Download the authoritative guide: Cloud Computing 2018: Using the Cloud to Transform Your BusinessFarecast, the slick new site that predicts whether air fares between particular cities will go up or down, has expanded its service to cover more than 55 U.S. airports.
When I first wrote about Farecast on June 27, the service made predictions only on flights originating from two airports: Boston and Seattle.
At that time, the startup -- which has attracted $8.5 million in venture capital -- launched a public beta that benefited from major media coverage in the New York Times, Time Magazine, and elsewhere. But with only two airports supported by its site, I wondered whether Farecast was making a mistake by seeking publicity so early. Wasn't it alienating potential customers, who would see that their local airports weren't included and leave the site, never to return?
The answer is crucial to any company that must decide when it should publicize a new and rapidly improving product or service.
In an interview, Farecast's vice president of marketing and product, Mike Fridgen, explained that the original wave of publicity for the public beta in June didn't hurt the site when it started covering 55+ airports last week. (This expansion had originally been scheduled for the end of 2006, but Farecast apparently accelerated the rollout because consumer feedback was so positive.)
Last week's rollout day more than doubled the site's unique-visitor traffic, Fridgen says, compared to the day in June when the more-limited, two-airport service opened to the public. That's true, he adds, even though the mainstream media coverage of Farecast's expansion was much lighter than the coverage of the original beta. The New York Times article about the enhancement, for example, was a small story on an inside page, whereas its original article had been a large illustrated piece on a prominent page.
Wisely, Farecast turned its limited number of airports into an advantage. A link on the site urged visitors to "Add My City." Fridgen says more than 50,000 visitors responded with airports that they wanted Farecast to support. A somewhat smaller number of visitors, which Fridgen wouldn't disclose, gave Farecast their e-mail addresses to get a notice when their favorite airports were added. Each of these people received a message the day Farecast unveiled its new, 55-airport selection and newspapers were carrying stories about the improvement.
Figures compiled by Alexa, a Web ranking service, confirm that Farecast's site received three spikes of traffic recently: one for an early pre-launch in May, one when the beta was first opened to the public in June, and a final spike last week when the expansion was reported. Alexa's rankings don't show that the latest launch generated twice the traffic as the one June, but Fridgen says his company's servers measure a significant amount of traffic that Alexa doesn't.
Predicting Air Fares the Farecast Way
It isn't as easy as snapping your fingers to add predictions for what the airlines call "city-pairs." Farecast is able to predict whether air fares are soon likely to go up or down only because the company's executives, many of whom have corner-suite airline backgrounds, have compiled more than 90 billion historical fares and analyzed them with sophisticated algorithms.
This doesn't mean that Farecast is psychic. It can only predict fares within a fairly limited range of dates. That's because the airlines, which years ago developed computerized models for pricing, determine ticket prices within three rough time periods:
• About 11 months before your travel date is when airlines start offering tickets. For example, on Sept. 1, 2006, you can buy tickets for travel on Aug. 1, 2007. Fridgen says people who buy tickets this early are determined to get to a particular spot on a particular date, so the airlines charge these customers "list prices."
• Around 90 days before departure, the airlines start to compare the number of tickets that have been sold to the number their computer models projected. If the numbers are too low, the airlines will drop some prices until the number of filled seats is back on target.
• Prices start to go up after the 21-day advance purchase deadline, and go up yet again at the 14-day and 7-day advance purchase deadlines. Within 6 days of a flight, the airlines know that any prospective travelers have little flexibility in their travel plans. Higher prices in this down-to-the-wire period are usually the result.
As a result of these pricing windows, Farecast makes predictions on future fares only if your trip is less than 90 days in the future. You also must be planning a stay of 2 to 8 nights. You can get price quotes for a trip that doesn't meet those criteria, but Farecast won't try to predict whether the prices are likely to go up or down.
A Business Model That Relies on Affliations
Since Farecast doesn't charge consumers for its service, you might wonder how the company makes money. The answer is that some airlines pay commissions when people who searched for air fares at Farecast wind up buying a ticket. Farecast sells no tickets itself, linking visitors directly to airline Web sites. (When a special fare or multi-airline itinerary is involved, Farecast sometimes links to Orbitz, which also pays commissions.)
Farecast recently started carrying travel-related advertising from the Yahoo network. Fridgen indicates that revenue from these ads is already significant, although he wouldn't indicate whether the ad income is larger at this early date than the commissions on airline tickets.
What he will say is that visitor traffic from the Boston and Seattle metropolitan areas made up less than 20 percent of the site's traffic, even when Farecast supported only the Boston and Seattle airports. There seems to be a genuine desire on the part of consumers to understand the airlines' computer pricing models -- and to use Web technology to fight back by finding prices when they're at their lowest.
Now that 55-some airports are supported, Farecast should grow its customer base even faster. "Seattle and Boston represent only about 4 percent of originations" in the United States, Fridgen says. "The airports we have now represent 74 percent of domestic travel." The site currently supports only trips within the U.S., but international flights will be added as soon as possible, he promises. Farecast also plans to someday support one-way ticketing and multi-city travel, not just round-trip fares.
Farecast is only one of several second-generation travel search engines that do more than an airline site can offer or the major portals, such as Expedia, can provide. Katie Fehrenbacker, a blogger for the GigaOM tech news site, says this includes such meta-search services as Sidestep, Mobissimo, and Kayak.
But for now, Farecast is the most interesting. It's the only site that's mastered enough mathematics to predict whether you should buy a ticket today or wait a few days because the price is likely to drop. For more information, visit Farecast.com.