Continued from Part One
If you want to launch a Web services or service-oriented architecture (SOA) project these days, it's going to have to pay its own way. Long gone are the days when technology projects
In this second of a two-part column on ROI, I'll look at what kind of return you might expect on your technology investment, and how to measure ROI after deployment.
Many happy returns?
There's, of course, no real way of knowing the ROI you'll get on a project before actual deployment. That can put you in a Catch-22 situation -- if you can't prove payback, you might not be able to get the project funded, but you, of course, can't measure payback until months or years after deployment. So what kind of ROI can you expect?
Perhaps the best way is to find out what kind of ROI some other firms have gotten from Web services and SOA projects.
Michael Liebow, vice president of Web services and SOA for IBM Global Services, said many of his customers "have seen tremendous ROIs in very short order."
Key, he said, is choosing the project carefully. It's best to focus on a specific business process, and solve that problem. If you do, he said, you can potentially expect very significant ROI.
Although Liebow couldn't name the specific companies involved, he did provide details of two their Web services/SOA projects:
- A large credit card company launched a Web services/SOA project that allowed it to truncate its billing cycle from more than 30 days to under 30 days. The payback, he said, was several hundred million dollars in a three-month period.
- A telecommunications company was bedeviled by a decentralized billing system that it couldn't rely on properly tracking usage and billing the correct amounts. It launched an SOA project to link its disparate systems into a single, unified system. The results, he said, were $500 million saved in the first year.
Even IBM, he said, has benefited from Web services/SOA projects, and they've carefully tracked ROI, and found a very big payback. The company spent $50,000 to add an easy-to-use Web services front end to a configuration engine that provides links between IBM and its business partners for transactions. The engine checks orders at the time of the transaction, and then validates the transaction. In the first three months, the project dropped the transaction error rate from 3% to 1%, and usage of it went from 5% of orders to 90% of orders because it was so simple to use. The result was multiple millions of dollars a year in savings, on a $50,000 investment.
These numbers, Liebow admits, are certainly at the top end of the ROI spectrum. And companies need to be careful in choosing their projects. He emphasizes that choosing a specific business process and then improving it will have the most immediate, clearest and largest ROI. Taking that into account, he said companies can typically expect to get a 20% to 30% cost reduction from a Web services/ROI project, and a payback period of a year or less.
"You can implement the project in months, and you can realize the business benefit and payback in a year or less," he said. "So it's really a self-funding model."
Start small, end big
Liebow said because it's a self-funding model, it should be easy to prove ROI ahead of time. But in the real world, that's often easier said than done. Ron Schmelzer, senior analyst with ZapThink, suggested that companies start with a very small project, measuring the ROI on that small project, and then expanding it beyond that.
"The smaller you start, the easier it is to measure ROI," he maintained. "Our suggestion is that you start with the smallest, finest-grade services, and measure the ROI on that. For example, it can be something as small and simple as getting information from a database, because even something that simple should have an ROI associated with it."
Make sure that the service is something simple and measurable. After launching the project, monitor and measure the results and calculate the ROI. He also suggested predicting the ROI ahead of time, for example, for a small portal, and then comparing the predicted ROI with the real ROI. That will help you learn how to predict ROI ahead of time better in the future.
Based on the initial project, he suggested doing the same thing with multiple small projects -- that is, if there was an ROI with the initial one. If there was no ROI, not much would be lost because the project was small and discrete. But if there was an ROI, going to multiple projects, and then larger projects, will yield even larger benefits.
The anti-ROI choice
IBM's Liebow is a proponent of measuring ROI, but he warned that if companies focus only on ROI ahead of time, they may lose big benefits. The time involved with doing an ROI project, then moving it through the chain of command, can be time-consuming. Many times, he said, if you just go ahead with the project, you'll save far more than if you have to delay the project for months.
So what to do if you're stuck in a situation where you know that the Web service or SOA will have a big payoff, but you don't want to have to spend the time moving through the bureaucracy?
"Any budget has some money in it devoted solely to innovation," he said. "Even if 80 to 85 percent of your budget is locked, most places still devote from 10% of the budget for innovation. In the confines of an innovation budget, you typically don't have to ask for management approval for spending. So use that money. The ROI will follow."
About the Author
Preston Gralla is an expert on Web services and is the author of more than
20 books, including How the Internet Works. He can be reached at firstname.lastname@example.org.
This was first published in April 2005