All it took was 20 minutes for a team of 30 people to lose $1 million. That’s how much Netflix, the online movie rental company, offered to the winning team who could come up with a way to improve the site’s movie recommendation system. Out of over 40,000 other teams from over 150 different countries, BellKor’s Pragmatic Chaos team of seven engineers sent in its results about 20 minutes ahead of their one remaining rival team, Opera Solutions, who had come up with the exact same prototype. Although the two teams were a dead tie mathematically, those 20 minutes made all the difference.
The contest took nearly three years to finish, beginning in October of 2006. The new algorithm developed by BellKor will improve the movie selection prediction of heavy users by at least 10 percent. Netflix considered this competition so successful that they’ve already announced a new one, this time to improve the prediction mechanism of users who seldom rent movies. The outcome of both of these contests, however, is not and will not be limited to just usage on the Netflix site. The winning model, which deals with extremely large data sets and large-scale modeling, can be applied mathematically to other fields of science and marketing.
In a contest like this, everybody wins. Obviously, the winning team gets a bunch of money. Netflix also wins. Instead of hiring a team of qualified mathematical engineers for an unlimited amount of time, they saved money, time, effort, received publicity, and have already reported higher customer retention rates. Even the losing team (which is a consulting company specializing in analytical data) wins. They applied the techniques and research conducted for Netflix to their own company, and have reported an internal payoff of millions of dollars. As mentioned before, other companies are also are sure to benefit from the winning model and techniques used in its creation when applied to their own business practices. Netflix’s contest has proved to be a clever and innovative marketing tool, benefiting many.