New avenues of attack are becoming increasingly popular, from social engineering and phishing to spoofed Calling Line Identity (CLI) and SIM swap. Fraudsters are targeting voice channels by faking identities and breaking through traditional security measures that were once considered highly secure.
The cost of these attacks is high, not only in terms of monetary impact, but in terms of brand and reputation. Negative experiences with fraud attacks can weaken consumer trust. A customer who has been a recent victim of fraud is likely to shop around and may switch to a rival provider. Reports of fraud can also damage a brand’s reputation and discourage new customers.
In parallel to this, consumers expect their interactions with business to be fast and frictionless – despite frequently forgetting the answers to knowledge-based authentication questions that they set. This results in a negative customer experience, higher average handling time and more calls escalated to a human agent – increasing the cost to serve.
So, how can you provide customers with the most efficient and secure fraud prevention and authentication tools possible, while meeting all these challenges?
‘Active’ voice biometrics, where users have to say a set phrase which is then compared against the stored voiceprint, used to be the norm in voice authentication and continues to have many applications. Thanks to advances in technology and Artificial Intelligence (AI), we now have more sophisticated verification measures available, such as text independent or ‘passive’ biometrics, where natural speech is compared to a stored voiceprint. Cutting-edge systems can also optionally monitor the dialogue of a call and analyse speech patterns, noting commonly used phrases specific to an individual.
There are other advantages too in using passive biometrics. User experience for all parties is improved as the caller isn’t required to repeat a specific phrase. This makes it easier to deploy for agents and cuts down verification time during a call. In addition, this kind of authentication is not language specific, making it ideal for international customers and global application.
Using new technology such as AI, security products can identify suspicious behaviour to help detect fraud based on unusual customer behaviour. Fraud prevention uses both real-time and offline analytics, using a variety of characteristics from a call and comparing them against a database of known fraudsters. Comparing call characteristics in real-time will immediately flag to the business or agent a risk score for suspected fraudulent activity, while analysing calls offline will allow them to search for patterns of suspicious behaviour and add to the fraudster database.
We can combine call signalling data with caller behaviour and characteristics of the device, to provide stronger authentication. The system can also flag international calls or attempts to access an account via repeat calls on different numbers. For example, caller X appearing 17 times using 11 identities would be flagged as a risk. Suspicious calls are sent for additional scrutiny while legitimate calls are streamlined for authentication, freeing up an agent’s time to focus on customers who pass these initial security checks. It’s a win-win situation for both the agent and the customer.
We work with many partners to provide integrated security solutions for voice with Nuance Gatekeeper and Smartnumbers Protect. By providing the security in our network, and the flexibility in the APIs we’re creating the ability for organisations to customise our solution features – removing the need for them to invest in managing infrastructure and software from multiple suppliers.
If you’re interested in working with us to help implement effective fraud prevention strategies for your contact centre customers, please get in touch with us. For more information visit Unlocking the potential for channel partners | BT for global business.