Air New Zealand's reverse auctions started only a couple of months ago but the company has already seen the benefits of the new system. Built as part of its Grabaseat brand, the reverse auctions now have their own website and grant customers new flight deals every 15 minutes.
According to Scott Giles, Grabaseat - Future, it is win-win situation: the last-minute empty seats get sold and customers get a better deal than they usual get when purchasing a regular fare.
The system is simple: the first person to bid wins the flight so they can wait as long as they choose to. However, if they wait too long to actually proceed with their purchase, someone else might buy it first. The always-on service means that there are deals running 24/7, with prices dropping every 15 minutes.
The backend, however, is far more complex than it might appear. Air New Zealand had to build a whole new reverse auction bidding system and flight booking engine, as the other one could not cope with dynamic prices.
"Normally, an airfare has to be filed in the system with a price and that doesn't work with dynamic prices so we had to start from scratch. The new system grabs a flight and the price on the flight is the start price for the auction," explains Giles.
Based on fish and flower markets
The idea of having reverse auctions, based on the Dutch Auction system used in European flower markets and Asian fish markets, was "floating around for a couple of years", according to Giles. The new system took about nine months to build and was soft-launched on 1 March.
The development process had a few challenges that the team had to work through. "Firstly, the correct conversion of invented prices (the bid price) to maintain the IATA fare standards," says Giles, who adds that the reverse auctions fully support IATA fare rules, rounding down for the customer where required.
The other technical challenge was to do with updating users who are watching the auctions. "HTTP is not a push technology, and we cannot afford the load requirements of constant polling of the server. So the compromise has been to create some custom caching of Ajax responses for specific controlled periods."
Automatic fulfilment was another great challenge, says Giles. "That is, being able to make reservations without human intervention, and for the booking to remain compatible with existing applications for retrieval and management of said reservation."
Highly complicated project
Sign up for CIO Asia eNewsletters.