Nov 04, 2003There's a great deal of talk about creating real-time companies, and a great deal of concern about how to cope with the expected flood of RFID data. Mike Gilpin, a VP and research director at Forrester Research, an IT research firm, says it’s important to use real-time data only when the business process you are improving requires it.
"You have to focus on process to determine what data you need and where you need it," says Gilpin, speaking at Forrester's Executive Strategy Forum in Boston last week. "If you bring data in the wrong way, you waste a lot of money, and you don't get business value."
The event focused on what Forrester calls the “extended Internet." The term encompasses RFID tags, smart appliances, cell phones and handheld computers, as well as other devices that will all be connected to the Internet before too long. The conference delved into strategies for taking advantage of the extended Internet.
Whether you need real-time or conventional data, Gilpin says, depends on the process and what you are trying to achieve. To illustrate the point, he gave examples of several companies that use real-time information from sensors on trucks, GPS devices and so on, and several examples of companies that use data in batch mode to analyze processes and continuously fine-tune them.
For instance, companies might want to use real-time data to forecast demand, but creating custom models in which to plug the data is expensive. "You can't apply predictive analytics everywhere in your business," he says. "You have to look at whether the benefits are significant enough to pay you back. The business process defines how urgent the need for data is."
Gilpin suggests that companies build two business intelligence software stacks—one for real-time data and one for batch data. At the bottom of each stack would be a data integration layer for capturing information from RFID readers, sensors and other devices in the field. On top of that would be a data staging layer—a data warehouse for batch data and a cache for real-time data.
Above that would be an analytics software layer, then an information delivery layer—desktop applications or a Web portal—that displays both the batch and real-time information. And finally he envisions a business process management layer that creates role-based portals and gives managers all the data they need to execute specific tasks.
The important thing is to focus on critical business processes, instead of on all business processes, says Gilpin. That reduces cost and provides the most bang for the buck. He also suggests that companies define key roles and decision points in their organizations to avoid gathering a lot of information that doesn’t create any real value for the company.
Nearly 95 percent of attendees at RFID Journal University in New York said they would recommend the program to a colleague. There’s still time to sign up for the final RFID Journal University in Chicago (Nov. 17). Register now by visiting RFID Journal University, or call 510-832-1501.
