The key success factor of your customer experience personalization comes from the effectiveness of your recommender system. How well can you predict what each of your customers is willing to buy when they go to your website or enter your shop?
In this article, we want to share with you some of the Netflix’s approaches to building a successful recommender system. Netflix is a company that is well-known for the accurateness of recommendations it provides to its users. Over 80% of what Netflix members watch comes from their recommendations.
If this educational content on using AI in marketing is useful for you, you can subscribe to our Enterprise AI mailing list to be alerted when we release new material.
Balancing Discovery & Continuation in Netflix Movie Recommendations
Hossein Taghavi is the Research & Engineering Manager for the Machine Learning team at Netflix in charge of user recommendations. In this talk given at the Machine Intelligence Summit organized by RE•WORK, he gives an overview of how Netflix balances both discovery and continuation in their recommendation algorithms.
You’re welcome to watch the full talk or simply continue reading our key takeaways.
Regardless of how sophisticated your recommendation algorithms are, the content you present to users on your website is a suggestion of what they should engage with next and has a massive impact on your engagement metics.
At Netflix, they recognized that to build a really good recommender system, it is important to balance for different modes of watching:
- some users come to continue watching a TV show they started yesterday;
- others want to discover a new movie to watch;
- yet other members are up to rewatching a title they enjoyed in the past.
So, when you go to Netflix, you’ll get recommendations for all modes of watching. Moreover, if the algorithms “think” that at this particular moment of time, you are more likely to continue watching TV-shows or movies that you are in the middle of, the Continue Watching row will be at the top of your Home page.
Netflix’s recommendation algorithms recognize that you’ve just completed all episodes of a show or just finished a movie, and so you are more likely to look for a new title. They also place the row of new titles you’re likely to enjoy next at the very top of the page.
The Netflix recommender system is also context aware. For example, if the titles you watch depend on the device you are using – let’s say, you watch cartoons on iPhone but go to the website to see the last episode of Sherlock, the recommendations will be different across your devices.
If you’re an intermediate or advanced practitioner, we’ve created premium education to help you and your team master recent breakthroughs in applied AI for marketing. Click here to purchase our AI in marketing research summaries.
RE•WORK is the leading global events company in AI and Deep Learning, highly respected by pioneers such as Yoshua Bengio, Yann LeCun, Geoffrey Hinton, Doina Precup and Joelle Pineau. At each event, they bring together leading experts in the field to share the latest technological advancements as well as practical applications in business and society. The brightest minds from a variety of industries cover topics such as Natural Language Processing, AI Assistants, Robotics, Speech Recognition, Reinforcement Learning, Computer Vision and many more. If you’re interested in learning more from the likes of Google Brain, Uber AI Labs, DeepMind, MIT, Facebook amongst others, take a look at their upcoming events here: https://www.re-work.co, or watch more video content here: http://videos.re-work.co/.
Enjoy this article? Sign up for more updates on using AI in marketing.
We’ll let you know when we release more content on applying artificial intelligence in marketing.