Throughout the 1990s, most large industrial companies installed enterprise resource-planning (ERP) systems—that is, massive computer applications allowing a business to manage all of its operations (finance, requirements planning, human resources, and order fulfillment) on the basis of a single, integrated set of corporate data. ERP promised huge improvements in efficiency—for example, shorter intervals between orders and payments, lower back-office staff requirements, reduced inventory, and improved customer service. Encouraged by these possibilities, businesses around the world invested some $300 billion in ERP during the decade.

What most attracted many a chief information officer was the opportunity to replace a tangle of complex, disparate, and obsolescent applications with a single Y2K-compliant system from a reputable and stable vendor; one major oil company, for example, managed to switch off 350 old systems when ERP went live. By entering customer and sales data in an ERP system, a manufacturer can generate the next cycle’s demand forecast, which in turn generates orders for raw materials, production schedules, timetables for shifts, and financial projections while keeping close track of inventory.

For many businesses, installing ERP was traumatic. Following long, painful, and expensive implementations, some companies had difficulty identifying any measurable benefits. Those companies that were...