OTT providers are quickly catching on to the benefits of user recommendation services being integrated into their platforms. However, both of the two main options for integration come with drawbacks. Off-the-shelf solutions can be costly to bring in, but cheaper custom deployments require the right expertise in-house. What’s needed is a meet-in-the-middle solution that saves on expenditure and is intuitive to use, while crucially providing recommendation services that impress viewers and ultimately helps to retain them.
The pitfalls of current strategies
A heavily saturated market and sustained pressure on business finances is leading OTT providers to restrict their budgets. It’s unlikely that they’ll want to invest heavily in an off-the-shelf solution with high capital and per subscriber costs, despite the ability of these platforms to help teams deploy personalized recommendation services quickly. Significant results need to be seen straight away, or else the board won’t hesitate in demanding answers from decision-makers as to the expense.
OTT providers are then exploring the alternative of an in-house solution, but they frequently find they don’t get far when the skills of their employees are insufficient to effectively build it. Instead, a solution that brings together data, analytics and machine learning (ML) in a flexible and cost-effective package is vital to delivering the most innovative recommendation services to end users.
Personalization in more ways than one
The answer is a unique personalization platform with serverless and scalable features, with AWS Personalize being a prominent example. It requires little working knowledge of ML, and is highly secure when implemented correctly. This kind of technology fits all the requirements for OTT providers looking to leverage the opportunities provided by recommendation services. While a “recommended for you” carousel is a great feature to have, consumers can now be engaged in a variety of other ways as well.
Providers have the opportunity to integrate a “more like this” or “other users also watched” carousel with the help of user interaction data. A viewer that’s just finished watching a documentary about the history of the Olympics might be interested in the origins of association football, for example. Historical viewing habits can also help shape the journey for other viewers with shared interests.
Optimal use of data can even play into how search results are shown. These lists can be re-ordered based on the individual viewer, or drive editorially curated content recommendations. This is an exciting prospect as search results can be individually tailored to the user and their interests. While AWS Personalize provides value to numerous markets, its personalized recommendation algorithms can be ideally implemented into content services.
Further opportunities for engagement
The technology can also take OTT providers in some other exciting directions. Say the OTT provider is aiming to promote an individual content item in recommendation results. The recommender can be tweaked to show a certain amount of content from a particular promotions category. Contextual recommendations are also coming to the fore. These recommendations can be delivered based on the device used, the time of day or the location of the user.
However, it must be said that using AWS Personalize in a media context is more complicated than its other managed services solutions. To ease this pain point, there are external solutions out there that can take the hard graft out of provisioning AWS Personalize. One is Merapar Development Kits (MDKs), which can enable the ability to transform and ingest significant amounts of data. With this solution, OTT providers can be up and running straight away. Modules can allow flexibility in how much of the technology is used, depending on requirements and available budget. OTT providers that want to deploy new features rapidly, be competitive in the market and trial new services can easily do so.
Testing of new features is vital to decipher what’s working and what isn’t, and can drive continuous improvements. There’s a few ingenious ways that the technology can do this. One example is the ability to deploy multiple recommendation models that are driven by user data. Parameters are adjusted automatically and then the best performer in the trial is selected to be used moving forward, with recommendations then becoming more accurate over time.
Alongside continuous improvements to accuracy, it is of course essential to know the levels of user engagement with suggestions provided to them. Clickstream data can be exported out to an analytics platform to enable organizations to measure performance against their main KPIs. Some examples of these include conversion rate of journeys from a recommendation click, the click-through rate (CTR) or even the levels of engagement from customers that watch or click on recommended content. These insights can power optimization of services and ultimately help increase user engagement and reduce churn.
Unique in the market
In a market that’s continuously witnessing new entrants on a regular basis, differentiation has never been more vital. This has been made even more pressing by the cost-of-living crisis as more consumers query the value provided by their current subscriptions, which they may be spending a significant portion of their income on. A varied set of recommendation services, from basic carousels up to contextual suggestions, are vital tools in pushing competitive advantage. The question that remains is how these capabilities can be brought in without breaking the bank and relying on the limited skillsets of current employees. Brought to life by a specialized toolkit, serverless, scalable personalization platforms can deliver recommendation services that meet viewer expectations.
[Editor’s note: This is a contributed article from Merapar. Streaming Media accepts vendor bylines based solely on their value to our readers.]
Personalized content recommendations are a hot topic in streaming. Extending them to your emails and other marketing messages is a logical way to increase repeat visits to your apps and counteract churn.
07 Sep 2022
Metadata quality is often the biggest challenge for any streaming service looking to increase content engagement with targeted recommendations. Can machine learning come to the rescue?
30 Aug 2022
If you’re looking to add or upgrade your data-driven recommendations, should you go for Recommended For You, Because You Watched, or Trending to bring the greatest increase in engagement? 24i’s SVP of Data Products, Damien Read, argues the choice of section names and the data models behind them should be tuned to your user base and your specific business goals.
16 Aug 2022