As a business leader, it’s important to do everything you can to prevent and reduce identity theft, which can otherwise lead to financial loss for you, your organization and your consumers. But sometimes, the go-to approach can create a frustrating customer experience.
As the president of an information technology company, I suggest using artificial-intelligence-based behavior tracking—which uses biometrics and contextual intelligence to judge the risks involved in a transaction—to help prevent identity theft. This type of technology helps enterprises assess the nature of a transaction and determine which authentication levels are needed to successfully authorize the purchase.
Enterprises face a lot of pressure regarding data privacy, so they are forced to set up a stringent process to validate a genuine user. This can lead to a lot of friction and unpleasantness, which can then lead to compromising the customer experience. Today’s customers often face two kinds of problems when they seek assistance from the contact center or customer support at a bank, telephone company, etc.:
First, they need to identify themselves and prove that they are who they claim to be. Second, irrespective of the nature of support or risk associated with the transaction, they need to go through the same process of proving who they are any time they call.
For example, imagine you’re calling your bank: No matter the amount of money in your account, you’ll likely be asked questions to verify who you are, such as your mother’s maiden name, pet’s name, what college you went to or even what the last transaction on your card was.
Similarly, whether you want to know your latest account balance or you want assistance transferring $100,000 from your savings to a checking account, the set of questions or verification process could be the same. All this can lead to frustration and even compromise security, as personal information is often easily available on social media or through social engineering methods imposters follow.
How would a new authentication process work?
Enterprises are embracing new technologies, such as AI-based behavior tracking and a composite risk-based authentication with biometric authentication as a tool to enhance customer experience with “passive” detection and authentication of user credentials. (Full disclosure: This is a service my company offers.)
It is clear from the words that, using contextual intelligence, the authentication process and necessary authorization for a particular transaction is dynamically assigned. For example, if someone asks, “What is my balance?” you might just do a PIN-based authentication. However, if the same user is asking for a fund transfer, you’d use voice biometrics and/or an additional process based on the risk of the transaction.
Contextual intelligence can be based either on user behavior or on the risk of a transaction. Sometimes, for information-related services, it’s best to use the behavior-based approach. In the case of a transaction-based authentication, the nature and quantum of risks involved with the transaction need to be weighed, so a multifactor authentication including biometrics should be used. This is called RB-MFA, and AI-based behavior tracking and biometrics are key to this.
Voice biometrics is emerging as a solution for AI-based behavior tracking and composite risk-based authentication. Unlike just a match on whether something is “true” or “false,” which is a binary result from other biometrics systems, a “confidence score” is the output of the voice biometrics system, so one can do RB-MFA with voice with ease.
For example, if the confidence score of a customer’s answer is in the range of 400-600 and the information or service sought by the customer is classified as “low risk,” you can pass the customer with the first level of authentication (i.e., when they describe their problem or request to an agent). But, if the transaction or service is classified as “high risk,” it’s important to increase the threshold to 750 and seek additional voice inputs with additional questions to the caller.
An enterprise can very easily adopt voice biometrics when a customer reaches out to their contact center for any service. Or, you can ask for explicit consent from customers and ask them to proactively enroll. Over a short period, regular users will get enrolled.
That said, it’s important to consider any roadblocks you might face if you’re considering implementing this tech. For example, a key challenge in implementing a voice biometrics system is collecting voice prints of existing customers and verifying that information with the original customer. For example, if an imposter gets enrolled in place of the original customer when the voice print is being created, how would the system know? This can be overcome by end user education about the benefits of voice biometrics and calling them on their registered mobile phone to make a verification call to complete the process of enrollment.
Another challenge is the use of speakerphones and noisy environments. User education is also key here to ensure that they speak from relatively noise-free environments and do not use speakerphones.
It’s also worth noting that passive behavior tracking using AI is another area under development by a few vendors. Many parameters of smartphones, including geolocation tracking, the way you swipe the screen, the pressure applied while typing, your style of walking and talking, etc. forms the part of this AI-based behavior tracking system. This can complement voice biometrics to ensure a reduction in fraud.
In summary, I believe AI and voice biometrics can play a big role in helping to improve the customer experience and secure their transactions. It’s relatively user-friendly, and user education can help you overcome any challenges you face, should you choose to implement the technology.