- The costs of compromised security in the customer care environment are high to both the enterprise and the customer, and the occurrences of security breaches continue to grow briskly.
- Although not widely used technologies today, the combination of voice biometrics and predictive analytics has great potential to enhance fraud deterrence.
The methods of customer identification and verification used in contact centers today take too much time and are a major source of customer irritation. Agents’ questions inquiring about personal identification numbers (PINs) or asking pre-arranged security questions, such as “What is your father’s middle name?”, have outgrown their usefulness and are often easily circumvented by fraudsters seeking illegal access to customer accounts and private corporate information. High on the list of technologies destined to replace these traditional techniques are voice biometrics coupled with sophisticated predictive analytics.
To date, automated voice biometric technology, the use of pre-recorded voice prints to identify a caller, has been slow to catch on in the contact center because of high cost, extensive preplanning requirements and long implementation times. Defined as the science of mining and analyzing patterns in historical transactional and real-time streaming data to identify security risks and opportunities, predictive analytics is a relatively new technology just beginning to capture the attention of contact center managers as a security tool. Several leading-edge providers of these technologies, including Nuance, NICE Systems, Pindrop and the Victrio division of Verint Systems, are currently making these offering more viable for companies of all sizes seeking to detect fraudsters trying to gain entrance to their customer interaction systems with security solutions that are relatively simple and inexpensive to implement.
Newer passive speech authentication methods have removed the requirement of enrolling users via the prerecording of a speech print. Today’s systems can analyze a caller’s voice in the first five or six seconds of introductory remarks, match speech patterns across 150+ variables to previously recorded customer service recorded calls and fraudster lists, and notify the customer service agents in real-time whether they are speaking to the customer or a pre-identified fraudster. More advanced and futuristic applications now appearing in the solutions of the aforementioned providers can identify a speaker using voice biometrics, analyze what he is saying with speech analytics, and decipher where the call is coming from using telephony analytics that can identify the network characteristics of a specific carrier’s network, which has a unique network “fingerprint.” This augmentation of traditional voice-biometric technology with behavior data, carrier signal analysis, supplemented with available CRM information has the potential to enhance the effectiveness and benefits of fraud detection technology greatly, and to improve authentication speed and accuracy of detecting and preventing security breaches significantly.
Also, when we consider that an increasing number of customer service calls are initiated from mobile phones, and now more often via smartphones, fraud detection processes have the potential to be improved even more than ever before in separating fraudulent calls from legitimate calls. The precise geographic location capability of a smartphone, which can be forwarded and read by an enterprise fraud detection system allows for cross checks on the caller’s location to ensure caller ID information is not being manipulated and that a caller is not simultaneously operating on different channels of a multichannel customer service system from separate and disparate locations simultaneously. Such logical testing of collected information will do much to enhance the predictive analytics capabilities of the fraud detection systems of the future.