The Benefits Of Voice Biometrics

Virginia Credit Union's voice identification deters fraud and boosts call efficiency.

Top-Level Takeaways

  • Virginia Credit Union added voice biometrics in 2016 to improve authentication for contact center conversations.
  • The technology has helped reduce handle times and weed out fraudulent actors.

In 2015, Kate Hopson read a troubling statistic: One in every 500 calls will be a fraudulent attempt by 2016.

For the contact center manager at Virginia Credit Union ($3.6B, Richmond, VA), the number was concerning. At that time, Hopson’s operation was fielding approximately 55,000 calls per month, and the credit union’s authentication process was disliked by members for being both laborious and long. On average, it took members 90 seconds to authenticate. But with fraud on the rise, VACU wanted to ensure it was protecting its members. Could it do so while providing an excellent experience?

Those processes were not efficient or well received by members, Hopson says. We needed to shift how we authenticated members over the phone.

ContentMiddleAd

In looking for ways to improve the authentication process, VACU landed on voice biometrics as a solution and, in partnership with Nuance, rolled out VACU Voice ID in October of 2016.

CU QUICK FACTS

Virginia Credit Union
Data as of 06.30.18

HQ: Richmond, VA
ASSETS: $3.6B
MEMBERS: 271,528
BRANCHES:18
12-MO SHARE GROWTH: 4.3%
12-MO LOAN GROWTH: 6.0%
ROA: 0.81%

At that time, everyone was just starting to unlock their phones with their thumb, Hopson says. We liked the idea of authenticating using something no one can steal.

Natural Conversation Technology

Before VACU implemented voice identification, its automated member authentication process required members to answer several baseline questions including name, member number, date of birth, the last four digits of the member’s social, address, and passwords on file before they could speak to a member service representative. For medium- or high-risk transactions, members answered additional out-of-wallet questions and provided transactional information.

This information was meant to ensure a secure transaction, but VACU found the process so difficult and exhaustive that the member fail rate on out-of-wallet questions was approximately 20% an unusually high number, according to Hopson. In addition, VACU’s abandonment rate was upwards of 16%.

Now, members can opt in to voice authentication and submit a digital print of their voice by engaging in 30-40 seconds of natural conversation. The system measures 140 different criteria and attaches the voice to a corresponding member number. When members call the credit union, those who have opted into voice identification speak naturally for six seconds to confirm their voice belongs to them.

It’s based on natural conversation because we want the rhythm and sound of our member’s voice, Hopson says. But it’s not only the characteristics of speech like tone, timber, and pace we want to learn the quality of the hardware they’re using.

Before members can opt into voice identification, however, the credit union must still authenticate them. Rather than continue its lengthy legacy process, VACU looked for a better way. In the Commonwealth of Virginia, driver’s license numbers are not available in the public domain. Only state and local police and the Department of Motor Vehicles keep these numbers on file, which made them ideal for baseline member authentication.

We’ve found that more people have this number ready to provide us, Hopson says. Plus, since few organizations have it, it’s something that stops fraudsters each time.

Want More Innovative Ideas?

A subscription to Callahan gives your organization unlimited access to content like you see here. With it, you’ll never be short on inspiration to help your organization grow.

Security And Efficiency

As of second quarter 2018, VACU had more than 270,000 members. And whereas not every member dials the contact center with the same frequency, to date, 36,000 members have opted in to voice identification. And according to Hopson, the contact center’s most active callers are all opted in to the service.

The contact center team worked closely with the marketing department in the lead up to launch. Primarily, they wanted to alleviate confusion or worries among members who viewed voice recording measures as a ploy of Big Brother. For Hopson, clearly and constantly communicating the increased security features and shorter call times was a valuable tool to put members at ease.

Additionally, the marketing department, which has more member touchpoints than the contact center, was a key ally. VACU updated its website with voice identification information, added a message to its call hold system, ran an article in its monthly newsletter, created a two-and-a-half minute explainer video, and held untold phone conversations with members in the months before voice identification went live. Some promotional efforts to support the program have slowed, but the credit union and Hopson continue to actively work on enrolling more members.

VACU wants to enroll 40,000 members in the voice identification system by the end of 2018. Hopson is confident the credit union will reach that but wishes the number was higher.

The more people who enroll, the more secure the credit union will be and the more efficiently we’ll be able to run the contact center, she says.

So, in the next few months, the 65,000 monthly callers who are not enrolled in voice identification will have to enroll before conducting any transaction over the phone. There will be exceptions, Hopson says, but contact center employees received full training on the system’s capabilities, including how the system works and what to do in the case of a voice mismatch. Each new hire to the contact center in the past two years has received the same training. Hopson is confident her team is armed with the knowledge to explain the service’s benefits and overcome objections.

If someone strenuously objects, then we won’t enroll them, Hopson says. But the environment is right for us to protect our membership and get more members enrolled.

In nearly two years, voice identification has identified a voice mismatch and stopped 8,500 potentially fraudulent calls, Hopson says. Additionally, pairing voice identification with a few other contact center updates has resulted in handle times 37 seconds shorter than non-voice identification calls. The contact center’s abandonment rate now sits at less than 5%, with voice identification an important factor in that improvement.

But there’s room to improve. For Hopson, enrolling a majority of credit union members in voice identification will go a long way toward battling fraudulent actors and improving call efficiency.

If we can cast a wider net, it will help us prevent fraud, she says. It doesn’t help as much when you just have a portion of your membership enrolled.

 

April 26, 2019
CreditUnions.com
Scroll to Top