Second is the value expected from the initiative. "Will I increase revenue or do I need it to protect revenue of specific clients because other competitors are providing the same service?" Balliet asks. The value component also takes into account any cost reductions that will result from a project through shutting down other applications or laying off staff. In addition, the value proposition looks at whether the technology will help to avoid cost increases if the technology must scale to keep up with increasing volumes or new regulatory requirements.

The third step is deciding the project's technological feasibility and whether it meets the company's technology standards. Business-unit leaders consult with executives who manage the IT architecture and overall technology to figure out if the project is going to function as it should.

The fourth area explores risks, including implementation and other risks related to doing the project, as well as the risk of not doing it.

Once a project gets through this process, the business-unit leader must report back about the initiative at least every quarter. This includes follow-up on whether the project is on budget and on scope and if it still has the proper support from the people who will use it. The bottom line: The business-unit leader needs to affirm that the project should continue and determine whether the anticipated value is the same as originally proposed.

The reasoning behind such frequent project health checkups? "The markets change so rapidly in financial services," Balliet says. For example, a project that had the goal of steadily increasing the capacity to process trades of Nasdaq-listed shares could have been a very viable project in 2000 but would be far less important today, he says. While most of Merrill Lynch's projects move faster than that -- finishing in 12 to 18 months -- they still can lose impact before the completion date.

Some of the larger business units must evaluate their technology portfolio every month to make sure they're keeping up with changing markets and technology. "They take a look at the portfolio throughout the year and ask themselves whether they're putting the money in the right places based upon what the marketplace indicates at that point in time," Balliet says. As an example of the importance of rebalancing, Balliet says that in 2000 and 2001, most technology initiatives were related to equities, with little development on the debt business because the returns didn't warrant the investment. This reasoning has now flip-flopped.

The governance model requires looking at many factors before making ROI decisions, Balliet says. "We have some target financial hurdle rates of return, but we approve projects that don't meet those rates because there are a lot of other nonfinancial measures that clearly indicate we should go ahead with them," he says. "And there are times when projects meet the financial hurdle rates, but we opt not to go with them due to some other nonfinancial reason."

Merrill Lynch developed this governance model for investments around applications rolled out to about 8,000 users in its Global Technology and Services Group, using the Web-based application Business Engine Network from Business Engine as the financial tool for technology investments. Merrill Lynch is replacing its project-management systems with Microsoft Project, which integrates with Business Engine Network.

Merrill Lynch, which buys Business Engine Network as a hosted application, uses it to track the full spectrum of what IT companies do. John O'Neil, chairman and CEO of Business Engine, says IT projects are usually divided between the "lights on" everyday technology maintenance and operation and projects such as entering a new market or building a new system.

An ROI process is embedded in Business Engine Network, and companies can set up their own templates for determining the standards a project must meet. "In addition to budgeting what the project is going to cost, users also budget what the proposed payback will be, so that two months into the project they can check back to see where they are and where the ROI is," O'Neil says.