Expert Voices: Frontier Airlines CIO Bob Rapp on High-Yield Business IntelligenceBy Debra D'Agostino | Posted 07-27-2006
Soaring fuel prices have affected every airline, large and small, forcing them to turn to IT to make processes more efficient and trim costs wherever possible. But IT can be used to improve profits, too. At low-cost carrier Frontier Airlines, for example, technology is working around the clock to assist the carrier's small team of 10 revenue management analysts, who oversee 280 daily flights, by delivering real-time business intelligence that helps them adjust fare prices and keep planes full. Senior Reporter Debra D'Agostino spoke recently with Frontier Airlines CIO Bob Rapp to discuss yield management, parallel processing, and the value of technology at Frontier. An edited transcript of his comments follows.
CIO Insight: The major U.S. carriers are obviously straining under the weight of skyrocketing fuel costs. How are the low-fare carriers holding up?
Rapp: Well, it's been a tough time for all carriers. And Frontier is still an emerging player in this industryeven a carrier like JetBlue is twice our size. But we still have the same business problems that any carrier has, and we're playing in the same field without the same level of financial resources that an airline like Southwest or Delta has.
So how does a small carrier like Frontier compete?
By doing very detailed yield managementbasically making sure that every seat on every plane is full, and bought at the highest price possible. You do that by analyzing the airline's historical data to determine how best to distribute fare prices and maximize value. Every airline does this; the concept was first started at American Airlines. But at Frontier, the process was very lengthy. The data took a long time to analyze, and as a result we weren't able to react as quickly as we wanted to.
Why is speed such an issue?
In the airline business, there is a very rapid turnover of inventory. And anything that's not used is lost. You can't sell an empty seat at a discount after the plane closes its doors. It's gone. So the ability to create a near-real-time booking scenario that puts passengers in all the seats and gets them to pay the highest fare possible is obviously the goal. You also have to factor in the competition and pay very close attention to sales that other carriers are having and anticipate how to react. To make those decisions, our revenue managers need a tremendous amount of historic data at their fingertips, quickly.
So we embarked this year on a project to enable that. By partnering with business intelligence vendor Greenplum and Daxpy LLC, a business analytics consulting firm, we created a new yield management system that uses parallel processing to speed up the time it takes to analyze all that data.
How does it work?
The Greenplum product uses parallel processing to analyze terabytes of data at a time. That information is then fed into an application called PROS (Passenger Revenue Optimization System), which isolates trends as it relates to a particular market or time of day. So now, instead of a query taking several hours, the query takes only a minute to come back. For example, if I am a revenue management analyst looking at the Denver-to-Dallas market, I would want to know what percentage of seats should be offered at full fare compared with those that can be discounted. So I look at several factors. The first is which events are going on in Dallas. Maybe Denver and Dallas are playing a football game, for example. We can stimulate traffic for that. But before the Greenplum rollout, that would have taken much longer. Now, instead of reviewing a flight only once or twice before its departure, we can look at it much more often, be closer to the true status of that flight, and take quick action as a result.
And have revenues improved as a result? What's the ROI?
We've had the system in place for about a month and are already seeing results. The Greenplum product is open source, so it's low-cost. To do it any other way would have been financially beyond our means. And our yields are improving by roughly 8 percent per month. Some of that can be attributed to outside market conditions, but it's also due to overall better yield management.