Recentralizing IT
Companies can run their IT systems more efficiently by creating new organizational structures in which IT departments and business units share responsibility.
Bradford Brown, James M. Kaplan, and Thomas Weber
May 2003
Companies know that centralized IT infrastructures—the hardware, operating systems, networks, and services that support IT systems—can cut their costs.1 What is more, centralized infrastructures are more reliable.2 Yet many business managers have resisted efforts to tighten up this crucial aspect of their operations, preferring a model that steers away from the highly centralized IT departments of earlier years and toward a more decentralized, customized approach, which increases not only flexibility but also inefficiency and conflict.
A number of large banks, telcos, and consumer goods manufacturers may have found the answer to that problem: a middle way that creates new financial and governance structures for IT organizations and the business units they serve. In the old versions of centralization, IT infrastructure costs were simply allocated among business units, which had little control over the system and few incentives to manage its usage effectively. The new model strikes a balance between the simplicity of central control and the transparency and accountability of local control (Exhibit 1).
At the center of this new arrangement are service managers, who usually have technical backgrounds but later graduated to management. Service managers work for IT organizations and support specific technologies, such as servers or mainframes. They gather requirements from their client organizations and develop price guides (for instance, a choice between high- or low-end servers and laptops), much as they would for hardware companies or systems integration firms. Before each budget year, they provide price quotations for available products and service agreements; a service manager might charge $300 a month for each server supported, for example, and guarantee 99.9 percent uptime.
Within the business units, demand managers, who may have some technical training but usually concentrate on business, assess the units' needs and select products and support from among the choices presented by service managers. At one global company, one unit's demand manager, given the means to control costs and set service levels, shaved 25 percent off its help desk expenses by differentiating between high- and low-priority calls and by deferring less urgent requests (Exhibit 2).
Operations managers—each of whom handles a technology cluster such as data centers, servers, or network equipment—have a strong technical orientation and work in the centralized IT infrastructure organization. They oversee the day-to-day delivery of IT services and are accountable for the business units' specific performance goals, such as the time it takes to install new hardware or to fix systems faults.
Central IT departments manage the supply of products and services—data lines, servers, technical support, and so on. The business units, in turn, base their requests for services on how much data they transmit as well as on the number of servers and the level of help desk support they feel they must have. Businesses get the service they require and pay for it. They therefore have more incentive to buy only what they need, while the IT group must hold to the previously agreed cost and service levels—an arrangement that alleviates the suspicions of managers that IT departments are either overcharging or underserving them.
This service-based approach has helped both to improve the reliability of corporate IT infrastructures and to reduce spending on them by as much as 30 percent; the more a decentralized infrastructure is consolidated, the more savings a company can expect (Exhibit 3). Within that consolidation process, the company can decide which other functions, such as the development of applications and the management of World Wide Web sites, ought to be centralized. Moreover, by clarifying its performance requirements and the cost of IT services, the company puts itself in a better position to assess its opportunities for outsourcing them. Ultimately, such decisions must be based on long-term strategic considerations, such as the uniqueness of a company's IT requirements (which would argue for in-house control) or the need to scale up its IT operations quickly (which would argue for outsourcing).
But IT infrastructures should be consolidated in any case. The system we describe is more complex to administer than are traditional approaches to IT; for one thing, the IT group must know how to set prices to reflect the true cost of services. Transparent pricing and clear accountability, however, bring economic and operational benefits—and may at last allow IT and business managers to find a model that suits both parties.
About the Authors
Brad Brown is a principal and James Kaplan is an associate principal in McKinsey's New York office, and Thomas Weber is a consultant in the Stamford office.
Notes
1Infrastructure typically accounts for some 40 to 60 percent of a company's total IT costs, and 20 percent or more of those expenses can be trimmed by consolidating and standardizing hardware and services.
2Quality control is a major concern. Well-managed, centralized infrastructures cut the number of "severity 1 faults"—systems outages that directly reduce a company's revenue or adversely affect its reputation—by about 50 percent. This kind of improved performance can save a large company tens of millions of dollars a year in operating costs.
Posts
