The 2017 Speech Industry Implementation Awards: FirstEnergy
Product: Convergys Natural Language Understanding IVR Platform
Akron, Ohio–based FirstEnergy supplies electricity to more than 6 million people across Ohio, Pennsylvania, New Jersey, West Virginia, Maryland, and New York. Since 1997, when the company formed following the merger of Ohio Edison and Centerior Energy, its electric systems have connected more than 24,000 miles of transmission lines from eastern Ohio to New Jersey’s Atlantic coast. Equally important, its customer service centers have connected more than 9 million live-agent calls annually.
Unfortunately, though, for a long time its technology struggled to keep up with the volume. Its old interactive voice response (IVR) system used directed dialogue, offering limited prerecorded selections that frustrated customers with their generality. Close to half of all callers were failing to provide their call reason at the IVR main menu, while 36 percent asked to be put in a default agent queue, resulting in far too many calls going to the wrong agents. Customers were frequently transferred multiple times, wasting time and creating confusion.
“Our goal was to get those calls to the appropriate agent to reduce that agent-to-agent handle time,” says Mary Alsayegh, FirstEnergy’s IT project manager.
Alsayegh, along with project leads Jamie Janosky and Brian Irons, knew the company needed to update its outdated directed-dialogue IVR with a more user-friendly experience. They turned to Convergys, which had provided FirstEnergy’s previous IVR system. Convergys’s suggested implementing its natural language understanding (NLU) application, which did away with preselected prompts and allowed callers to communicate their issues in their own words. “We did look at other vendors, but decided to stay with Convergys because of our successful relationship in the past and because they had the technology we wanted,” says Alsayegh.
FirstEnergy quickly upgraded its entire system to NLU, thoroughly altering its customer service model. The changes began at the main menu, where the system simply asks the customer “in a few words how can I help you today?” NLU can recognize any response, from something as general as “outage” to something as specific as “What’s my account balance?”
Using directed dialogue, callers might have had to go through four menus before arriving at service. Natural language immediately places them with the appropriate agents or routes them to the proper menu for self-service.
The revamp, which took place late in 2016, produced quantifiable results in short order. The conversational automation inherent within NLU made self-service rise by 11 percent, lowering the average amount of time agents spend on the phone. The transfer rate from agent to agent was reduced, as the customer was properly identified more often. Most importantly, the volume of calls being routed to the default agent queue slashed by 58 percent and the total number of default calls fell by 36 percent. “To give you some perspective on those numbers, we only anticipated reducing [default agent queue calls] about 25 percent to 30 percent,” Janosky says.
It also turned out to be a profitable move on FirstEnergy’s part. The company had predicted seeing a return on investment (ROI) two years after the overhaul, but actually achieved it in only one.
The rebuild also improved workflow and communication at FirstEnergy’s call centers. “The director of the contact center is ecstatic over how well the system is performing,” Alsayegh says.
While FirstEnergy is open to expanding customer service with other technologies like chat and text, the immediate focus is to expand and refine its implementation of natural language. One goal is to provide agents more information about every incoming call, such as account numbers and customer history. Long term, it’s also looking at other call flow areas besides customer service where the technology could be implemented. “Right now our IVR deals with customer issues,” Janosky says. “Next step would be to expand it to begin to provide products and services.”
Using Convergys’s natural language understanding IVR platform, FirstEnergy saw the following results:
• a 36 percent decrease in default call types within the IVR;
• a return on investment within one year when original predictions suggested two;
• a 58 percent reduction in calls going to the default agent queue;
• handle time avoidance of 60 and 109 seconds per default call avoided, depending on customer issue; and
• an 11 percent increase in self-service call containment.