Voice biometrics for CX detects customer age, fraud
As of now, Nuance voice age-detection features are for security teams that want to prevent fraud and to route service calls. Analyst sees potential for marketing personalization.
Your voice changes as you age, getting a little rougher each year. Nuance Communications Gatekeeper voice biometrics technology analyzes that and 1,000 other "micro-characteristics" that can't necessarily be heard by humans to confirm an elder customer is on the line.
The company developed the senior age-detection voice biometrics, released last month, for its customer Telefonica, to help with increased contact center volume during the pandemic. The Madrid telecom uses it to route seniors to priority customer service, with shorter wait times and protocols to prevent fraudulent account takeover.
Numerous vendors, including Sestek, Verint and Nice, offer AI-based voice biometrics tools, mostly along the lines of voice prints to authenticate individual users. Nuance Gatekeeper is a security package that integrates with contact center platforms to provide that, as well as customer identification through conversation syntax and device authentication.
Customers in the financial vertical have expressed interest in using it as a weapon against account fraud there, too, said Brett Beranek, vice president and general manager of Nuance's security and biometrics. Seniors are particularly vulnerable to social engineering attacks, often from younger relatives, that clear out their finances.
"Imposters sometimes aren't even the same gender of the name of the individual they're calling about, but they have the answers to the challenge questions," said Dan Miller, Opus Research analyst. "In many instances, a biometric or other detection technology is the only line of defense. Agents want to be helpful, want callers to accomplish what they're trying to do. They can be the weak point in any security scheme."
One Nuance customer that manages retirement funds for seniors told Nuance that 30% to 40% of financial fraud committed against its account holders came from the victim's children or grandchildren, and that it often goes unreported because they don't want to send their relatives to jail.
"They might have passwords," Beranek said. "But we're able to generate a flag that says, 'Hey, the person that's calling is not in the right age category, this is not somebody in their 60s, 70s or 80s.'"
Nascent marketing potential
The senior age-detection technology for Telefonica branched off from a similar feature Nuance developed to detect minors, Beranek said. Many countries have regulations surrounding conversations with children, such as making it illegal to record them, so customers use that age-detection voice biometric for compliance.
While Nuance did not discuss specific accuracy rates, Beranek generally characterizes Gatekeeper age-detection as weaker in the age group where people approach senior age, with "double digit" inaccuracy for customers between the ages of 60 and 65.
[Fraudsters] might have passwords, but we're able to generate a flag that says, 'Hey, the person that's calling is not in the right age category, this is not somebody in their 60s, 70s or 80s.'
Brett BeranekVice president and general manager security and biometrics, Nuance Communications
Nuance Gatekeeper is a voice biometrics security tool, and Beranek said that the company does not have plans to move it into other realms of customer experience, such as sales or marketing.
There is, however, potential for personalizing customer service by giving agents age-specific content and marketing upsells using voice biometrics for inbound and outbound contact center calls, Miller said. In fact, some companies such as Chase have replaced passwords -- or offered them as password alternatives -- with voice prints for customer convenience. But the industry is reticent to move in that direction.
"It is surveillance, so there's an understandable pushback," Miller said. "But it could be baked in as benign surveillance, a very valuable tool for personalization.
