Blogs

Predictive analytics: The key to reducing outages and increasing customer satisfaction

Technical Writer

Customers are not pleased when they can't make a call or their Internet is down. What if you could foresee an outage and prevent it from happening? Using predictive analytics, telecoms can monitor service patterns and predict service interruptions. The goal? To fix issues before the outage even occurs.

Predictive analytics allows telecoms to analyze real-time data against empirical data to flag triggers and situations that typically proceed an outage and monitor the network for triggers.

Telecoms that successfully use data analytics to address outage issues will have a competitive edge in the industry. The American Customer Satisfaction Index (ACSI) uses the "ability to keep service interruptions and outages to a minimum" as a benchmark in its Fixed-Line Customer Service Index. The industry is currently trending toward lowered customer satisfaction in that area, recent ACSI research has found. In 2014, 82 percent of fixed-line customers were satisfied with their provider's ability to keep outages to a minimum; in 2015, only 76 percent were satisfied.

The Internet of Things (IoT) revolution has put a strain on telecom networks and made preventing outages an even higher priority. As Don DeLoach analyzes in Data Informed, the IoT explosion has dramatically increased the load on telecom networks. Verizon's "State of the Market: Internet of Things" report predicts that the number of connected devices will increase from 1.2 billion in 2014 to 5.4 billion in 2020. That puts a tremendous strain on the telecom networks and impacts service calls, network performance and outages. "Seven days of call data records (CDRs) in 2008 would be a mere fraction of seven days of CDRs recorded today," writes DeLoach.

https://kapost-files-prod.s3.amazonaws.com/uploads/direct/1453746205-30-0340/predictive-blog.jpgClient-side analytics often provide the key to outage prevention

Telecoms often simply use the data they have on their network. But that only paints part of the picture, especially when it comes to usage patterns that help predict outages. According to RCRWireless, 70 to 80 percent of content goes through Wi-Fi carriers that rely on network-side data and not client-side data. Client-side data gives insight into the behavior of people using the network, such as increases in dropped calls or people accessing specific data-heavy websites. When telecoms collect both types of data, they know what their customers are doing as well as how the network responds.

"Client-side analytics provide carriers with a granular, 360-degree view of user activity, offering insight into the best ways to improve services and optimize user experience," RCRWireless reports. "Network-side analytics, in turn, provide a bigger-picture view, showing the best places for offloading data to Wi-Fi and the ideal locations for small-cell placement."

Since privacy is often a concern with client-side data, stripping out personally identifiable information (PII) to make individuals unidentifiable is a common strategy with big data. CSO Magazine reports that there is a less than 1 percent chance of a person being identified if de-indentification is done correctly, according to Daniel Castro and Ann Cavoukian, lead authors of a white paper on the topic sponsored by the Information and Privacy Commissioner (IPC) of Ontario, Canada.

Use predictive analytics to take action

Identifying that a problem is likely to occur is the first step, but you still must pinpoint the specifics to fix the issue, often very quickly, to prevent loss of service. DeLoach says that telecoms must find the nature, location and cause of the issue to address the problem. Telecoms can then use data tools to analyze data sets such as device types, operating systems, browsers, applications, cell tower locations, communications technology (3G, 4G, LTE) or with a specific type of transmission, such as voice, data or video, says DeLoach. Once the cause of the problem is identified, the telecom can then correct the issue before outages occur.

Instead of having to wait for the network to go down or predict outages based on anecdotal evidence, use data to know when your network is about to go down. Your customers will thank you.

Learn more about IBM solutions for telecommunications.