The Technology Beast

y now, it's a truism that a big information technology project has less than a 50-50 chance of achieving its goals-whether in the private sector or in government.

In fact, when the Health Care Financing Administration pulled the plug on its Medicare Transaction System development project in 1997, it almost elicited yawns around Washington. No one seemed at all surprised that the agency could spend $50 million or more over six years and then declare its most important systems project an un-fixable disaster.

What is surprising, is that hundreds of big federal IT projects have run their courses, with varying degrees of success, but the same fundamental errors keep occurring.

There's an entire industry devoted to advising chief information officers and program managers on how to succeed at this very difficult task. Whether a national accounting-consulting firm, the General Accounting Office, a technical services vendor, the Office of Management and Budget, a professional journal or any of hundreds of speakers at any month's management conferences, they all say much the same thing:

• Get top management commitment to your project and stakeholder buy-in.
• Have a performance plan and measure your progress. "If you're not measuring it," says National Weather Service project manager Paul Nipko, "it's getting worse. If you are measuring it, it will improve."
• Be sure your IT project is aligned with the agency's mission and strategies.
• Don't bite off more than you can chew. An eon ago, the General Services Administration's Frank McDonough warned us more elegantly to forsake "grand designs."
• Re-engineer business processes before automating them.
• Define an agency IT architecture; then use it as a management tool.

Enough. Determined not to produce another didactic management tome, we at Government Executive still felt the topic of IT project management warrants further examination. Its importance is not diminishing. In fact, given the widespread agreement on the need to embrace e-government, project management can only become more important.

So we zeroed in on five good-sized projects representing a spectrum of challenges. Some, such as the Weather Service's new digital nervous system, have millions of customers. Others, like the Army Flow Model, are internal systems for use by staff units. You'll find successes and failures, and a couple that could go either way. They're a mixed bag and, we hope, a representative (if unscientific) sample of what our readers who manage IT projects are up against.

While researching this special report, we talked to experts and came away with fresh insights into the challenges federal managers face. For example, it's hard to find people to take on these tough assignments.

In Short Supply

Paul Wohlleben, former deputy chief information officer at the Environmental Protection Agency, points out that the government does not have a corps of seasoned project managers who are available for massive IT efforts. Typically, he says, agencies turn to someone within their own walls -perhaps an IT manager who has not headed a large project before, or a mid-level program manager with little direct IT experience. It's hard to get such people trained and seasoned before the first real problems crop up.

The Senior Executive Service once was envisioned as the kind of management corps Wohlleben describes. But SES members have turned out to be much less mobile in their careers than the service's founders expected.

Now that government is outsourcing more work, especially IT, "they're going to need more project managers to manage industry," says Wohlleben, who now heads the government group at the Grant Thornton accounting and management services firm. With experienced federal managers retiring at an accelerating rate, he advises agencies to beef up their management training and development programs now.

Carol Kelly, a vice president of the META Group consulting firm and a former California state government manager, agrees. Her company predicts that DoD-style program management offices will become more common in federal IT organizations during the next couple of years.

Internal Consultants

The Information Technology Resources Board, a little-known intergovernmental body that quietly advises agencies on how to salvage troubled IT projects, has issued a series of useful publications on the issue, including "Project Management for Mission-Critical Systems."

Among its insights: Look for a program manager who can build consensus while being willing to take risks to move the program forward. Match project team competencies with the tasks at hand, then get additional expertise as needed. Balance the mix of contractor and government personnel. (This and other ITRB publications are available at www.itrb.gov.)

ITRB counsels managers to make sure they know where the risks are. For example, the organization's project management pamphlet says that "stakeholders' expectations are rarely spelled out in legislation, executive orders or formal memoranda." That certainly has been the case for the Customs Service's so-far-fruitless effort to find financial and political support for modernizing its major business systems.

Is COTS the Answer?

Another ITRB report advises agencies on how to assess the risks of the using commercial, off-the-shelf (COTS) software, as opposed to hiring a systems integrator to build a custom software system. Many agencies are choosing COTS software in hopes of reducing the risks associated with developing a system from the ground up and in the belief that COTS can be deployed faster. In this supplement, we look at how the intelligence community used COTS software in developing the new Intelink network.

Harry B. Heisler, vice president and general manager of Micron Government Computer Systems Inc., advocates the COTS approach. "The overall risk in any project," he says, "is not having enough respect for the things you do not know. The fewer of those inadvertent dark corners you leave unlit, the higher your chances of success." With standard COTS products as an element of a solution, agencies get market-tested technology, Heisler says.

What's more, he says, "your leverage is much smaller" if you find yourself locked into customized technology that's out of the mainstream. And upgrading or updating commercial products, as will always be necessary, tends to be less expensive and less technically challenging than doing the same updates for custom systems.

But the COTS route is by no means risk-free. "The majority of COTS solutions require extensive customization to meet the needs and support the business processes of the federal environment," according to an ITRB report, "Assessing the Risks of Commercial-Off-The-Shelf Applications." The report concludes that "federal agencies must make major business process reengineering changes to use COTS solutions as delivered."

Scale Matters

Perhaps more important, COTS products seldom are built to handle the huge and highly complex tasks of the federal government.

In a recent account of difficulties with installation of a new financial management system for NASA, the project manager was quoted as saying the prime contractor may have underestimated the size of the job.

Scale is an issue for many federal projects, whether the planned system will employ COTS products in only a minor role or to a great extent.

There's no way to shrink the ultimate size of a national tax processing system or weather system that collects, moves and processes thousands of bits of information every second. But it is possible to reduce the scale-related risks of big projects by breaking them into manageable modules.

That introduces another risk-the possibility that the modules won't work seamlessly together as an integrated whole at the end of the project. To guard against this, a well-thought-out architecture and plan must be in place from the start.

Reap the Rewards

But a modular development approach has rewards as well as risks. It allows the project manager to produce some results early. The experts recommend sequencing the project so that the first modules are those with customer impact. That means they'll be visible to a large, external audience. Positive response to the initial modules can be a wonderful vaccine against the inevitable arrival of the project flu-the difficult times when success is a distant memory.

"The slow, systematic, gradual approach to improvement just doesn't cut it," Wohlleben says, now that young entrepreneurs are the nation's heroes and many companies are operating on "Internet time." Like many of the other people we interviewed, he says smaller projects with narrower scope-projects that can be completed faster-are the order of the day.

In the end, the best recipe for success may be to avoid mammoth modernizations and elephantine systems development projects altogether. The Health Care Financing Administration seems to have adopted that strategy now, as it quietly proceeds to consolidate its major systems and bring them into the 21st century.

If agency executives adopt a posture of continuously refreshing the technology in their IT infrastructures and business systems, so that each year they invest a similar amount of money and upgrade an important portion of their IT, then they will never need to beg funding sources for big pots of extra money. They will never have to scour the woods for people with experience to manage a huge IT project. They will always have relatively up-to-date systems and satisfied users. And they never will have to explain to a congressional committee or a GAO auditor or a magazine reporter what went wrong with their project.