Xavient says deep analytics will help OTT providers retain customers, reduce churn

analytics

Consumers may be tuning more towards over the top (OTT) video content, but the diverse set of content options and flexible options is driving ongoing customer churn.

Xavient Information Systems, an IT consulting and software services company, estimates that the churn rate for OTT is higher than traditional video services.

Unlike a traditional video service from a cable provider, a consumer does not have to lock themselves into a long-term contract or rent equipment like a set top box (STB).

“Churn rate for OTT has been significantly higher than traditional broadcast and on-demand operators,” said Bob Hallahan, VP of technology at Xavient Information Systems, in an interview with FierceOnlineVideo. “Of course, that’s due to the flexible nature of the packages, services, billing and promotion campaigns are structured.”

RELATED: Parks: 19% of U.S. households cancelled an OTT service in last 12 months, but churn remains stable

The company’s thesis is not unfounded.

A recent Parks Associates study revealed that since the end of 2015, 20% of U.S. broadband households had cancelled at least one OTT video service in the past 12 months.

However, the research firm found that households with OTT video subscriptions increased their spending from an average of $3.71 per month in 2012 to $7.95 in 2016.

Understanding customer characteristics

Xavient sees a number of two types of OTT customers that are behind the churn trend.

The first category are customers that will sign up for a free OTT service in trial order to binge on a particular program commercial. After they complete viewing a series, these customers will end their subscription.

Likewise, the second segment will sign up for a paid service, but only for a short duration to also binge on a particular show or series.

“It’s all about the conversion rate,” Hallahan said.

What Xavient has been doing to help our over the top providers address these challenges is developing analytical algorithms.

These solutions provide actionable insight into the probability that the customer may or may not convert into a long-term paid subscriber or will they churn and the reasons for the decision to churn.

“We can identify early stage symptoms, which is a category of analytics called anticipatory analytics,” Hallahan said. “Similar to technology that’s in cars today, it anticipates something that has not happened, but has a high degree of happening in the next few minutes or days.”

The OTT industry has gone from sentiment analytics, which reflects the mood of the customer, to more predicative analytics that focus on whether a customer will either continue subscribing to the service or not.

Hallahan said usually that “type of information is a little bit too late in the game or in the cycle for an operator to react quickly enough to head off churn.”

Isolating churn symptoms

As a result, Xavient can equip an OTT provider with symptomatic analytic engines based on deep learning algorithms.

These deep learning analytics are focused on three areas: content packaging and pricing, customer service, and service quality.

Leveraging Kafka open source technology, Xavient’s approach is to streaming video performance by collecting data off the network, the device as well as real-time metrics on usage. The system will then pull a wide range of sources during the real-time broadcasting of the streams into a data ingestion engine.

Xavient is also using a voice recognition engine that monitors and transcodes text to scan for key symptom words and correlating that in real-time to understand what’s happening with phone calls coming into customer call centers.

“This looks not just at what the customer says, but also what the customer service rep is saying,” Hallahan said. “Are they saying the right things or not saying the things they should be saying, which could turn into potential churn.”

In order to minimize churn, OTT video provides also need to make what Xavient calls a “course correcting” action.

This could be everything from offering a coupon or something that’s tailored to a customer’s likes or dislikes.

“By getting into specific algorithms, we can give an OTT provider the precise ability to make a recommendation that statistically will yield the type of result they are looking for,” Hallahan said.