Ann All spoke with Carlos Navarro, chief markeing officer for Elavon, a US Bancorp subsidiary that offers payments services to 1 million customers worldwide. Elavon uses speech analytics technology from Verint Systems to mine recorded calls coming into its call center to identify what it calls "closure triggers." These verbal cues, words or phrases have helped it rescue more than 2,000 at-risk merchant accounts. The algorithms inside Verint's speech analytics application, which at Elavon include a library of more than 500 words and phrases, make correlations to help identify these at-risk accounts.
"It's not an application where you put it in and you're done. It's very much a tool to be used as part of your customer lifecycle management, which is part of your overall customer experience. It's another tool in your overall CRM arsenal."
All: What specific call center issues did you hope to address using Verint's application?
Navarro: I knew we could improve our overall customer experience. I had been associated with these types of solutions in the past. First and foremost, we are using the solution to address the customer experience. We were looking for particular phrases and sentence structures that wouldn't necessarily pop up on a traditional customer service end-of-call survey or ranking by a CSR (call center representative). We started with a base library for our industry, but we tweaked that library and added to it to make it very specific to the nomenclature our customers are used to hearing. We build the library and then can add and subtract from it on an ongoing basis. It's a living, breathing organism, in that we add things as topics and issues come up in our industry.
Two years ago we weren't searching for "Durbin" or "Credit Card Protection Act" or "PCI." It can be fairly complex. So for example: If someone says a certain word within a certain proximity to another type of word, or early in the call, or late in the call, it might signal that we need to call the customer back.
While we look for emotion and certain words that can't be repeated, that's not our primary focus. Those are too obvious. We are looking for certain phrases that we felt needed attention, based on our interviews with current merchant customers as well as ones that had left us. We built some sophisticated models based on those interviews.