Similar problems arise when IT tries to satisfy business needs too quickly. "Sometimes these things were built as 'Let's just get something up and see how it works,'" Curtis says. "Things that were designed as a demo suddenly have to grow. Or even if something was designed appropriately for what they thought would be the use, people kept adding new requirements and features until it became a kludge."
Perhaps worst of all is the tendency to customize licensed software in an effort to fulfill business requirements -- whether or not those requirements have any real bearing on the organization's goals or success. "We talk about business capability -- the list of things a business needs to do to be successful and achieve its goals," says Nigel Fenwick, an analyst at Forrester Research. "Out of 30 high-level capabilities, maybe two or three are differentiators." When senior executives understand this well, he says, they encourage IT to focus on those key areas and seek standardized, easy-to-maintain solutions for everything else.
Unfortunately, such understanding is rare. "It's hard to get the CEO to stand up and say, 'This is the way we're going to do it,'" Fenwick says. But if the CEO doesn't do that, he adds, "every little department will want to customize the technology to make their part of the business run more efficiently -- and so they should." After all, each department is being judged on its own efficiency, and anything that can make it run better is a good thing -- from the point of view of the department's managers. But the approach leads to systems that are difficult and costly to maintain.
"Over the past 10 to 20 years, we've plowed millions of dollars into software customization to support generic capabilities," Fenwick says. "It has made IT more complex, made interfaces more difficult, reduced IT's agility and added cost."
There's one last reason it can be difficult to contain keep-the-lights-on costs: You may become a victim of your own success. "We've determined that it'll be pretty tough to get to 50-50," says Peter Forte, CIO at Analog Devices, a semiconductor maker headquartered in Norwood, Mass., with annual revenue of $2.6 billion. "The reason is, the more successful you are on the right-hand side, that drives more activity to keeping the lights on. Every new system we deploy is a system that needs to be maintained."
Here's a look at strategies that can help CIOs who want to spend less on keeping the lights on and more on innovations that will help the company reach its goals.
If you haven't gotten around to virtualizing servers, you may find that doing so is an effective way to cut keep-the-lights-on costs. Forte discovered that when a normal cyclical low in the semiconductor industry coincided with the worldwide economic downturn of 2009. "We lost 30 percent of our revenue almost overnight," he says. As a result, IT had to quickly cut 30 percent of its costs, leading to significant layoffs.
At the time, Analog Devices was about 45 years old, with the legacy infrastructure to prove it. "The first thing we did was calculate what percentage of our investment would be needed to keep the lights on," Forte says. "It was in the low 80s." For a technology company whose success depended on its ability to rapidly bring new products to market in large numbers, that was not acceptable. So IT launched a three-year effort to shift that balance. Today, Forte says, Analog Devices spends 62 percent of its IT budget on keeping the lights on and 38 percent on growing the business. That's not 50-50, but it's a meaningful improvement.