4 Ways AI Can Improve User Engagement in OTT Video Services

Artificial intelligence (or “AI”) technologies is one of the hottest topics out there today. Not only does its implementation help elevate inner functionalities of hardware and software, AI also plays an increasingly crucial role in user engagement, especially when related to over-the-top (“OTT”) video services. 

Mixed reality 01_web-01

But let’s start with stating the obvious - an OTT platform is not simply a video player. It’s a much more complex solution, encompassing a variety of elements from technical to editorial, including reporting tools, customer support integration, subscription and billing management, marketing, social media, technical support and operations.

As an additional layer, OTT for sport-specific environments need to deliver a number of key elements:

1. High quality, low-latency video streaming, both for live and VOD

2. Global scale

3. Very narrow load and usage peaks

4. Fine-grained content rights management

Deltatre’s architecture for OTT platforms, built from the ground up for sports events, presents itself as an intricate combination of all of the above. Cloud-based or hybrid scenarios for digital data are supported; up to 11 end-user platforms are supported too, with the same UX/UI on web, mobile, smart TVs and gaming consoles.

From venues to end-users, covering the entire workflow, a unified management system sits at the core (subscriptions, payments, analytics, CRM, etc.).

How can AI fit into this big picture? Here are four ideas.

#1. AI for Content Editing

What if service providers could tailor specifically targeted marketing campaigns, optimising their strategy through an “intelligent” editor on Forge?

Greater personalisation of the content offering naturally leads to monetization, empowering commercial teams to quickly identify the most effective customer engagement patterns.

Applied to Deltatre’s CMS, Forge, the AI-powered smart editor would enable automatic analysis of the text to support auto-tagging and entities identification. This would help operators to write outstanding pieces of high-quality content that are primed for virality and a broader sports fans engagement rate.

It also suggests additional content, images and or videos to reference, picking from the internal CMS or from the web.

#2. AI for Video Indexing

A video content indexer makes video content actionable, anticipating needs customers might not even realise they actually have. It makes content more discoverable, improving engagement.

When integrated on a CMS such as Forge or in Diva video workflow, a video indexer would let you upload your video and start finding insights right away, without even writing a single line of code. The metadata extracted, finally, helps making the content consumption interactive and therefore memorable.

Not only does this involve the usual “search, discover, recommend” functionalities already common in sophisticated OTT solutions - it takes them to the next level. This can happen through audio transcription and speaker recognition, auto-captioning, as well as a deeper understanding of emotions through a variety of specific cognitive services.

Learn more about Deltatre’s Diva video player after the release of its 4.5 version.

#3. AI for Recommendation

Hands up if you, at least once, have been staring at your screen, clueless on what to watch next after enjoying a live sport event or consuming all the on-demand content available from your favourite team? A sophisticated, data-driven recommendation engine is that ever-present virtual friend who suggests which video to watch, based on the users own watching history. It integrates with AI to make sensible recommendations and increase the longevity of your watching experience.

“Collaborative filtering”, in particular, is a key concept here: a custom machine learning model matches content to users, according to other users’ similar behaviours, which are considered here as latent variables, not directly observable yet impacting the final outcome. So, think about this: if 99 fans of House go on to watch ER, it’s logical than the 100th fan of House will also perhaps like the mid-90’s hospital drama.

Amazon is well-known for personalization and recommendations, a “collaborative filtering” model that has helped and helps customers discover items they might otherwise find. The company underwent over two decades of challenges, creating great recommendations as the business has grown from a bookstore to selling just about everything.

In 2010, YouTube reported using the technology for recommending videos. Netflix is famous for its introduction of recommender systems, using them so effectively that their former Chief Product Officer, Neil Hunt, had indicated that more than 80 percent of movies watched on Netflix came through recommendations, and placed the value of Netflix recommendations at more than $1 billion per year.

Let’s put it this way - if the most successful media companies in the world do it, don’t you want to do it too?

#4. AI for Next Gen Content

Content creation clearly does not look like it used to. It’s hard work to build a closer and more profitable relationship with an audience.

Among the different approaches content creators will need to take, there is the ambitious field of creating immersive 3D videos taken from 2D video source file.

Learn more about Massive’s AXIS, a tool to leverage AI to deliver a fully targeted UX experience

The University of Washington, along with Facebook and Google researchers, built the first PoC prototype, named “Soccer On Your Tabletop” .

It’s a system that transforms a monocular video of a football (soccer) game into a moving 3D reconstruction, in which players and field can be rendered interactively with a 3D viewer. How? Through a mixed/augmented reality device. The project had as an output a Mixed Reality concept created for Microsoft HoloLens. It clearly responds to the needs of sports organisations to bring fans closer to the action by revolutionising their experience. As if they actually were at the stadium.

A Bright Future Awaits…

AI helps to enhance users’ watching experience and to overcome the belief that an OTT platform is a mere video player. It’s not.

It’s a mean for fans to be taken to the centre of the action through their very own screens and devices. From content management to a sophisticated architecture of recommendations, the implementation of such technologies is by far among the most exciting sides of a future for the OTT world and what will come next.

Gianni Rosa Gallina
R&D Senior Software Engineer