Harnessing big data is crucial to the day-to-day operations of Netflix, the online video streaming service which sees users watch more than a billion hours of on-demand films and television shows a month.
That's what Justin Ward, Netflix's manager for data science and engineering, told the audience during his speaker session, titled "Netflix: Big Data-Driven Decision Making in the Cloud", given at MicroStrategy World Forum 2014 in Barcelona.
"We have people enjoying over a billion hours of content a month now, which is pretty cool, and it's safe to say we have some big data coming into our warehouse," he said, before describing how no decisions at Netflix are made without analysing big data algorithms.
"We use a lot of data in our decision making day to day, it's crucial to our business. We don't push out any new feature live to our site, we don't change any algorithm, we don't change any email without testing it first."
That process, Ward explained, involves A/B Testing, randomised experiments with two variants, the results of which are collected and analysed in order to determine whether the result brings benefits to Netflix and its customers through its "great recommendation algorithms".
"We go through this vigorous A/B experimentation because we don't know if our ideas are very good when we have them or not," he said, describing how Netflix ensures every piece of data is measured and analysed.
"So we take a small sample group that's representative of the population we're targeting and make sure we have every piece of data about them that is relevant to the test so that we can move forward and measure it against the same metrics across the whole company."
However, Netflix doesn't just use big data algorithms to assess its user-facing interface; they're also used to decide which television shows and films the on-demand video-streaming company should purchase for its subscribers.
"There's a lot more to our experimentation to what is customer facing," said Ward. "We use it for things like picking which content we're going to buy for you. At the price point we give, we can't afford to have all the movies and TV shows, so instead we have to be smart about which content we do purchase for you. We want to pick the right content, what you'll want to watch."
[Please turn to page 2]
This paper seeks to provide education and technical insight to beacons, in addition to providing insight to Apple's iBeacon specification
Focus on cost efficiency, simplicity, performance, scalability and future-readiness when architecting your data protection strategy