HOW NETFLIX USES BIG DATA FOR CUSTOMER SATISFACTION

Vedant Kapoor

Every 2 days we create as much data as we did from the dawn of time until 2003. The act of accessing and storing large amounts of information for analytics has been around a long time but big data gained momentum in the early 2000s.


Netflix, a world leader in streaming services, has been a data-driven company since its inception. It is one of the world’s most successful companies using big data and AI to boost its sales and improve customer experience. Did you know Netflix customizes different trailers for different users? Ever wondered how you immediately get a recommendation to watch ‘Designated Survivor’ after watching ‘House of Cards’? These are some questions we will delve into in this article.


With a market capitalization of about 232 billion USD, Netflix has managed to surpass Disney and all other streaming services including YouTube and Amazon Prime. Netflix has a retention rate of 93% compared to Hulu’s 64% and Amazon Prime’s 75%.

Figure 1. Low Churn Rate


But the only reason behind its success isn't the low churn rate. Netflix is ahead of its competitors because it also makes more successful movies and TV shows. It has managed to identify what the audience wants. How has Netflix managed to do this? Netflix collects all kinds of data from each and every customer. This includes what you watched, when you watched it, where you watched it, which device you used, if you finished the whole series/movie, your ratings and even when you paused.

Netflix has around 193 million subscribers and even if it manages to collect data from all these customers, what does it do with it? This data is analysed carefully and used in its recommendation algorithm. According to Netflix, about 75% of its user activity is based on personalized recommendations. As soon as you finish watching a show or movie and go to browse, Netflix will have created a ‘You may also like….’ section where you’ll find movies and series similar to the ones you've already seen. According to Netflix, they earn more than a billion in customer retention because the recommendation system accounts for about 80% of the content streamed.


Data is useful only when analysed well. Over a period of time, Netflix manages to keep a detailed profile of every customer. By keeping track of what you've been watching, it knows exactly the kind of shows you'd prefer. This not only helps in the recommendation process but also allows Netflix to make certain changes that would make it more likely to watch a particular content. How does Netflix convince you that a title is worth watching?


Let’s take the example of artwork. After conducting several experiments, Netflix found out that the artwork representing the title was something that caught the customer’s eye.


Figure 2. Different images cover various themes in the show which go beyond what any single image portrays

However, due to the diversity of tastes and preferences, there could not be a single perfect artwork for all users. That’s what led Netflix to find the best artwork for each of its users, highlighting those aspects which are specifically relevant to them.

To give you an idea, let’s see how two different customers might end up watching ‘Good Will Hunting’.


Figure 3. Personalisation on the basis of genre


In the first case, the customer has been watching romantic movies so he may be more interested if he sees the artwork with Matt Damon and Minnie Driver.

In the second case, the customer has been watching comedies, so he may be interested in an artwork with Robin Williams, a well-known comedian.


Similarly, trailers have also been customised for different customers. If you've been watching a lot of TV shows centred around women, it will alter the trailer and provide one with more focus on the female characters. Similarly, if it is observed that you've been watching contents with a particular actor, the trailer and/or the artwork will be focussed on that actor.


Figure 4. Personalisation on the basis of actor


Netflix uses a row-based ranking system. The users will see the strongest recommendations on top (across rows) and on the left (within each row). One of the reasons for using rows is that it is easier for Netflix to collect feedback. A scroll-down would indicate that the user is not interested in that category. On the other hand, a right-scroll on the row would indicate that the user is developing interest. Some of these rows are Top 10, Recommended For You, Trending now, Crime, Comedies and several other genres. To illustrate, Netflix uses the Trending Now ranker by capturing the temporal trends which the algorithm deduces to be strong predictors. For instance, the current situation of the Coronavirus pandemic has led to a growing interest in documentaries like ‘Coronavirus Explained’ and ‘Pandemic’. Moreover, seasonal trends like Valentine’s Day lead to increasing views of romantic videos.


Conclusion

The amount of data Netflix collects is limitless. As we have seen, almost everything that Netflix does is driven by data and powered by smart AI algorithms. After reviewing several case studies, we get the idea of how Netflix manages to stimulate customer satisfaction by focussing on details like artwork, trailers and the row-based ranking system. Thanks to big data, Netflix knows which shows are most watched, which actors are most liked and what genres are preferred. This information is used to stream movies or shows of the people’s choice-either by getting a license or by making its own in the form of Netflix Originals.


With more than 1/4th of the world’s population under lockdown, the entire world is streaming more than ever. Netflix has witnessed 16 million new sign-ups i.e. more than 60% of its subscriber base. Following an approach that provides superior data and instant feedback is how Netflix is surviving and growing in today’s innovation-intensive and data-driven world.

SPONSORS

Platinum Sponsors

Gold Sponsors