Background

The Company is a well-known and substantial company in the financial services industry with two main areas of business: retirement & pensions, and insurance. They have a substantial call center, doing queries and new business from both their end clients as well as brokers and corporate customers. They also use the call center to make outbound calls as part of marketing campaigns and for call-back follow-up purposes. Their clientele is generally older and not always open to technological changes, so any new solutions must be easily introduced.

Customer Requirement

The Company had several requirements that needed to be addressed. As a US-based organization, a common initial caller identification is based on the caller’s Social Security number. This is an insecure and easily discovered number, available via several sources derived from past data breaches. The second customer identification key is the caller’s mobile or landline number in the form of their CLI. Again, this is easily spoofed, especially from landlines, and has limitations as repeat callers do not necessarily use the same numbers consistently.

Since the contact center deals with both individual and corporate accounts, a caller could belong to both categories, and it was important to determine quickly and accurately whether they were calling in a private capacity or as part of corporate administration. This was handled by entering a number like a Social Security number but containing a different number of digits.

A large percentage of callers are elderly and retired and are reluctant to adopt new solutions, particularly those that give the impression that they are being monitored or watched. The Company was looking for a voice biometric solution that answered all these key requirements.

ValidSoft Solution

ValidSoft offered their cloud-based Voice Biometrics solution integrated with their contact center partner, Five9.

Following the discussions and considering the caller profiles, ValidSoft recommended that the voice biometric solution should be based on a text-dependent model which uses a known and fixed phrase as the basis for the voice biometrics. Text-dependent ensures that the caller is aware of the voice biometric process and can identify it as part of the identification with little explanation. At this stage, we suggested that text-independent models should not be used since, although they involve the use of natural speech rather than the potentially stilted fixed phrase, they risk giving the impression that the caller is being tracked and monitored during the call. The company’s client base was likely to be averse to this until more education could take place. The chosen user journey for all the inbound calls uses the existing toll-free numbers routing to Five9’s platform, which triggers a request for the caller to enter their Social Security number or corporate number.

Benefits

The company has gained a new and effective tool in its fight against fraud. With the added advantage that their contact center agents are not spending upwards of 75 seconds identifying a caller using legacy and insecure methods (e.g., “what you know” KBA) and instead are connected immediately with a pre-authorized caller who has probably taken no more than 5 seconds to pass through the process.

We are told that their callers appreciate the smoother process and are easily convinced that there is an added level of security being applied to their own accounts. 

The anti-fraud process is enhanced in tangible ways in that it is a simple barrier to fraudulent attempts to access an account with a known KBA, learned, overheard, or guessed by the fraudster. It is also enhanced in intangible ways in that the caller is prevented from talking to an agent until some VB attempt has taken place and thus prevents the opportunity for social engineering-derived fraud attacks.