Prevent Contact Center Fraud With Voice Biometrics

Fraud is a Real Threat

Fraud directly impacts a company’s finances and branding, not to mention customer loyalty. Yet fraud protection measures often increase operational costs and inconvenience customers.

Most organizations dedicate significant resources to prevent fraud online, but this leaves few resources for call center security. A contributing factor to this is that companies focus their call center(s) on providing customer support, not preventing fraud.

Yet, as fraudsters become increasingly sophisticated many have mastered today’s common security measures, such as security questions and passwords. They’ve also mastered call centers, leveraging information gathered from sources like social media or manipulating helpful agents with weak authentication methods at their disposal.

The prevalence of this type of fraud shows that knowledge-based authentication (KBA) does not go far enough to protect consumer account information. Furthermore, KBA is costly, time consuming, and frustrating for callers who have to constantly repeat personal information to access their account.

To Voice Bio, or Not to Voice Bio, that was the Question

Voice biometrics and phone-printing technology, however, help prevent fraud in the contact center while simultaneously balancing customer service needs, resulting in a minimally invasive authentication process.

In order for voice biometrics to work effectively in this manner, baseline data is necessary for each customer. The question then becomes how to obtain this data without inconveniencing every last customer with process of “registering” their voice. Getting the technology to work is one thing, but getting thousands of customers, potentially, to comply with the request can be a challenge in its own right.

When Plum first started researching voice biometrics several years ago, the roadblock to implementation was the voiceprint enrollment process. In some instances, that process was so arduous that it required callers to recite passages from Shakespeare to properly train the engine to recognize a caller’s voice.

Even after the engine was supposedly trained to recognize a voice, our testing showed a high number of false positives when attempting to authenticate using a voiceprint. At that time, the negative effect on customer experience outweighed the potential to provide account security.

Voice Bio Grows Up

Fortunately, technological advances now allow many voice biometrics providers to recognize a caller’s natural voice pattern without having the caller first spend time registering their voice. This, and other improvements to voice biometrics make it a viable replacement for KBA authentication because it balances security and usability, acting as a strong failsafe against fraudsters while preventing caller frustration.

How has voice biometrics bypassed the need for time-consuming enrollment? There are two different approaches: 1. Voice biometrics can “actively” authenticate a user through a simple pass phrase, or 2. The software can “passively” analyze a caller’s voice patterns through natural speech.

The passive method of voice authentication is primarily used in the contact center while a caller is speaking with an agent. The active mode is most commonly used in IVR or mobile applications and requires a user to enroll by stating a simple passphrase.

Passphrases can be text-dependent, requiring the user to say an exact phrase, or text-independent, where the caller can say whatever they want. Neither the passive nor the active method of user voice authentication requires training anymore.

Both methods can provide a high level of security through analysis of hundreds of unique voice traits. Using voice biometrics instead of KBA lowers call handle time, provides a more reliable layer of security and improves the caller experience.