Understanding content key to success in streaming wars
By Marcus Bergström, CEO, Vionlabs
The rapid growth of streaming services like Netflix, Amazon Prime Video, and Disney+ means that consumers can access more on-demand content than ever before. Along with other challengers such as Apple TV+, Peacock, and HBO Max, the OTT streaming market is pretty flooded, making the battle for viewer attention incredibly tough. The services that will thrive will be those that offer a differentiated user experience, retaining subscribers and increasing engagement through intuitive and ultra-personalized content discovery systems.
One of the main reasons viewers decide to cancel streaming subscriptions is due to time wasted navigating vast content libraries: on average, users spend more than 25 per cent of their screen-time choosing what to watch. Content discovery recommendations are usually based upon information on the video content coming from an operator’s catalogue, supported by external metadata sources. However, the resulting viewing suggestions are generally fairly simplistic and inaccurate. As a case in point, a user may have decided to watch Uncut Gems, a recent crime-thriller film starring Adam Sandler. A regular content discovery platform that solely relies on analyzing metadata would churn out recommendations leading the viewer to some of Adam Sandler’s other movies – for the most part, light-hearted comedies, such as Grown Ups or Happy Gilmore. The platform might also direct you to its ‘Hidden Gems’ genre after reacting to the name ‘Gems’ in the title.
The output of recommendations is only as good as the input, and if providers don’t know enough about their content then recommendations will be, for the most part, highly unreliable and, at times, irrelevant. How likely is it that viewers who enjoyed Uncut Gems, a critically-acclaimed drama, will be immediately on the hunt for one of Adam Sandler’s older, goofier romcoms? Ultimately, viewers can be put off by an apparent lack of relevant content choice and decide to leave their service for a competitor.
Streaming providers can increase user engagement and improve subscriber retention by stepping up the game when it comes to content discovery. Accurate and personalized recommendations are crucial to keeping viewers on-screen, and the sooner that providers enhance their platforms with higher-quality content discovery based on more than just metadata sources, the better they will fare.
Typical metadata will still play an important role in content discovery, but it isn’t anywhere near enough by itself. AI has the capabilities to solve content discovery challenges that almost all operators face, combining highly-detailed video content analysis with the user’s viewing history. At Vionlabs, we use a number of different neural networks to identify patterns in audio, color, camera movements, pace, stress levels, positive/negative emotions and many other content features.
Our analysis of these variables means that we can extract information determining the emotional impact of movies, television shows and other video, producing the highest quality content description available: our content fingerprint timeline. Our AI engine understands how changes in the fingerprint timelines are linked to what content users enjoy. A vital component of this is a technique we call content similarity analysis, which compares each content timeline with every other timeline to evaluate the grade of similarity between each content asset. Our engine uses this knowledge and that of the viewer watchlist to provide users with highly accurate and personalized recommendations. A viewer that has just finished watching a horror film, for example, may want to move onto something more light-hearted and less intense afterwards. Our platform will be nuanced enough to understand this, serving up more intuitive recommendations.
AI-based content discovery will undoubtedly have a crucial role to play in the AVOD and SVOD markets, as well as in linear programming. OTT providers distributing both live and VOD services can better understand their content libraries by using content discovery platforms. They can then utilize their VOD libraries and provide new experiences, including ultra-personalized linear programming. Online music services such as Spotify have been using similarities in songs and music streaming behaviour to formulate channels and curated playlists for users. In the case of video, a similar deployment of content analysis can enable providers to create hyper-personalized viewing experiences through personally curated channels such as a weekly discovery, Mediterranean cooking shows or historical documentaries, all based on the user’s watch behaviour.
As well as being able to provide more nuanced and intuitive recommendations, a deeper understanding of content can benefit OTT providers in other ways, too. One example is in making more informed decisions when it comes to content acquisition and production, giving them an edge over competitors. Knowing what content viewers enjoy, and why they enjoy it will strengthen a provider’s position and allow them to strategize better. From the user’s perspective, the streaming platforms that have a more detailed appreciation of their content libraries will be able to offer better user interface (UI) experiences. Often, they can be flooded with inaccurate and basic recommendations, leading to a less enjoyable experience. However, with a better understanding of content, platforms can offer a clearer, more relevant display, even moving from countless tiles to a smaller, more focussed carousel, for example.
While content analysis reinforces OTT providers’ understanding of their audiences and offers a more personalized user experience, so too can it open up exciting new opportunities for advertisers. Increased personalization of content discovery can be used to inform more contextual, targeted advertising opportunities, creating an experience both relevant and engaging for the viewer, while also adding value, and another key differentiator, for providers.
In the increasingly saturated VOD market, providers need to stand out from the crowd and engage viewers by offering the best user experience possible. If not, they will simply move to another service, and that will become their new go-to content app. The OTT providers that offer a differentiated user experience, with the most accurate, relevant and personalized content discovery platforms will be in a much better position to survive and thrive in an increasingly competitive marketplace.
AI and machine learning have unlocked capabilities to enhance our understanding of video content through detailed and intelligent analysis. There are some fantastic opportunities afforded to OTT providers through AI to keep viewers watching and save them from switching apps. Increasing viewer engagement must be the battleground as the streaming wars enter their next phase. If providers achieve that, then they will be well placed to remain standing when the wars reach their conclusion.