Is Voice Biometrics Right for You?
cases that investment can be recouped within the first six months.
On the other hand, Miller maintains that there isn't any more of a downside than there would be for a single factor authentication method. "All authentication can be defeated," Miller says. "The bad guys are usually ahead of the good guys in terms of fraud. What we're learning is that you don't want everyone to be a single factor zealot, so it's not a downside to voice, it's a downside to any single factor solution."
Khitrov stressed that, while the technology itself works well, "the pitfalls that we see usually fall [into] two categories—integration and communication. It is very important to start the project with clear implementation and integration plans."
Tips for Successful Deployments
While voice biometrics experts are telling the enterprise, "Come in, the water's fine," some companies remain hesitant. The following recommendations for successfully deploying voice biometrics may help more reluctant companies.
Nuance posits that the first step is to set clear business objectives that align with the organization's overall customer care strategy. This includes key performance indicators (KPIs) that can be measured pre- and post-deployment.
"Companies should evaluate customer experience metrics, such as internal metrics like Net Promoter Scores, to see if authentications are a major factor in the customer experience," Beranek says.
Mark Lazar, global vice president of identity and fraud solutions at Verint, says that companies should establish a baseline by measuring repeat fraud instances. "While many contact centers suspect that professional fraudsters are rampant, they can't always quantify the problem," Lazar stated in an email. "By conducting a voice biometric analysis across a small sample of fraud calls, an organization can quantify the problem to find eye-opening results."
Additionally, Khitrov recommends deploying a proof of concept, a pilot to a small group of users before thinking of going to live production. He suggests that it's important to clearly identify the use case and requirements, refine them, and understand the legal considerations.
Following these steps, it's important to refer back to your objectives and KPIs at every step of the project to make sure you don't veer off track. "The objective may be to improve the self-serve customer experience, and the KPI is Net Promoter Score," Beranek says. "In this case, the voice biometrics system will be designed very differently than if the objective is to eliminate fraud, and the key KPI is dollar losses due to perpetrated fraud."
A voice biometric solution also requires tight integration to back-end systems and likely further development. It's critical that a vendor has experience working with a company's back-end systems and/or has professional service resources that can do the work if a company can't.
"Voice biometrics is very different from knowledge-based authentication technologies, such as passwords, PINs, and knowledge questions," Beranek says. "It's key to bring on board any voice biometrics initiative resources that understand the intricacies of voice biometric technology."
Additionally, companies should be sure to maintain high accuracy rates as a voice biometrics system scales. The accuracy is measured not only by the detection rates (correctly recognizing a fraudster), but also by false positive rates (mistaking a customer on the phone as a fraudster), according to Lazar. "The system is successful when it can maximize detection without too many false positives," he stated. "Field results show that system accuracy is greatly increased by including other factors in the scoring process. Adding metadata to the voice biometrics scoring process is an important strategy for a company to maintain high accuracy at scale."
Companies should also consider wrapping a PR campaign around a launch, which could including giving the biometric authentication system a special name, Khitrov stated. This could include advertising the new system to customers by telling them the benefits of voice biometrics. "We can't stress enough the importance of communicating the value of this great technology to the customers. People don't like change, and we need to explain to them the benefits of the new approach to authentication." Otherwise, he indicated, adoption rates will never hit the target.
"Be prepared to work across the organization," Miller says. "There's some marketing that has to be done [as well as] conditioning, but once they start rolling out, they'll have the tiger by the tail."
Even from day one, Nuance maintains that companies should plan for a voice biometrics implementation for multiple channel deployment. This will ensure that not only will the right technology be selected, but also that all design decisions are not made in isolation. For example, a company can decide how voiceprints are collected and on the wording of passphrases, which can enable one voice biometrics system to be leveraged in different contexts. "Ultimately, this will minimize the financial impact on an organization, enable a speedy implementation of voice biometrics across the organization, and provide a consistent customer experience across multiple customer-facing interaction channels," Beranek says.
While companies are making sure that their voice biometrics systems are bullet-proof, Khitrov stated that the number-one mistake organizations make is trying to conceive the ultimate security system. "The key is to refer back to the business objectives and at each step of the design process, ask if the decisions being made are aligned with these objectives," he said.
"Some organizations have created such robust enrollment and verifications processes that they lead to low adoption of voice biometrics," Beranek says. "As a result, these organizations were unable to improve security [or] improve the customer experience."
Finally, as with every implementation, voice biometric systems need to be monitored and tuned. Regular analysis and turning can dramatically improve detection rates, Lazar says, because of the many parameters involved in a configuration. "With access to this information, analysts can regularly inspect data to gain unique insights into fraud trends, and new attack vectors can benefit the contact center," Lazar says.
The Future Is Catching Fire for Everyone
As voice biometrics technology is gaining a foothold in the enterprise, the future will see small businesses looking to differentiate themselves in their space. Nuance maintains that there is a business case for smaller organizations to bring the technology to their customers via intermediary companies that can offer it as a service. For example, voice biometrics firms are already beginning to offer the technology as a service to smaller companies, such as credit unions.
"They don't have to make a massive investment and can use a pay-per-use model," Beranek says. "That model is becoming more and more prevalent with third parties and there's going to be many more of them offering this as a service."
Staff Writer Michele Masterson can be reached at mmasterson@infotoday.com.
Related Articles
Solution offers new personalization feature, Nuance Voice Print
03 Apr 2014