While I’m thinking about Netflix, a quick question: why does Netflix care so much about improving their recommendation system?
Netflix made headlines years ago when it announced its Netflix Prize, a $1,000,000 bounty to whoever could build an algorithm that would improve the accuracy of its recommendations by 10%. When this award was claimed in 2009 by BellKor’s Pragmatic Chaos, Netflix gave them the million. They then announced a new prize for whoever could improve the system further (which was later canceled.
What’s in it for them?
My understanding of the Netflix business model is low fixed costs (giant DVD warehouse) plus low variable costs (envelope stuffers, postage, replacing scratched DVDs) vs. hefty subscription fees = profit. Even if Netflix buys its DVDs at retail prices – which they don’t – they only need to rent out each DVD three times to make back its cost. The margins are in the low transaction costs and the quick turnaround time.
I don’t see how a more accurate recommendation system adds value to that.
Someone help me out here? What am I missing? Netflix knows more about the video rental market than me, obviously (or even Blockbuster), so I’m willing to bet they know something I don’t. But what is it?