Allianz: Speech Analytics Identify Call Center Employees for Accent Neutralization Training

Article Featured Image

Company: Allianz Partners

Allianz Partners, owned by Allianz is one of the world's largest insurance and assistance companies. One of the company’s signature products is travel insurance, and it has recently branched out to provide insurance for event tickets. With 16,000 employees across 34 countries, Allianz Partners is truly global.


Business Challenge

One of the biggest challenges of call-centers is making sure customers have a good user experience—and more of than not, that comes down to the interaction they have with your agents. If there is a language barrier—or even if the agent has a heavy accent that your majority of callers have trouble understanding—that can cause user satisfaction problems. Allianz wanted to find a way to solve that problem.

Vendor of Choice: Verint Systems

Verint understands the importance of data and focuses on helping deliver actionable intelligence to its partners and customers. According to its website, “With Verint solutions and value-added services, organizations of all sizes and across many industries can make more informed, timely, and effective decisions.”


The Problem in Depth

Allianz Partners works with a sister company in Mexico to handle some of its call-center needs. According to Nate Ford, speech analytics analyst at Allianz, at his last company, “The associates were easier to understand than callers who had more diverse accents and languages.” So communication difficulties can go both ways. While you can’t do anything about who is calling you, you can help make the associates in your call centers easier to understand to the majority of your customers.

“Decreased communication leads to poor customer experience,” Ford says. But it wasn’t all about the customers. Keeping associates more engaged and productive was also a priority for Allianz. “It’s sort of obvious that if you have trouble understanding someone the experience is not going to be as good,” he adds.

Allianz was already working on “accent neutralization,” as the company calls it. But Ford thought he could make the process more efficient.

“It clicked with me that I could use analytics to quantify the agents who needed help speaking clearly,” says Ford.

The Solution

Luckily for Ford, he didn’t have to go searching for software to help him with this new task. Allianz was already using Verint Speech Analytics. Here’s the system he devised to help identify the associates most in need of coaching:

First, Ford takes the desired population, in this case, the travel services associates in Mexico. He pulls all of their calls within the Verint software from within a given time period—usually a month. He excludes outbound calls, calls under three minutes, and any associates that have less than 100 calls within the population.

Next, Ford pulls a report to find out how many calls each associate has that have not been categorized. If a call isn’t categorized, it’s usually because Verint’s speech analytics couldn’t understand the call well enough to add that data. “Every call should have at least one category attached, so if it’s not, speech just didn’t understand the terms with those calls,” says Ford.

“I then see what percentage of an associate’s calls have no category attached,” Ford says. If Agent A had a 100 calls, and 20 were missing a category while Agent B only had five uncategorized calls, it begins to come clear who can benefit from accent neutralization.

It isn’t all left up to the machines. Ford tests the veracity of the results by listening to three random calls from the associates he’s identified and then ranks them from 1-5 on how well he was able to understand the call.

The Outcome

Just landing on Ford’s list doesn’t always mean you go straight to accent neutralization training. “I’ll color code top 25%, the middle 50%, and the bottom 25%,” he says. But he also has a separate list of names of associates who have been in the bottom 25% for three months in a row. He says the list usually has between 10-15 people on it.

Those associates who were so often near the bottom are important to monitor the success of the accent neutralization program. When Ford looks at their transcription rates after 3 months, they have improved 4.3%. The training tends not to make a difference in the general population of travel associates, because new associates are constantly being on-boarded.

This is just one of many programs in place that Allianz uses to help improve customer experience, so Ford isn’t comfortable attributing any enhanced customer satisfaction ratings to just this program. That being said, Ford is working to recalibrate the process to work with Verint’s 15.1 release of Speech Analytics. As the system gets better, Ford has to reimagine how he uses it.

SpeechTek Covers
for qualified subscribers
Subscribe Now Current Issue Past Issues
Related Articles

Higher Education Growth: Improving Quality Assurance with Call Recordings

HEG agents often have phone conversations with students. These calls play an important role in enrollments, but agents must adhere to strict Telephone Compliance Protection Act (TCPA) compliance guidelines and processes for efficiency. There are many opportunities for errors and compliance missteps throughout the EDU lead lifecycle — particularly during phone calls. The company needed a more efficient way of dealing with this challenge.

NWEA: Speech Recognition Technology Evaluates Out-Loud Reading for Students

The NWEA sought a way to streamline oral evaluations to save school districts time and money while still ensuring students are on track.

Bluegreen Vacations: Using Speech Analytics to Improve Customer Experience

When its customer service call volume had unexplained spikes, the resort vendor turned to CallMiner for answers.