Project Too Big? Break It Apart

In the more than 20 years I’ve been a chief information officer, I’ve found the most difficult thing in introducing new technologies isn’t putting in the hardware and software. It’s changing how people work that is difficult. When you don’t prepare people for change, that’s when you get breakage or projects that aren’t successful.PDF Download

Another key is having detailed metrics, detailed tracking. I have a requirement on any large project that every two weeks, all the principals sit down and discuss where they are. This is a variation on the theme of establishing milestones—we call them “inchstones.” There are two advantages to the inchstone approach. First, nothing becomes a huge problem before you know about it, and second, you know every minute of the day exactly where you are in relation to your schedule and cost on a project.

Out of all the lessons I’ve learned, that’s the big one. I don’t care how difficult or how big a project is: If you break it into small pieces, you can manage it.

This is the infancy, still, of information technology—we’re just starting. It may not seem that way in the wake of the dot-coms turning into dot-bombs and the market falling apart. But one result of the ride up has been the development of an incredible amount of new software—a lot of which is still around even though some of the companies that created it have gone bust. Indeed, I’d argue that for the first time in the history of computers, the software actually matches the hardware capability. We now know the software isn’t going to be the linchpin for a New Economy that changes all the rules, but we also know it has a lot of value inside of companies for productivity purposes. Information technology essentially has gotten a windfall here.

Honeywell’s digitization program, which we launched on Sept. 12, takes advantage of the new Web software that’s been developed. Our goal is to digitize 90% of all our major processes—including engineering, manufacturing, and finance—by the year 2004. We now have 700 projects identified, 300 of them under way, which will bring a minimum of $150 million to our bottom line this year and $500 million in annual savings within three years. It’s the presidents of the individual units, from those that make jet propulsion engines to those that make spark plugs for cars, that have to deliver this—with my help.

It’s hard to organize hundreds of projects like this, so what we’ve done is stack them into three tiers. Tier One are projects that we do in zero to 30 days, that cost up to $100,000. Projects in Tier Two take 30 to 180 days and cost $100,000 to a few million dollars—they’re a little more integrated. Tier Three, as we call it, is where you get into actually transforming the things your company is doing. And that is what we would call inventing. We’ve told our people that they have to earn their way from Tier One to Tier Two to Tier Three.

Of course, we never do anything like this without metrics. For every Tier One and Tier Two project, we’re looking for an internal rate of return of 100% or greater. It’s got to pay back that quickly, because we don’t want our money sitting out there for long before we reinvest it in the next thing.

We still require metrics for Tier Three even though it’s more complex—and we still insist that the project be broken down into nine-month deliverables. We don’t let anything go beyond a year without paying back. And so every project is broken up into chunks so that we can show the business that we can deliver on these things as we go along.

People are saying, “No more big systems. Give me something I can use today and then throw away.” Having lived through building all these big, monolithic systems, you come to realize flexibility is what you need. The thinking is quite different now.

—written with Robert Hertzberg

Larry Kittelberger is Chief Information Officer and Senior Vice President/Administration of Honeywell International, the $24 billion company formed from a merger with Alliedsignal. He was previously CIO at Lucent and Tenneco.