Hidden in plain sight: how OTT platforms can mine user preferences

How challenging is it to predict what users want to watch on OTT platforms? Our guest author weighs in.

Wall Street guru and statesman Bernard Baruch had a simple mantra to get ahead. “Most of the successful people I’ve known,” he said, “are the ones who do more listening than talking.” 

The same holds good with running a successful OTT brand. A treasure trove of great content and effective advertising are just the beginning. But what really makes some OTT platforms click, better than others, is their ability to capture viewer feedback and act on it. In an industry where churn rates were at 45% in Q3 2021, it’s the key to retaining customers. 

But predicting what viewers prefer, is challenging, since COVID-induced changes have triggered massive shifts in users' emotions, psychology and behaviour. Many of these are difficult to fathom. Or, are they?

User preferences: a goldmine of insights for OTT brands and creators

With so much OTT content available, it may take a viewer a lifetime just to consume all that a single platform offers. This is why successful OTT players need to be acutely mindful of users' choice of content. The 'whys' and 'whens' of their choices, if accepted, studied and acted upon, form the bedrock of successful customer engagement strategies.

Conversely, without a personalised approach and sharply focused recommendations, even great content platforms can fall by the wayside. Netflix, for example, garners as much as 80% of its watch time from personalised recommendations. 

If acting on user preferences is so crucial for an established industry giant, you can only gauge its importance for challenger OTT brands to succeed. Viewers who experience AI-based recommendations, have been shown to consume three times more content per viewing session.                          

Beyond addressing the growth and customer retention needs of OTT platforms and the viewing needs of users, user preferences help creators too. With user preference insights, content creators can gauge which genres, storylines, star casts and other parameters are working for them. This helps them up their content game, while shaping result-driven content partnership strategies for OTT platforms.

You have them, but are you using them? Executing a robust content personalisation strategy   

Content personalisation strikes a chord with viewers, as it seems like the platform truly cares for their likes and dislikes. Offering a personalised piece of content increases the positive perception towards the platform, making viewers strikingly more loyal towards it. 

In addition, it helps the OTT marketers in evaluating each content simultaneously and accurately gauging the change in consumption patterns over time.

The challenge here is not gathering user preferences, but having a strategy to utilise them. In most cases, viewership data exists in the OTT platform's vast data lakes. 

But using this data to provide a meaningful and enjoyable OTT user experience is as much an art as a science. It involves complex type of analysis and predictive analytics to drive the best value for the end user. By crunching viewership data available to them, OTT service providers can mine insights, notice trends and build viewer affinity.

Different folks, same strokes: the content personalisation toolkit

OTT platforms are constantly innovating to use recommendations better – Netflix’s ‘Surprise Me’ function is one such example. But, at a fundamental level, pooling and analysing data remain at the core of content personalisation. 

Data and AI algorithms should be able to access and crunch data about your users' behaviours, such as preferred genres, actors, teams or sportspeople. It should be able to accept both explicit – likes, shares, watch duration and ratings – and implicit data points. 

These could be demographics, weather, local and global events, connectivity and time of viewing. Finally, the platform’s AI algorithms should then be able to automate to serve appropriate content based on these data points.

In analysing these data points, AI and machine learning play a vital role. These technologies intuitively recognise patterns in viewer behaviour, which can be scaled to predict how other viewers, with similar attributes or demographics, consume content. 

Chorki: a content personalisation success story

Chorki, a Bangladeshi regional OTT content pioneer, approached ViewLift to help in its recommendations game. Using a combination of analytics and processes, we accurately captured user preferences, likes and dislikes on each content piece. 

We then used this collective data to help Chorki figure out its top-performing movies and shows. Based on these insights, Chorki was able to offer highly personalised content to viewers, right on the site's home page.

The approach of accepting user preferences has reduced Chorki’s churn rate by a significant margin. It has also enabled Chorki to modify its content creation and distribution strategy to mirror user demands. This has led Chorki to enjoy a significant CAGR.

(The guest author is a front-end architect at ViewLift, a cloud-based OTT platform)