Oct 25 2012

How to Avoid Slow IT Death

The Veterans Affairs Department has improved the success rate of its IT projects by using short, tight delivery timelines.

When the Department of Veterans Affairs paused development of its IT projects in 2009, we knew our system was broken — but we didn’t know how or why. We just knew that we weren’t delivering. VA received significant attention for stopping 45 projects and initiating a top-to-bottom review of what was going wrong. The department wasn’t unique in its inability to deliver. What was unique was that we said, “Stop!”

Before putting the brakes on delivery, we examined what high-performing organizations in the private sector were doing to successfully deliver projects. We learned that the most important constraint was the time taken to deliver a project. To use a football analogy, projects were most likely to succeed when delivery was focused on a series of first downs as opposed to throwing a Hail Mary pass for a touchdown.

Of the 45 projects we paused in 2009, most were several months to more than a year behind schedule. Largely because of the time slips, nearly every project suffered from other kinds of project-killing constraints: Half of our projects were significantly over budget, and nearly all had an increasing number of defects.

Time is an important measure of risk. Over time, nearly every facet of a project can change: requirements, budgets, acquisitions processes and rules, infrastructure and people. These changes reduce the probability of successful delivery. Our proposed approach was to manage projects by time and mitigate change risks with incremental deliveries.

By using short, tight delivery timelines, we’ve tamed many of the delivery problems that plagued us in 2009 and before. Projects must produce a customer-facing deliverable every six months; even shorter delivery timelines are encouraged. As long as the customer and the project manager understand that the delivery date must be met, the project can adapt to change in other areas — budget, people, requirements, etc. — and still deliver a customer-facing product on time.

Our incremental delivery approach has been successful. Since we implemented incremental, time-based deliveries, we have delivered more than 80 percent of our increments on time. Of those that were late, half made their delivery within two weeks of the deadline. Industry estimates are that 42 percent of IT projects deliver successfully.

Enforcement Mechanisms

One of the most important aspects of keeping our project managers true to their delivery schedule is the VA’s “three strikes and you’re out” rule. Any project that misses a milestone — even by a day — receives a strike. Just as in baseball, three strikes means you’re out. At VA, that means the project is stopped: The project’s contracts are terminated, we send the project manager for additional training and we reevaluate the requirements.

Our flag system lets project managers indicate areas where leadership should be paying attention to a project. Red flags indicate that senior leadership involvement is needed to avoid an imminent strike. Green flags recognize projects delivering ahead of schedule. Yellow flags are a new concept we introduced. When a project manager raises a yellow flag, it lets leadership know that the project is encountering some kind of change in a critical area and risks missing a delivery date. If other risks threaten to throw the project off of its timeline, the yellow flag lets leadership intervene. Yellow flags are proving especially useful when used at the beginning of a project’s lifecycle and then managed throughout the delivery process.

So far, we’ve been effective at delivering IT products by focusing on time to delivery. We’ve scored touchdowns with software solutions, such as the G.I. Bill claims processing, Blue Button health data downloads, eBenefits portal and MyHealtheVet electronic health records. We’ll continue to use this approach toward time-based incremental delivery as we continue to innovate at VA.