In the never-ending quest to reduce churn, cable operators are turning to predictive analytics not just to anticipate network concerns, but also to look at subscriber activity. By tracking data related to an individual customer, particularly those who have a history of churning, a cable operator can take preventive measures to decrease the likelihood that the customer will churn again.
"Algorithms can run real time, looking for indicators or symptoms of problems or potential problems. They draw on historical trends and what (customers) are saying (or doing) right now," said Bob Hallahan, VP, solutions architecture, at Xavient Information Systems.
The three issues driving turn are quality of service and experience (QoS and QoE), customer service, and content packaging and pricing, Hallahan said. To address the first from a consumer perspective, work is being done on self-service portals to allow subscribers a deeper insight and visibility into the service and more control over resolving issues.
"The more educated (they are) and the more they understand what is involved with delivering these services to the devices in their home, the more likely it is they will be more understanding what the causes are," Hallahan said, noting as an example that in-home WiFi problems are often a result of how the customer configured the router.
Regarding customer service, voice recognition technology can be used to monitor live and recorded calls to determine if customer service agents are handling situations appropriately. For example, are they upselling when the chance arises or apologizing when necessary?
Additionally, this data can be correlated with e-mail chats or website searches, which ties into packaging and pricing. Is a customer looking at the billing page or packaging info, for example? Perhaps the subscriber has come off an intro package and has a history of switching providers when the price rises. Knowing this, the cable operator has the chance to intervene and offer a solution before the customer leaves.
"The operator can more precisely target existing customers with the highest propensity to take some negative action. The worst case is churning or downsizing and dropping down to the smallest package," Hallahan said.
The use of predictive analytics can also help with locating and helping what Hallahan called the dissatisfied, non-complaining customer. Relevant data ranges from the state of the network or the IT systems website to a tracking of social media and comments made about the operator.
"Having a bigger set of eyes and ears out there to pick up on any kind of negative indicators … understanding the source, who is making the comments … (the company) can be proactive and offer resolution or change perception," Hallahan said.