Скачать или смотреть New Account Fraud Prevention: Behavioral Biometrics and Population Data (Feedzai Spotlight Sessions)

Thanks for watching our latest Spotlight Sessions on new account fraud prevention. For more on the money mule challenge, cost of onboarding for banks, and behavioral biometrics, please subscribe:    / @feedzai  

TRANSCRIPT:

If you have one device making three applications on three different users, that would be highly unusual. Why would somebody be doing that?

What we're now doing is we've now built up a picture of behavior, not just at the individual level, but we're starting to see this at a more of a population level.

Absolutely.

So now we can take that and apply that to new account fraud - a user that we’ve maybe never seen, but we can say, how does that play within the population? Give us some more detail in terms of the kind of things we're doing there.

So that's the key difference. You haven't got the luxury of that user baseline when they come to your bank for the first time to open an account. They're an unknown entity, you haven't seen them before. You have no idea what to expect in terms of what their normal behaviors will look like.

So you're right. What you have to think about is, how do you move the needle from thinking about, is this different relative to my user norm? And you lift that up a level to say, is this different relative to my population norm?

Examples would be, if you have one device making three applications on three different users, that would be highly unusual. Why would somebody be doing that? Perhaps a slightly more complex example might be, analytically speaking, you never see any applications from a certain group of IPs, and then you see a spike in activity from a particular part of the world or a particular part of the country, which is highly anomalous relative to what type of traffic you normally see.

And I think the final thing and the more complex thing is taking that population principle, but starting to think about it with more complex data principles like behavioral biometrics. So you think about how you, Rich, would enter your personal information into a web form. You know your name, you know your address, you know your email, you know your phone number. So the way in which that data goes through the form would be quite fluid. There wouldn't be much hesitation in there.

But if you were trying to impersonate me as a victim, you might be able to remember my first name and surname. But when it comes to perhaps some of the more complex data that you need from me, knowing a full address, knowing a full phone number, it's unlikely that you're going to be able to enter that in a way that's as fluid as I would as the genuine user.

So things like field timings become really interesting. If it on average takes 10 seconds for the user to type in their email to the email box, and all of a sudden you've got somebody taking 30 seconds, that would be highly anomalous.

#aml