Netflix profiles: The big data logic behind personalization

A couple of months ago I rushed to an office where I do some regular work, opened my laptop and was greeted with the loud, happy-go-lucky singing of Dora the Explorer. As I scrambled to turn it off, a few of the women in the cubicles near me turned around with bemused smiles. They are used to this. “What are we watching today?” one of them asked.

Damn you, Netflix! It’s one thing when the kids forget to turn it off before I take the laptop with me. It’s another when I’m finally ready for some downtime, visit the site and see “Top 10 for Shane” recommendations that are dominated by Max and Ruby, Go Diego Go, Kipper and Caillou. Obviously I’m not alone, given that Netflix this week announced the introduction of personalized profiles that would be available to multiple family members using the streaming video service in their household. Don’t mistake this as some kind of friendly customer-service play, however. Netflix, like everyone else, is waking up to the possibilities around big data.

Though it has certainly blazed a pioneering path in offering movie and television content on nearly any device, Netflix’s trajectory followed that of companies operating in almost every other form of entertainment content.

Think about music, which began as something enjoyed primarily in community at concerts before vinyl records, the Sony Walkman and finally Apple's iTunes made the consumption of songs more of an individualized experience. The same thing has happened with newspapers that might have gotten passed around the breakfast table, which are being increasingly replaced by Twitter, FlipBoard and other tools to aggregate information according to personal preferences. My kids still watch Netflix together at the moment, but it’s not hard to imagine the day when they’ll demand profiles that reflect their unique interests there, too.

Netflix was already giving some hints about its profile strategy a few weeks ago, when the company was among the exhibitors at a big data conference in San Jose, Calif. In an online video interview from the show floor, Netflix’s manager of the data science platform architecture Jeff Magnusson was asked point-blank about the possibility of more personal sign-in methods.

“There’s a lot more interaction happening on the service now,” he said. “There’s a lot more analysis we can do, and we can start digging into some of those problems (around preferences) and hopefully even providing even better solutions and even better recommendations for content.”

Netflix’s approach reminds me of a recent conversation I had with Wayne Ingram, managing director of technology for consulting firm Accenture Canada, who talked about the notion of building “relationships at scale” using more personalized customer profiling.

“The real value is about asking the right questions,” he said. “I believe that the data sources, the volume, the plumbing, we’ll figure that out. I think the real value is in trying to understand what are those questions to ask. You need deep skills in business, in strategy, and in the operations of the organization to understand what insights can I pull to derive that extra value above and beyond?”

Companies like Netflix also need to think what defines “value” in this case. Does it mean customers use the service more often or for longer periods, or that they somehow monetize them through advertising or premium content services? Is it about cementing loyalty amid increasing competition? Should Netflix focus on better recommendations (the way to help customers passively make decisions) or better search (the more active way)? Personal profiles are a first major step, but there will be and should be much more. Stay tuned.