Five ways SOA pays... eventually

Show me the money, now -- or at least within six months.
Written by Joe McKendrick, Contributing Writer

Show me the money, now. Okay, you have six months. The pressure is on technology folks to start showing returns -- in the form of hard cash -- within six months for new technology ideas. Languish for a year or more, and it's considered a dud.

Where does that leave SOA, which is a highly cognitive, feel-good, transformative process that requires that C-level types be convinced that the current spaghetti-oriented architecture residue vendors have left in their enterprises will be somehow become the magic carpet ride to endless profits? How is this done in six months, let alone six years?  (Don't worry, the vendors will be harping about something else by then.)

IBM, vendor of vendors, says, 'don't worry, we'll show you the way. Here's something you can take to your CFO.' DestinationCRM recently extracted, from a recent IBM white paper, some of the quickest ways SOA has been known to deliver some returns.

It's not exactly clear how or if some of these endeavors can be up, running, and pumping money within six months (without the guidance of some high-priced IBM consultants, of course), but these five points at least give us something to ponder:

  1. Deliver cost savings through service reuse. This, more than anything else right now, is the ace in the hole for SOA -- a service only needs to be designed and built once, and made accessible across the enterprise for reuse in other systems. The cost alone in development staff time means significant savings. IBM cited a fairly weak example of an Asia-Pacific government agency "that had deployed an automated verification service for documents like passports and licenses was able to share it with four other agencies in that government. The process took five weeks and saved the cost of four separate systems."
  2. Automate processes across multiple users and companies. A financial services organization estimates that it has saved $200 million a year by implementing a Web-based, automatic customer dispute-resolution system. Frankly, a $200 million a year savings from one system sounds wildly blown out of proportion. But there are some solid documented calculations behind it -- "the organization's cost per transaction was reduced by 90 percent, and resolution periods fell from 20 days to three hours."
  3. Make existing resources more efficient. This makes perfect sense. Keep your powerful legacy systems -- such as mainframe CICS -- and put a service-oriented front end on it. But the savings will become evident only if you were planning to scrap and migrate that legacy system. (And within the six-month time frame.)
  4. Build a single source of information. I'm not sure if they're talking about SOA here or some kind of data warehouse/EII effort. SOA and enterprise information management are headed for an inevitable convergence, but that's not going to help us within the next six months. IBM reports that a retailer bumped its revenues by $20 million a year by implementing a delivery notification service. The service "linked the systems in its fulfillment chain so that they communicate continuously and reliably, automatically informing customers of status changes."
  5. Sell through partner Web sites to increase sales. E-commerce redux. I'm surprised there isn't more chatter about the e-business/e-commerce angle to SOA. "SOA allows resellers to give customers direct access to the original vendor's sales system, easing the process for all." It increases sales and strengthens relationships. Makes sense.
This also brings into play that whole start big/start small argument that I have had going in recent weeks at this blogsite. If anything makes the case to start SOA "small," this whole six-month ROI scenario is it, agreed?
Editorial standards