At Pindrop, we love voice. It is our company’s mission to provide security, identity and trust for every voice interaction. We believe in a world made more human with simple interfaces like using your voice to not only command, but to authenticate simultaneously. We have found a way to use voice to help us authenticate genuine customers so they don’t need to answer so many security questions or remember a PIN, and we use voice to help us catch fraudsters. But much like our technology we don’t solely rely on voice. In this post we will look at the role voice plays in stopping fraud.
For organizations fraud is a zero value expense that only detracts from the bottom line and stopping fraud is an ongoing effort. In times of financial uncertainty, firms can seek to tighten their grip on costs to insure they can survive and thrive. Reducing fraud doesn’t have to be painful and when done properly even provides value in operational efficiencies at the agent and for fraud teams.
Your Fraud Solution Can’t Rely Solely on Voice
In certain situations, like the IVR, voice may not be available for analysis or maybe unavailable for a number of reasons, so we have strived to give our solutions the flexibility to use what factors are present to make the best possible decisions. Checking a suspected fraudster’s voice against a data set of known bad actors is one method you can use voice to detect for fraud. We use this tactic as part of a larger solution, but often hear that using voice bio as a primary fraudster detection method results in detection rates below 25% or more.
Fraudsters don’t always use the voice channel, or a single channel for that matter. So relying on a fraud detection that requires a voiceprint to be effective will also limit your total detection rates. For example, how would a voice solution catch fraudsters who mined an account number and personal info from an IVR, and then used the information online. No voice was used by the fraudster, nor was there an opportunity to print a device, since IVRs are rarely monitored. The fraud might have just appeared as just an unauthorized login on the web portal. With voice based solutions that don’t do any risk analysis, use Phoneprinting™ technology, or ANI analysis, not only will catch rates be lower, but there will be fraud that slip past a voice biometric if not use in conjunction with other additional factors.
Pindrop Uses Multiple Factors to Stop Fraud
Pindrop has been stopping fraud for our customers since 2012 and has prevented hundreds of millions of dollars in fraud losses. Along that journey we have uncovered not only how to detect fraudsters over the phone but also how they use other channels and at when in their processes. For example, using behavioral heuristics such as account changes, failed step ups, non-monetary transactions, reputation of the phonenumber accessing the account in the IVR, all are analyzed to assess if an account is at-risk of being defrauded. Stopping the most amount of fraud includes watching accounts over time and analyzing calls as they come in and providing an accurate depiction of the call risk and account risk. These are both much better indicators of fraud than only checking against voices that have defrauded you previously.
So while we love voice, if we only focus on it, we miss the larger picture. Having a channel dedicated solution might not have enough visibility to track fraud effectively. Fraud is going omni-channel so tracking them will take insights from every channel to be effective. Fraudsters are using every advantage they can to extract money from customer accounts, and today they are using the IVR as a validation tool that is being left unmonitored.