When wireless networks go down, the speed with which operators can respond is critical. The first step is identifying the problem, which can mean piecing together data from disparate service calls and even social media. Nokia, which supports 1.5 billion mobile devices with its Motive care and provisioning software, has found machine learning can help diagnose problems faster and also prevent them from ever occurring.
“That’s where you see such dramatic numbers like 90% reduction in truck rolls or 85% reduction in help desk calls,” said Rich Crowe, Nokia’s head of marketing for customer and network operations. “If I can identify that major outage and correlate a number of different calls to one big thing, I can obviously proactively notify my customers so they don’t have to call. Also, I know I’ve got one problem instead of ten little problems.”
Motive, which Nokia acquired as part of its Alcatel-Lucent acquisition, uses machine learning algorithms developed at Bell Labs. Those algorithms are now part of Motive’s service management platform and care analytics solution. Crowe said machine learning enables “self-optimizing workflows,” meaning customer care agents and applications follow a sequence of tasks designed to deliver a higher probability of resolving billing, subscription and network service issues in the shortest amount of time.
“When a care request comes in, there might be three or five or seven different things that possibly could fix this problem,” Crowe explained. Recent history of care requests and outcomes will tell the system what is most likely to work. “Based on that learning, the service management platform is able to apply what we call the next best action, in other words the action that is most likely, at that moment in time, to solve that customer’s problem,” Crowe said.
Machine learning will be even more important as the number of devices multiplies, Crowe said, noting Nokia’s care solutions need to move towards artificial intelligence to get ready for the time when automated service calls come from connected devices that self-report.
“If you think about 10 billion devices going to 35 billion devices by 2020, with most of the increase coming from things, all of these things are going to need to be cared for,” Crowe said. “You need automation, you need machine learning and in some sense this release is a first step in that direction. We kind of have [“internet of things”] in our headlights as we start thinking about the need for all this.”
Nokia said Motive’s enhanced care analytics breaks new ground by automatically correlating customer help desk calls and self-care actions with network, service and third-party application topologies to identify call anomalies. Once anomalies are identified, Motive is designed to initiate actions through its service management platform.
The top drivers of customer care interactions are slow internet/low-quality mobile video, device activation and billing issues, Nokia said. The company said the Motive software can diagnose last-mile issues in wired broadband networks as well as mobile service issues, and issues with customer premises equipment and third-party applications.
“Service disruptions are often hard to identify because they happen in the access network, on customer equipment or on customers’ devices. Traditional customer care may only address a small part of a larger problem and the time-consuming, step-by-step troubleshooting process can lead to customer frustration and the risk of lost business,” said Bhaskar Gorti, president of applications and analytics at Nokia.
The risks associated with service disruptions will go beyond lost business as operators connect millions of machines to their networks. Crowe said Nokia is keenly aware of the challenges and opportunities that the “internet of things” presents for customer care, and will have more news to announce at Mobile World Congress in February.
