Right now, as you read this article, billions of people are sitting at home with no access to cinema halls or restaurants, and only the internet for their source of entertainment. This sudden shift has created a massive impact on all sorts of businesses including the entertainment industry. As filmmaking came to a screeching halt but consumer demand remained ever-present, over-the-top (OTT) media services witnessed a sudden surge in demand that was not foreseen by any of the players.
While the past few weeks have obviously benefitted all OTTs, but it also gave rise to questions like will the OTTs continue to enjoy the same growth as the lockdowns ease across the globe?
Currently, giants like Netflix have witnessed an uptick in their subscriber count and their shares went up by 35% in the first quarter of 2020. We also saw rising popularity in home-grown content (John Krasinski’s Some Good News comes to mind), during these uncertain times. Quibi, a bite-sized content platform, finished its latest round of funding in March 2020. It has shown of 10-minute episodes with many big stars and titles.
However, with a plethora of OTT players rising to this unanticipated golden opportunity, customers are spoilt for choice more than ever. The quality of experience that OTT players provide during this phase may also define their future once the lockdowns completely phase out. In such a scenario, we know what they can use to maintain this success prolong: Big Data.
Big Data: The unsung hero
Unlike their counterparts, cinema and television, OTT viewers have the option to watch anything at any time. In the case of televisions, you depend on the channel’s schedule, but with an OTT platform, it is content on demand. The lack of a fixed schedule thus puts the ownership of program selection on the consumer.
In a sea of choices, how does an OTT player deliver options that a viewer probabilistically has the highest propensity to watch, even if it means suggesting the viewer watch F.R.I.E.N.D.S for the nth time over any of the new releases since they are most likely to watch that! However, with no new content available on television barring daily news, OTTs have an opportunity to attract new subscribers to binge on their carefully selected content.
This is where big data comes into the picture (pun intended).
To a large extent, big data solves the issue of ‘finding what the user might like’ for OTTs through its recommendation systems. To understand this, we should take a closer look at the most prominent OTT platform out there, Netflix. Their relentless pursuit to perfect their customer experience is not an unknown fact.
Netflix quickly rose to become the largest OTT platform in the world by pre-empting customer demands and offering seamless experience at every touchpoint. It was almost as if they could read their customer’s minds and knew what their customers wanted to watch every time they came on Netflix. It is a lesser-known fact that Netflix owes a large slice of their success to big data.
They started with personalized movie recommendations, but after they switched to streaming, they began using data-based recommendation systems.
Demystifying the role of big data
You might already be aware of how advanced Netflix’s recommender system is. It uses multiple algorithms to solve complex issues. One such algorithm is the PVR Algorithm, where PVR stands for Personalised Video Ranking. It filters the titles according to specific criteria (Genres, Actors, etc.) while considering the products of user-based filtering. Netflix also incorporates RNNs for their time-sensitive predictions.
The ‘continue watching’ recommender not only looks at the content the user hasn’t finished watching (incomplete movie, unwatched episodes of a series), the algorithm also considers the point of abandonment, the device used to watch individual titles, and other context-based signals to optimize its recommendations. Recurrent Neural Networks are excellent for contextual sequence data, so it should not come as a surprise that Netflix employs them in their recommender.
During COVID-19, Netflix’s large library size and its various algorithms proved to be its key competitive advantage. Netflix used one of its algorithms, Trending Now Ranker to recommend shows and movies to its audience. This algorithm focuses on short-term trends that cause a rise in the popularity of individual titles.
For example, the OTT giant which got over 15.8 million new subscribers in Q1 of 2020, launched a new show at the beginning of the lockdown in February 2020 called ‘Pandemic’, which is a documentary that questions Man’s readiness to face a fast-moving virus-like Coronavirus. Our first thought would be – How is that for timing?
However, it is not a matter of timing. This was a topical series purposely launched by the OTT player to capitalize on people’s increasing curiosity around the virus.
Netflix also uses hybrid filtering systems to provide enhanced personalization to its users. Collaborative filtering helps it to identify the most popular titles while user-based filtering allows it to recognize the titles a specific user would like the most. They also change the image of its titles according to its user’s preferences. They perform A/B tests to realize which thumbnail would convince the user to click on it.
The other platforms are not far behind
While Netflix has been enjoying the sweet taste of success, its competitors have also been keeping busier than usual. OTTs like Disney+, Amazon Prime Video, Hotstar, and plenty of other platforms that compete for a similar customer base has been continuously using big data to combat the competition.
Amazon, which is considered the leader in customer experience, has also been a leader in recommender systems since its advent. They use topic diversification algorithms to enhance their recommendations. Through collaborative filtering and item-to-item matrices, they can keep their implementation scalable for different users. They also use content-based filtering methods to predict user behavior, where the impact of other users wouldn’t be too influential.
In the starting days of Amazon Prime Video, they solely focused on using collaborative filtering, which proved detrimental because people weren’t getting personalized recommendations. When Amazon identified this problem, they started using more user-focused methods, which led to a 2x increase in user engagement. Disney+ and Apple TV+ are relatively new in the market, and it will be interesting to see how their recommenders enhance the viewing experience.
OTTs aren’t the only platforms that use big data to personalize recommendations. Music streaming platforms like Spotify analyze the music a user listens to for improving its recommendations. Similarly, Pandora takes user feedback explicitly (through user ratings) and generates new music recommendations through that data.
Where do we go from here? A question on everyone’s mind
The current pandemic forced people to stay at home, which led to an upsurge in the viewership of OTT platforms. Surely, OTTs are witnessing a rise in their popularity, but will this trend continue? Will it change how people perceive entertainment? Will the pandemic mark the end of the television era? Netflix has officially announced that it is bracing itself for a drop in customers as the lockdowns are lifted and people step into the outside world again.
The player is not alone. As COVID-19 passes and the world moves on to the new normal, the steps taken by OTTs during the lockdown may define their future. OTT players that truly milk big data and accordingly rethink their content strategy and customer experience may continue to enjoy the success they are currently witnessing.
If you are curious to learn about big data, check out IIIT-B & upGrad’s PG Diploma in Data Science which is created for working professionals and offers 10+ case studies & projects, practical hands-on workshops, mentorship with industry experts, 1-on-1 with industry mentors, 400+ hours of learning and job assistance with top firms.