The Netflix Prize paid $1 million for a 10 percent gain

The Netflix Prize, launched on October 2, 2006, offered a grand prize of one million US dollars to the first team that could improve the predictive accuracy of Netflix’s Cinematch recommendation system by 10 percent, measured as root mean squared error on a held-out test set. Netflix’s own description of the contest, by James Bennett and Stan Lanning, documents the prize amount and the 10 percent target.