Machine learning is becoming a crucial part of many businesses in our world today. There are more and more examples popping up for how it can be used to help companies in many different industries grow. Netflix is using machine learning to continually train a robust recommendation engine that provides highly targeted content to users. Their machine learning algorithms are helping to produce around $1 billion dollars every year from customer retention.

Before Netflix started to dominate the video streaming space their primary service was shipping DVDs to customers. With a small monthly membership fee you could pick from a wide variety of DVDs and have them shipped to you, once the DVD was returned, the next one on your list would be sent. You could rate the DVDs you had watched and Netflix would suggest other films with similar qualities you may be interested in.

In 2007 Netflix was quick to jump aboard with global streaming. This opened up a vast collection of information including real time data on what members are watching, when they are watching it, where on the menu the video was found and the popularity of the videos in the catalog. This shift to global streaming also greatly shifted how Netflix customers interact with the service. Before, the recommendations were targeted for people selecting a DVD to watch in a few days, weeks, or even months. This selection process customers took was more in depth because the whole exchange would take more than a day so people were constantly thinking about what they might want to see in a couple days or weeks.

Once streaming was brought into the mix Netflix realized the algorithms they were using needed to include the real time data they were receiving. Users now have the ability to sample videos before selecting one to watch, they can watch multiple videos in one sitting and all the information about each session is being reported back. If a user stopped watching a movie halfway through or they completed parts 1 & 2 in a 3 part series, this is all information that can be used to help determine what shows up for their recommendations.

See more examples of Machine Learning in our Everyday Encounters blog series >>