Twenty years ago, as Jack Welch was getting ready to take over as chief executive of General Electric Co. (www.ge.com), Peter Drucker asked him a simple question: “If you were an outsider, which of GE’s businesses would you want to buy?” As Welch stewed over the question, he arrived at a famous rule that underpinned GE’s phenomenal success. Welch decided that he wanted only businesses that were No. 1 or No. 2 in their industries.

It’s time we all asked ourselves a variant of the Drucker question, about our information-technology operations. The question is: Which of our I/T projects, systems, and people would we want to “buy” if we were outsiders in the process of building an I/T department from scratch?

Companies are already poring over their I/T budgets, trying to weather the recession by looking for cuts. At the same time, companies understand that they are increasingly competing based on their use of technology, as they try to improve their ability to understand and serve customers, to operate efficiently, and to cooperate effectively with partners. In other words, companies are hoping to cut while still putting themselves in a better competitive position.

It’s hard to accomplish that complex task through incremental changes. Instead, you need to take out a clean sheet of paper, as Welch did, and figure out what your ideal I/T operation would look like. Even if you don’t end up restructuring your operations as ruthlessly as Welch reshaped GE, you can still keep the ideal in mind as a goal to work toward.

As you develop your ideal, I suggest four principles:

You need to synchronize your I/T investments with the overall business’s strategy. That may seem obvious, but many companies’ I/T projects are out of synch with the overarching strategy. A recent study found, for instance, that while 85% of spending on new I/T projects was related to customer relationship management, only 15% of senior executives said CRM systems were critical to the business.

Aligning I/T with corporate strategy mainly requires building processes that will force precise, continual communication between I/T and senior operating executives. There also are frameworks that can help spot problems and facilitate communication. The best I’ve seen in some time comes from Peter Weill, director of the Center for Information Systems Research at MIT (www.mit.edu). His extensive research helps companies see how their mix of I/T investments matches up against corporate goals that range from extreme focus on cost-cutting to extreme focus on agility and innovativeness. [See “Measure for Measure,”] Weill says that if two companies spend the same amount on I/T, the returns they get on their investments can range from 40% better than average to 40% below average, depending mostly on how well investments line up with strategy.

You should consider setting up your I/T operations like a stand-alone business. Many companies already use a “shared services” model that has I/T basically selling its services throughout the company. Some chief information officers carry the title of CEO of the I/T business.

It makes sense to think of the CIO as the chief strategist, making sure that I/T stays in synch with corporate strategy. But at many companies, the CIO is also expected to have an incredible array of additional business and technical skills. Instead, in many instances, it’s smart to have a separate chief operating officer, who makes sure that the trains run on time, plus a chief financial officer and a chief marketing officer, who helps sell I/T projects to potential users. I/T now accounts for 50% of all capital expenditures, so providing additional, senior people makes sense at many companies.

Like projects at other business units, each I/T initiative should have a manager who focuses on business, not technology, issues. Having someone continually evaluating whether a project will deliver the value that it was initially designed to deliver will keep projects on track or, at the least, kill off projects before they eat up too much funding.

I/T operations should operate on a companywide basis. I know that many business units want their own I/T capability because a centralized I/T operation may move too slowly for their needs. But the need for companywide cooperation on data generally should override those concerns. As a fundamental part of their competitive strategy, lots of companies are trying to get a complete picture of their customers so as to manage relationships better and to sell them more related products and services. That cooperation simply can’t happen if individual business units handle data differently and don’t share information.

The CIO of a Fortune 500 company told me recently that departments within the company had ordered up 25 customer-relationship-management systems—all designed to provide a single view of the customer. That’s silly. And she isn’t alone.

In addition to providing better data, centralizing operations can lead to the elimination of such overlapping projects and, our experience shows, cut 15% to 20% from an I/T budget.

Buying I/T products and services on a companywide basis can also produce huge cost savings. Simply moving from two personal-computer suppliers to one can produce enormous benefits.

I/T departments should take a more flexible approach to people. At the moment, departments generally try to use internal talent on projects and try to grow those talents through training. The problem is that technology changes faster than training can keep up with. Just in the past few years, the most common programming language has gone from C++ to Visual Basic to Java. Many I/T departments are finding their skills obsolescent.

Instead, companies should map out the I/T skills they possess or are close to developing internally, then figure out which they should go outside for, whether on a part-time or full-time basis. These outsiders can be expensive, but they can also be extremely cost-effective. Studies have found that top programmers can literally be 20 times as productive as average programmers—and now is a great time to hire the rock stars full-time because so many of the start-up companies that they went to have failed. Meanwhile, being more flexible about where you find talent means you can keep a much smaller core I/T staff—as little as half the current size. You also complete projects faster because you can quickly line up all the talent you need.

While broad change is always hard, it has been many, many years since there has been a better time to use principles such as these four to rethink our approach to I/T. The recession and the Afghan war have convinced people that they simply can’t keep doing business as usual.

Curran is a partner with DiamondCluster International Inc. He can be reached at chris.curran@diamondcluster.com.