Call monitoring gets emotional

Telecommunications Americas, Dec, 2004 by Bridget Mintz Testa

"Poor service." Competitor names. "Cancellation." Terms and phrases like these don't gladden the hearts of telco customer service managers charged with retaining subscribers. But up until now, call monitoring software at telco call centers couldn't identify such charged terms. Recording calls for "quality purposes" has been mostly about legal compliance and agent performance, not about customer satisfaction. According to Bar Veinstein, director of product marketing for call-monitoring giant NICE Systems (Ra'anana, Israel), call centers randomly sample five or 10 calls per month per agent. So quality monitoring of calls really hasn't helped customer service managers step in to deal with dissatisfied subscribers.

Enter NICE Systems' Perform software, which adds speech analytics technology to the company's call monitoring suite. Perform's speech analytics uses proprietary algorithms to detect things like significant words or phrases such as competitor names; emotions via parameters such as voice pitch and volume; and talk patterns such as lengthy pauses and agents interrupting customers (or vice versa). "The software flags 'interesting' interactions and generates a report," Veinstein says. The report goes to designated managers, who can respond as quickly as the company review process allows. That could be in near-real-time, Veinstein says, but it won't be instantly.

Perform's speech analytics goes beyond customer satisfaction. Veinstein says the software can detect customer intent and note conversations involving products and competitors. It can even be used to evaluate marketing campaigns. "Perform turns call-monitoring into a business solution. It lets you improve processes and products for that purpose," he says.

Quality monitoring, including speech analytics, is a segment of "workforce optimization technologies." According to London-based Datamonitor WOT analyst Tom Pringle, besides quality monitoring, the field includes agent analytics (evaluation of agent performance), e-learning (agent training by listening to recorded calls) and workforce management (monitoring, forecasting and planning call center staffing levels). However, speech analytics, and especially emotion detection, are both very new, according to Pringle. "The usefulness of speech analytics is to bring out those calls that really need attention," he says. "It allows those call centers' supervisors to pay attention to calls, see what's wrong, and try to put it right."

[ILLUSTRATION OMITTED]

Telcos, with their large call centers, are in a good position to deploy brand-new technologies like speech analytics, Pringle says. Although Veinstein wouldn't name customers, he asserts that telcos and financial companies are the top two leading early adopters of Perform. "Telcos are really looking at this because of tough competition," he says. "They're looking for an edge over their competitors."