Nice Systems' Nice Perform Voice Analyzer Picks Up Clues

This app hears nuances in call center recordings that can be used to improve service quality.

Did you ever stop to wonder about the sheer volume of time we spend talking to customer service reps in call centers? According to Incoming Calls Management Institute Inc., there are 50,600 call centers in the U.S. containing 2.86 million agent positions. Add another 4,500 call centers in Canada (212,000 agent positions) and probably another million or so agent positions overseas devoted exclusively to the U.S. market. Let's say there are 4 million agents each talking to customers 1,000 hours per year -- that's 4 billion hours. Imagine if that data could be captured and mined effectively.

Enter Israeli company Nice Systems Ltd. and its Nice Perform application, which analyzes voice recordings (typically retained for quality control) and combines the information with a wealth of other data, including voice over IP, e-mail, chat and agent screens.

"The drivers for customer behavior are increasingly hard to get due to limitations placed on outbound communications by legislation such as the [Federal Trade Commission's do-not-call] list," says Seema Lall, an analyst at Frost & Sullivan Ltd. "Nice Perform offers enterprises the means to tap into the hidden information gold mine within agent/customer-interaction recordings."

Lall says creative application of this technology could significantly affect the bottom line. The word-spotting and data-mining capabilities, for instance, could be used to monitor compliance with regulations such as the do-not-call list and avoid hefty penalties. Contact centers could harness the data to improve quality management, raise productivity, conduct root-cause analyses and enforce agent adherence to scripts. Alternatively, sales and marketing could use it for sales planning and monitoring, or to prevent customer defections.

"Nice Perform analytics turn millions of customer interactions into structured data that can be utilized by existing business intelligence and customer intelligence solutions," says Yoel Goldenberg, director of contact centers and enterprise solutions at Nice Systems. "This dramatically shortens the ROI period for these solutions, makes them more efficient and means that many more people in the enterprise can take advantage of them."

Nice Systems already had a background in word spotting, emotion detection and speech recognition, so this product grew out of existing expertise. The entire development was undertaken in-house. It's built on Microsoft Corp.'s .Net technology.

37Hnetscout_chart.jpg
Nice used extreme programming methodologies within a structured source management environment, with regular code reviews. Along the way, developers ran into some hurdles - primarily as a result of the great variety of customer environments that had to be accommodated. For example, Perform had to work with a wide range of telephony switches, computer/telephony integration (CTI) servers, storage management systems, firewalls and more. This led to one major technological change during the process -- a move from ASP .Net to smart-client technology, which improved functionality and simplified deployment.

Another stumbling block was figuring out how to filter out the music that customers hear when they're on hold or being transferred to another agent. "The speech analysis engine often interprets music as a new speaker or as a very emotional reaction for the current speaker," says Goldenberg. "In order to eliminate this problem, we included the CTI indications in the analysis process so that the engine doesn't analyze on-hold segments."

The product has been on the market for about a year, and minor enhancements are planned for next year. According to Lall, Nice provides a comprehensive analytics system. She says speech analytics products such as Nice Perform hold great promise for enterprises looking to harness the contact center as an effective source of customer information.

"Speech analytics products are poised for immense growth over the coming years as enterprises get better educated on the capabilities of these types of products and utilize them creatively to derive business value," says Lall.

Robb is a Computerworld contributing writer in Los Angeles.

Special Report

horizon_special_report_imag.gif

Computerworld Horizon Awards

Stories in this report:

Related:
5 power user tips for Microsoft OneNote
  
Shop Tech Products at Amazon