Machine-based data helps telecoms improve customer service and reduce churn
Image: iStock/Andrew Ostrovsky
Gartner predicts that by 2016 70% of high-performing companies will manage their business processes using real-time predictive analytics or extreme collaboration. "The impact of integrating real-time analytics with business operations is immediately apparent to business people because it changes the way they do their jobs," said Jim Sinur. "The most dramatic change is the increased visibility in how the company is running and what is happening in its external environment. Individual contributors and managers have more situational awareness, so they are able to make better decisions faster."
Nowhere is this better applied than in the area of IT network endpoint management, where two-thirds of IT pros responding to a 2013 Ponemon Institute survey commissioned by Lumension said they did not believe their networks were secure, and 80% cited endpoint devices such as laptops, smartphones, and tablets as the greatest areas of risk.
"Organizations have thousands of endpoints around the world in their networks," said Puneet Pandit, CEO of Glassbeam, an IoT analytics company. "What they are learning is that the machine-generated data for these devices contains hidden information. With the right learning techniques, highly useful analytics can be developed."
Pandit cites the example of a global provider of wireless controllers and access points that must manage access points across companies, schools, consumers, and other customers. "There are all kinds of scenarios that happen when thousands of devices must be managed for a diversity of customers," said Pandit. "Some customers will tell you that their Wi-Fis aren't working -- or perhaps that their Wi-Fi applications are rebooting every five minutes."
Harnessing data generated by machines and then developing effective analytics can cut operational costs and efforts by IT support staff. True enough, an IT technician can solve a customer wireless issue by having the end appliance rebooted 20 to 30% of the time, but then there is the other 70% of the time, when the problem can't be solved without analyzing log data to get at the bottom of the problem.
"These are the situations when tech support has to start digging," said Pandit. "Has there been a configuration change? Has there been a machine change? Is there a difference in the traffic pattern? All of this information is stored at the access point log data. An engineer has to pull up this data and then look through screens of it. It is a tedious and time-consuming process."
Pandit's company applies machine learning to the network endpoints. The technology ingests endpoint-generated data, weeds out unnecessary data, and then runs the data through multiple rule and alert algorithms. Data is also processed iteratively to detect false positives and negatives based on known machine behaviors. A final set of data is distilled into a dashboard that a network technician receives for purposes of problem diagnosis and drilldown. For the technician, a complicated issue that could take several hours to resolve can be reduced to minutes -- this improves mean time to repair and customer satisfaction.
Improved performance and customer satisfaction are critical for the telecommunications and wireless industry, where nearly 75% of subscribers signing up with a new wireless carrier every year come from another wireless provider and annual customer churn rates average between 10% and 67%. A 2014 customer satisfaction survey conducted by J.D. Power revealed that resolving a service issue the first time is the single most important factor contributing to customers renewing their contracts with their current wireless providers.
If new ways of harvesting and actionizing machine-generated big data with analytics can speed diagnosis and issue resolution, wireless carriers might gain yet another tool to assist in reducing churn while improving customer loyalty.
