to bring the machines to the work rather than the work to the machines. It is cheaper to have the work flow according to its own inner logic, even if it requires a few more machines, than to cart materials around. Similarly, we must always bring the special activity to the work, never the work to the special activity. For ideas and information cost even more to cart around than materials.
The stress on functional organization by related skills is thus a misunderstanding of what functional organization properly should be organization by stage of process. This is illustrated by the unsatisfactory experience with those functions that are typically organized as bundles of skills: accounting and engineering. The typical accounting department is in constant friction with the rest of the organization. The typical engineering department has constant difficulty working out its objectives or measuring its performance. Neither condition is an accident.