U.S. airlines, which spent US$14.8 billion on 20 billion gallons of jet fuel last year, go about topping off their tanks much the same way you do: They buy fuel where it's least expensive and plan stopovers and fuel contracts where it makes the most economic sense.
But for the airlines, that's no seat-of-the-pants operation. Digesting the variables of route structure, time and passenger loads requires sophisticated IT operations. And what the airlines do routinely to keep their fuel costs below the average of 77 cents a gallon demonstrates how densely technology can be woven into complex business decisions.
At Delta Technology Inc., a subsidiary of Delta Air Lines Inc., chief architect Jeff Tyler is making it possible for business units to extract data from a variety of sources and subsequently recombine the data into meaningful information. Tyler applies business intelligence software to graph, sort and compare data that arrives from reservation agents, unit managers, flight crews and even marketing departments. The system Tyler oversees uses Informatica Corp.'s PowerCenter and PowerConnect to integrate data from 25 sources, including SAP AG's R/3, legacy databases and third-party booking sources such as travel agents and Web sites.
"We try to optimize where we put gas in the plane," says Tyler. "A 1 percent or 2 percent savings is a lot, and with gas prices differing from airport to airport, we want to match least expensive fuel to most efficient time to put gas in the plane."
This decision is made using information mapped via the Teradata warehouse running on top of Hewlett-Packard Co. Unix servers. Informatica provides interfaces directly with SAP at the application layer (rather than at the database layer) for speed and efficiency. The information can be modeled and used for comparisons to see, for example, where fuel prices have jumped the most during the year.
Tyler says Delta has four development squads assigned to the airline's business units: revenue, customer, airline operations and corporate management systems. He uses the same business intelligence approach for each.
"Sometimes the business unit knows what they want and the relationships they're trying to look at, while others request help to build a model," he says.
In marketing, Delta was able to match passenger revenue with its frequent-flier program. "We knew how many miles someone was flying but had no way of knowing how profitable a customer they were," says Tyler.
Now Delta links revenue data, ticketing information and the customer loyalty program. Maybe it can also tell passengers where to fill up their cars?