Get Air Fare Deals by Forecasting Them

Kim Peterson of The Seattle Times has found a new company which is developing software to help you get a better price on your next flight ...

There are more than enough travel-related sites on the Web, but Farecast hopes to stand out from the rest by actually predicting a rise or fall in airfares.

The Seattle startup, currently testing its technology in a private trial, is combining data mining with a vast library of historical airfare prices.

Put those two together, the company said, and you can actually start to make sense of the seemingly wild fluctuations in plane fares from day to day, even hour to hour.

Farecast expects to open its trial to the public sometime in the summer of 2006. The concept has been solid enough to attract venture funding, $8.5 million in all, and executives are confident it will appeal to consumers flummoxed by airfare pricing.

"There is a need for a sort of Consumer Reports to emerge," said Hugh Crean, president and chief executive. "There's a need for a Kelley Blue Book. What is a fair price?"

For now, Farecast is focusing on plane trips originating from Seattle and Boston, with plans to expand nationally by the end of the year.

Here's how it works: A user picks a date and destination for a plane trip. Farecast gives a list of current airline prices and a prediction about whether those prices will trend up or down over the next week.

When queried Wednesday about a round-trip fare from Seattle to Denver in the first week of June, Farecast said it was more than 80 percent confident the lowest fares would rise by $45.

But for a round trip to Columbus, Ohio, over that same period, Farecast said it was 58 percent sure that fares would drop by $32.

The idea strikes a chord with people who hear about it, Crean said. And no wonder. Consumers are savvy about how to buy tickets online, but are often at a loss when it comes to figuring out when to buy.

Oren Etzioni, a professor of computer science and engineering at the University of Washington, came up with the concept for Farecast years ago while flying from Seattle to Los Angeles to attend his brother's wedding.

During the flight, he started asking passengers how much they paid for airfare. He was dismayed to find some people who had bought a ticket after he did got a better deal.

"It was very frustrating," he said. "I was like, 'Wait a minute. I want to do something about this.'"

Etzioni originally named the project Hamlet, with the unofficial slogan, "To buy or not to buy." The technology belongs to the UW, but the university has licensed it exclusively to Farecast, and Etzioni is on the company's board of directors.

Farecast started in April 2003 and now has 22 full-time employees. The company gets its data from ITA Software, a Massachusetts company that specializes in collecting airfare information, and says it now has 50 billion historical price quotes from routes to study.

Farecast sounds like a company that would make airline executives spitting mad, but Crean said he's received a positive response. That's mainly because Farecast doesn't sell tickets itself; instead, it links directly to airline Web sites.

"It's certainly very strategic for an airline to have a customer go straight to them," Crean said.

Some airlines pay Farecast a fee for the referral, though Crean said he hopes most of the company's revenue will come from ad sales in the future.

Using technology and data mining to make predictions is a hot business, said Matt McIlwain, a managing director at Seattle-based Madrona Venture Group, where Etzioni is a venture partner. Madrona has invested in Farecast. Kirkland-based Inrix, for example, aims to predict traffic patterns so that it can tell drivers about trouble spots.

But Farecast's technology is very difficult, if not impossible, to replicate, and no other company is in the same business, McIlwain said.

"We've all had a Farecast moment where we're sitting there and we're looking at a fare that costs 400 bucks," he said. "Farecast is trying to help the consumer answer a fundamentally different question, which is when to buy, versus the question of what to buy."