It’s a tug-of-war… in today’s economic climate, enterprise executives across business, security, and operations are faced with seemingly mutually-exclusive objectives of reducing costs and driving revenue, while striving to deliver a superior customer experience. What takes priority? Adding to this conundrum, industry analysts confirm the increasing consumer concerns about fraud and identity theft, leaving enterprises to answer these three key questions:

• What applications would benefit from additional security?
• Where do regulatory agencies dictate multifactor authentication?
• What applications require extensive and expensive agent or self-service challenge/ response authentication?

The solution that can successfully link enterprise security with consumer convenience is one with voice biometric authentication at its core. The voice authentication market is projected to grow at least five-fold from today’s level to the year 2011. Why the explosion in demand? Because voice authentication can help banks, card issuers, cable and telecommunications, healthcare, manufacturing/distribution, and technology companies cut fraud by more than 90 percent and improve service across both agent-assisted and self-service channels. 

Voice biometric technology has indeed become more robust, handling variations in channels (landline vs. mobile) and in individual callers' voices from call to call. Now, coupled with application best practices by implementing voice authentication as part of a multifactor authentication solution, this “total solution” combines elements of “what you know” (account number and PIN), “who you are,” (biometrics), and potentially, “what you have” (card, token, etc.). 

As we all experience, traditional authentication techniques are usually confined to the “knowledge” arena, prompting for an account number or user ID, plus a PIN or password. The drawback is that if somebody finds out both the ID and the PIN/password, then he can gain entry 100 percent of the time. (This is particularly manifested in mobile devices where account information is commonly stored with the phone number entry.) To overcome this inherent weakness, agent and Web applications increasingly use obtuse challenge questions like, “What elementary school did your spouse’s best friend attend?” In some high-security cases, applications require a secure token, but these are expensive, can be lost, and are generally inconvenient.

Where do we go from here…intelligent Voice Authentication

With these various authentication strategies in mind, the best approach is to establish a set of business rules that can be created, managed, and applied in real time as part of an intelligent multifactor voice authentication solution. More than 90 percent of the time, sufficient confidence is obtained with the voiceprint comparison; if, however, a voice authentication score is questionable, then rules should make use of consumer profile information, usage patterns, and transaction contexts to make a decision, thereby increasing security while ensuring that valid users are not prevented from accessing their desired information.

Ironically, voice authentication is not just for a speech IVR. It can be used in DTMF IVRs and agent transactions and to authenticate users who are logging into a Web site—triggering, for example, an outbound call to the user’s known phone number, and then passing back the necessary authentication to the Web application via the data channel.

Managing multichannel deployments across lines of business requires the need to surround the core technology with an enterprise deployment framework that benefits from a single administrative interface to apply consistent security thresholds, authentication logic, and integration with multiple platforms and back-end databases, as well as manage add/ deletes/changes across millions of customer accounts.

Recent announcements heralding new options for deploying voice authentication in an on-demand fashion represent a refreshing business model that uses a software-as-a-service (SaaS), pay-as-you-go pricing, which eliminates capital expenditure. Furthermore, a hosted model saves the enterprise from having to upgrade voice authentication software over time. Finally, an on-demand hosted option accelerates early-stage limited deployments to within a month or two, along with the ability to quickly scale to large volumes. This allows an enterprise to try it out without any large up-front investment.

Finally, enterprises should consider the following when evaluating voice authentication solutions.

Consideration #1. Consider an enterprise-wide, multichannel strategy from the start, rather than treating each application as a custom and independent effort, which requires far more effort to deploy, with no ability to leverage one application to the next. A hosted platform- and channel-independent solution should allow voice authentication to be deployed across IVR (DTMF and speech) platforms, as well as agent, Web, and mobile device applications.

Consideration #2. Be cautious when considering what phrase to use for the voice authentication utterance. An application-specific phrase (such as an account number) may be justified if voice authentication is used in only one application, but doing so requires a consumer to separately enroll in each application. Also, avoid phrases that contain account- or personally-identifiable information (PII). The benefits of short phrases like “my hero” are that they are easy to repeat back (enrollment is quick… about 10-15 seconds), and enable a higher degree of confidence than digits.

Consideration #3. Seek a solution where the platform does not require a specific voice biometrics engine. To ensure portability and compatibility across various IVR and Web platforms, the voice authentication enterprise platform should accommodate the option to update or swap-out the underlying voice authentication engine with best-of-breed technology without requiring customers to re-enroll their voiceprints.

Voice authentication can reconcile opposing forces of customer experience, security, and cost effectiveness. When used as part of an intelligent, multifactor authentication solution, voice authentication can leverage a single consumer enrollment in one channel and leverage that across the IVR, Web, agent, and mobile device environments. Furthermore, by adopting an on-demand model with transactional pricing, the capital expense hurdle can be overcome, and deployments can be easily rolled-out within a month or two, making it easier to start small and scale quickly, accelerating the return on investment for a solution that delivers enterprise-wide security in a customer-friendly fashion.

Paul Watson is general manager of agent, speech, Web, and mobile device on-demand solutions at Convergys.