Benefits of automating identity verification

There are many situations where simple means of authentication – such as a PIN or password – cannot be used, either because the caller doesn’t have one or has forgotten it. For most of these cases, caller authentication then needs to be performed by contact center agents, which results in significant costs to the organization. Automating this identity verification process can bring two key benefits to organizations:

• Reduce costs – It reduces the amount of time an agent has to spend per call. For large contact centers, this alone can produce multi-million dollar annual savings.
• Create opportunities – It enables the introduction of self-service applications requiring caller authentication.

Leveraging its unique expertise in speech application optimization, Nu Echo has developed NuID™, a high performance application that automates the identity verification processes used in most contact centers.

High accuracy automated caller identity verification

The key challenge in automating identity verification with speech recognition technology is to minimize automation failure while maximizing security. Because standard speech recognition approaches can break down quite rapidly, leading vendors recommend the use of “verification grammars”, dynamically built based on the expected answer. But even then, failure rate can increase quite rapidly as the number of security questions increases. In order to successfully address this challenge, Nu Echo has spent over three years developing fundamentally novel —and patent pending — approaches and technologies that today enable NuID to deliver truly superior performance.

Remarkable field performance

NuID’s remarkable performance is eloquently illustrated by the chart on the right, showing the actual results obtained in a recent deployment at a large government agency. The verification process used the four security questions shown above. Of all calls processed by NuID, only 5% were incorrectly rejected (including zero-outs), while greatly exceeding the agency’s requirement of a maximum per-question false accept rate of 0.5%.