Artificial Intelligence (AI), along with its various components, such as Machine Learning, Neural Networks, Image Recognition, Text Recognition, Speech to Text, and a number of others, is a hot topic these days. In the media industry content providers are examining how AI can help them improve engagement, reduce churn, improve revenue, lower costs, and improve margins.
According to Christopher Whitely, Senior Director of Applied Analytics at Comcast, “there is a trend towards using machine learning models to deliver the most relevant content to consumers to keep them engaged, whether that is programming that they might watch or advertisements that are of interest to them.”
To help Comcast, FOX, Netflix, AT&T, Hulu, and other content providers achieve these goals are a host of AI vendors, all with very professional and polished presentations and demos. As a result, expectations are high and the hype even higher. But can AI meet these lofty expectations?
To answer this question we spoke with TiVo and IBM to learn whether the use of Machine Learning and other AI techniques to improve personalization has delivered improved business results.
In a case study published by Tivo in conjunction with one of their content provider customers (The ROI of Personalized Content Discovery: Increased Views and Engagement, January 2017 – April 2018), TiVo showed how the use of a Personalized Content Discovery (PCD) platform utilizing Machine Learning and other AI techniques yielded the following results
- Increased engagement by 35% over manual curation w/out testing or analytics.
- Personalized recommendations out performed hand curated title 4 to 1.
- Personalized search and recommendation results increased the number of channels watched by 25% over those w/out personalization.
- Personalized Content Discovery (PCD) increased overall viewing by 40%.vs. no personalization.
- Conducted A/B testing within the PCD platform for new fall season content. Found the click-through rate for a lesser-performing TV show increased by 30x when promoted within a “Recommended for You” carousel.”
A second study (Personalized Content Discovery, Q3 2018) conducted by TiVo in conjunction with three pay TV provider customers (two EMEA, one U.S.) over six months showed that subscribers on PCD platform utilizing Machine Learning and other AI techniques were 3X less likely to churn than non-PDC subscribers.
IBM has seen similar results to Tivo. In general, media & entertainment video publishers using IBM Watson Media's AI-driven Video Recommendations solution to personalize video content for viewers have seen the following results.
- A 70% increase in video views over no personalization.
- A 7% bounce rate reduction over no personalization.
One international publisher saw the following after implementing IBM Watson Media's AI-driven Video Recommendations.
Problem: This publisher had 100K videos in their content library but with little to no video metadata and taxonomy, they weren't making the most of that content. Specifically, the company struggled to monetize its U.S. based viewers.
Solution: IBM Watson Media's AI-driven Video Recommendation solution extracted keywords to create categorized taxonomy automating the search and discovery process and improving video recommendations. With this algorithm they were able to promote original content, and play branded video as a first recommendation to U.S. viewers. In addition, the publisher was able to personalize its video players to accommodate individual viewers and increase their engagement.
Result: Video views increased by 125% and monetization by 87%.
While the use of Machine Learning, Neural Networks, Deep Learning and other AI techniques are surrounded by a significant amount of hype, type Artificial Intelligence into Google and you will find 84.2 million news stories, it is not without some merit. As seen here, the implementation of AI on personalization can yield tangible improvements to engagement, reduce churn, and improve monetization, all of which will ultimately improve the bottom line.