Opportunity

Allianz Partners has some 800 associates who assist customers with general benefits inquiries, first notice of loss, policy cancellations, and claim status. It uses Verint Speech Analytics to transcribe approximately 96 percent of these customer calls.

The company had around 50 “industry-standard” call categories in place for traditional deep-dive call analysis, which it conducts to understand trends to guide improvement strategies and associate coaching. A big focus is on mining calls to uncover areas of customer confusion or frustration.

“From the call analysis we were doing with Verint Speech Analytics, we understood that transcription accuracy in part was based on how clearly the technology comprehends the voice of our associates,” says Nate Ford, speech analytics analyst at Allianz Partners. “We realized we could leverage the software to provide data points to quantify associates’ speech clarity, which can impact customer experience. If agents can’t be understood, calls can quickly deteriorate.”

To identify a sample of calls to review, Ford worked with contact center managers to compile a list of words deemed difficult to pronounce and which are critical to associates’ explanation of insurance coverage and benefits to customers. The list included terms such as determination, limit of liability, payable, physician statement, receipts, reimbursement, underwriter, uninhabitable, and unintended.

Man and woman walking with business suits on

Measuring Speech Transcription Accuracy

Next, a call category was created to flag calls missing these terms. The intention was to understand instances where the terms where actually spoken by associates but not recognized by the speech transcription engine.

Ford then reviewed and scored random calls from the top and bottom 25 percent of associates. In doing so, he determined there was a strong correlation between associates he found difficult to understand based on his subjective review and associates where speech transcription accuracy was poorer.

Ford also found that transcription accuracy (associated with accent) was only one of several factors impacting Net Promoter Scores (NPS) and Agent Satisfaction (ASAT) scores, but that speech clarity had a much greater bearing with complicated call issues.

Group of people talking around a table

Solution

Allianz Partners has some 800 associates who assist customers with general benefits inquiries, first notice of loss, policy cancellations, and claim status. It uses Verint Speech Analytics to transcribe approximately 96 percent of these customer calls.

The company had around 50 “industry-standard” call categories in place for traditional deep-dive call analysis, which it conducts to understand trends to guide improvement strategies and associate coaching. A big focus is on mining calls to uncover areas of customer confusion or frustration.

To identify a sample of calls to review, Ford worked with contact center managers to compile a list of words deemed difficult to pronounce and which are critical to associates’ explanation of insurance coverage and benefits to customers. The list included terms such as determination, limit of liability, payable, physician statement, receipts, reimbursement, underwriter, uninhabitable, and unintended.

Man sitting in an airport looking at iPad

Benefits

While the findings indicated that an accent neutralization initiative would be beneficial to customer experience improvement efforts, Allianz Partners recognized associate speech clarity was one of many factors impacting customer satisfaction. To this end, the question became to what extent accent neutralization was needed.

“We employed a data-driven approach to understand how much emphasis we should place on accent neutralization,” Ford notes. “We correlated it to our customer satisfaction goals to pinpoint the appropriate amount of associate coaching needed.”

Ford matched transcription accuracy rates with post-call survey data containing customer satisfaction scores. With this data in hand, the company administered focused accent neutralization coaching to a subset of associates – those who had the greatest amount of missing words or phrases in their customer call transcriptions.

Group of people working around table on laptops

Results

The performance of these associates was then reviewed over a three-month period. Associates who had repeatedly scored in the bottom 25 percent reduced missing information in their transcriptions by an average of 4.3 percent at the end of the three months. With the transcription rate improvement, the company also saw a rise in customer satisfaction scores.

“Verint Speech Analytics is an adaptive tool that can be used in so many different ways – from helping understand the big picture in terms of trends to uncovering opportunities for associate coaching,” concludes Ford. “In this case, it gave us an objective data-driven means to evaluate associate performance and deliver targeted coaching, providing a novel approach to accent neutralization.”

Two men having a conversation

Verint Solutions Featured in This Success Story

Featured Customer Success Stories