Written by Aryan Sethi
With over 139 million paid subscribers (total viewer pool - 300 million) across 190 countries, 15400 titles across its regional libraries, and 112 Emmy Award Nominations in 2018 — Netflix is the world’s leading Internet television network and the most-valued largest streaming service in the world. The amazing digital success story of Netflix is incomplete without the mention of its recommendation systems that focus on personalization.
What should I watch after such a hectic day? This is the first question that pops up in your mind when you sit down and have no idea about what to watch. Fortunately, we do not have to worry much about this as the recommendations given to us by Netflix are according to our likings and usually helpful too. But have you ever wondered how it recommends the content? Do they spy on us? Do they secretly know us? Or do they just guess a movie/show and pray to God that we like it? NO. It is through the marriage of Machine Learning and Data Science!
Netflix uses machine learning and algorithms to help break viewers’ preconceived notions and find shows that they might not have initially chosen. To do this, it looks at nuanced threads within the content, rather than relying on broad genres to make its predictions. This explains how, for example, one in eight people who watch one of Netflix's Marvel shows are completely new to comic book-based stuff on Netflix.
THE 90 SECOND RULE
Researchers at Netflix have found that if a user is not able to find appropriate content within the first 90 seconds after opening Netflix, he or she is likely to switch to another platform. It, therefore, utilizes its recommending engine to compel its users to start playing content within those 90 seconds.
“Our personalization efforts, including the global recommendation system, are about helping members find something they will love to watch as soon as they open Netflix,” says Gomez-Uribe. “Knowing that we have 60 to 90 seconds to help you find something great, it is our goal to develop the most personalized experience as possible, based on your unique preferences and tastes, so we can surface the titles you will enjoy as fast as possible.”
BUT HOW DOES IT PREDICT THE CONTENT?
It’s all about ‘big data’. Big data literally translates into an extremely large chunk of information or data that is processed by optimized algorithms for shaping better business strategies and outcomes.
Companies like Netflix collect millions of data points from their users and use machine learning to analyze and refine the algorithms to rank content according to your preferences. Machine learning helps automate millions of decisions based on user activities. It takes several factors into consideration while predicting videos from its catalog such as:
Interactions such as your browsing history or your ratings of other titles
Other users’ interactions with similar tastes
Relevant information like genre, categories, titles, the cast of titles
The time of the day you usually watch content
The devices on which you watch them
The usual time duration of your watching hours
All these user interactions act as data points for the algorithms to deliver better and personalized performance.
For example, if you liked watching FRIENDS, Netflix will recommend you to watch Brooklyn Nine-Nine based on your watch history, ratings, and other interactions.
NETFLIX’S WAY OF RANKING TITLES
It is quite clear that Netflix utilizes a two-tiered row-based ranking system, where ranking happens:
Within each row (strongest recommendations on the left)
Across rows (strongest recommendations on top)
Each row highlights a particular theme (e.g. Top 10, Trending, Horror, etc), and is typically generated using one algorithm. Each member’s homepage consists of approximately 40 rows of up to 75 items, depending on the device the member is using.
The advantages of using a row-based ranking system can be seen from two perspectives — 1) As a user, it is more coherent when presented a row of items that are similar, and then decide if he or she is interested in watching something in that category; 2) As a company, it is easier to collect feedback as a right-scroll on a row would indicate interest whilst a scroll-down (ignoring the row) would indicate non-interest (not necessarily irrelevance).
With the help of machine learning and data science, Netflix has managed to build a robust recommender system to keep its viewers hooked on their screens and they have indeed done a great job. The company has confessed to save about $1 billion every year due to its smart recommendation engine.