Unbundling the corporation

Abstract

The forces that fractured the computer industry are bearing down on all industries. In the face of changing interaction costs and the new economics of electronic networks, companies must ask themselves the most basic of all questions: what business are we in? In the late 1970s, the computer industry was dominated by huge vertically integrated companies such as IBM, Burroughs, and Digital Equipment. With their vast scale advantages and huge installed bases, they seemed unassailable. Yet just ten years later, power in the industry had shifted: the behemoths were struggling to survive while an army of smaller, highly specialized companies was thriving. What happened? The industry's sea change can be traced back to 1978, when a then-tiny company, Apple Computer, launched the Apple II PC. The Apple II's open architecture unlocked the computer business, creating opportunities for many new companies that specialized in producing specific hardware and software components. Immediately, the advantages of the generalist--size, reputation, integration--began to wither. The new advantages--creativity, speed, flexibility--belonged to the specialist. The story of the computer industry illustrates the crucial role played by interaction costs in shaping industries and companies. Interaction costs represent the money and time expended whenever people and companies exchange goods, services, or ideas [1] The exchanges can occur within companies, among companies, and between companies and customers, and they can take many everyday forms, including management meetings, conferences, telephone conversations, sales calls, reports, and memos. In a real sense, interaction costs are the friction in the economy. Taken together, interaction costs determine the way companies organize themselves and form relationships with other parties. When the interaction costs of undertaking an activity internally are lower than the costs of undertaking it externally, companies tend to incorporate it into their own organization rather than contract it to outside parties. All else being equal, a company will organize itself in whatever way minimizes its overall interaction costs. Apple's open architecture sharply reduced interaction costs in the computer industry. By conforming to a set of well-documented standards, companies could, for the first time, work together easily to produce complementary products and services. As a result, tightly coordinated webs of specialized companies, with names such as Adobe Systems, Apple, Intel, Microsoft, Novell, and Sun Microsystems, could form and ultimately compete effectively against the entrenched, vertically integrated giants. Many of the new companies grew very large very quickly but never lost their focus on specialized activities. The moral of the story is that changes in interaction costs can cause entire industries to reorganize rapidly and dramatically. Today that fact should give all managers pause, for we are on the verge of a broad, systemic reduction in interaction costs throughout the world economy. Electronic networks, combined with powerful PCs, are permitting companies to communicate and exchange data more quickly and cheaply than ever before. As business interactions move on to electronic networks such as the Internet, basic assumptions about corporate organization will be overturned. Activities that companies have always believed to be central to their business will suddenly be offered by new, specialized competitors that can undertake those activities better, faster, and more efficiently. Executives will be forced to ask the most basic and discomforting question about their companies: what business are we really in? The answer will decide their fate in an increasingly frictionless economy. One company, three businesses Look beneath the surface of most companies and you will find three kinds of businesses: a customer relationship business, a product innovation business, and an infrastructure business. …

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 299 citations exist upstream.