Just let them commit fraud
Predicting whether a user will commit fraud can be really tough.
You may be better off just letting them do it.
Why do I say that?
Here’s a strategy that could be more effective:
When you detect that a user has committed fraud, put all your focus on never letting them do it again.
I’ll share an example to illustrate.
Recently a ride-hailing platform shared with us that they had a problem with offline rides.
Offline rides happen when drivers and riders work together to cut the platform out of their share of profit for the ride.
For example, a rider might request a ride, and then once their driver arrives, cancel the ride in the app. They then make a verbal agreement with the driver in person to do the ride for cash instead. This allows the rider to pay less and the driver to earn more, but the platform doesn’t profit at all from the ride that they facilitated.
So this company knows their users are doing these offline rides, and they were asking us how they could use Incognia’s data to predict this kind of behavior so they could stop it before the ride happens.
We were frank with them: There’s not a good way to predict this kind of behavior.
Recommended by LinkedIn
Instead, we suggested this:
We told them that if they did those things, their offline rides problem would likely go away.
It’s a classic case of working smarter, not harder.
Some types of fraud are very difficult to predict before they happen.
You could spend a lot of time and money trying to do that, only to fail.
In some cases, you’re better off detecting when a user first commits that type of fraud (or noting when they break the policy).
And then putting all your focus and investment toward ensuring they can never do it again on your platform.
Other things you should check out:
Fraud | Verification | Abuse | Authentication @ Incognia
1moEffective ban evasion detection and prevention delivers outsized business impact for platforms 💡