Decentralization inevitably drove costs up as economies of scale were lost, and as little pockets of generalists attempted to do work formerly done by teams of specialists. And a lot of little outsourcing deals drove costs up too. Furthermore, corporate synergies were lost, the pace of innovation slowed, and quality suffered. Again, the net effect of the cap was to increase, not decrease, IT costs.

The lesson here is straightforward: You can't control spending by controlling supply. The only effective way to contain costs is by limiting clients' spending power.

Bureaucratic Governance Processes

How can executives constrain the amount that business units spend on IT?

Some people advocate a "governance" process that imposes bureaucratic hurdles in an IT project-approval process - ROI thresholds, committee and CIO approvals, and piles of paperwork. Their theory is that the committee or the CIO is smarter about investing than clients, and they'll stop business leaders who try to buy something that's a poor investment.

In their cynical moments, advocates of bureaucracy might admit that simply making it difficult to do business with the corporate IT department will reduce demand, since people won't make the effort to run the gauntlet for any but the really important projects. While these hurdles may, in fact, dampen demand, they certainly don't engender optimal resource governance. Some people make up ROIs. Others may not bother with projects that are really good investments. People break up their requirements into small "enhancements" to circumvent the committee. No doubt you know all the games people play.

Furthermore, bureaucratic hurdles chase business to the competition (decentralization and outsourcing), where clients are treated as customers (not like children in a candy store who need to be controlled) and where it's easy to get what they need. Like caps, bureaucratic project-approval hurdles drive costs up.

The alternative to bureaucracy is market economics. In a market economy, your spending isn't limited by the size of the store. It's limited by your chequebook. Similarly, within corporations, cost containment is best achieved by limiting the size of the business units' chequebooks, not by limiting the size of an internal service provider like IT. In other words, the corporation must constrain demand, not supply. Then, the business units can be free to spend whatever they have on whatever they see fit. They'll naturally use their limited funds to buy what's most important to them and skip the rest, without time-consuming hurdles. Resources will automatically flow to the best investments. I call this a market-based internal economy.

There are three things required to make this marketplace work: business units that care about their bottom lines, a limit to their chequebooks, and the information they need to make good investment decisions.

As to the first requirement, most business unit leaders already care about their bottom line. If this concern doesn't ripple down through their ranks, then they have a management problem - one that will affect spending on everything, not just IT. There's nothing IT can do about this.

So presuming that clients want to spend their resources wisely, the two remaining challenges are to limit their chequebooks for IT, and then to provide the information they need to make good spending decisions.