A twelve-year old data mining company that deployed its first personalized multi-screen recommendation engine in 2006 contends that intelligent navigational technology can benefit Pay TV operators and subscribers alike. Counting 20 large service providers among its customers, Think Analytics brings a number of techniques to bear in the arena of proliferating content choices.
Think Analytics’ longevity and customer breadth distinguish it from many of the start-ups in the recommendation-engine space. Technology, however, may be the underlying differentiator.
“We’ve seen early basic recommendation systems being deployed, the if-you-like-this-then-you-like-that collaborative filtering approach, like Netflix,” said Founder and CTO Peter Docherty. “But from our perception, these are sort of basic, early steps, good but early.”
A commonly used technique on the Web that enables automatic predictions based on the preferences of many users, collaborative filtering is both basic and limited. One problem is that the technique assumes that viewers know what they want. “What we’re trying to deliver is a much more intelligent navigation,” Docherty said.
Think Analytics’ approach expands upon collaborative filtering with multiple algorithms that aim for better understanding of customer’s likes and dislikes, incorporates content awareness through systematic metadata analysis, including a massive base of pre-tagged movies, and supports a wide range of use cases and customer experiences. And it backs up its software with “industrial strength” scaling and no-zero-downtime reliability since 2006.
As with other recommendation engines, Think Analytics draws a demarcation line between its own platform and the customer user interface (UI) that service providers tend to claim as their own. The two nonetheless work in tandem. Docherty offered the BSkyB guide as a case in point: “When you select a program, you see the details of the program you have selected, and also down the right-hand side, you can also see the recommendations.”
This creates a win-win dynamic, Docherty said. The operator is able to promote content and increase revenue by marketing content outside of a customer’s existing package. “But customers also have benefits,” he said. “They’re being shown content that (the recommendation engine) knows they’re going to like. It’s being proactively presented to them. And customer satisfaction goes up.”
Results that Think Analytics obtained from another large Pay TV operator are impressive. They showed 70% of individual recommendations being executed within 24 hours, a 50-60% above-average purchase rate among customers receiving recommendations, and 90% positive ratings from viewers that received content recommendations.