People who commit insurance fraud over and over again. They do it because they get away with it. But how do they get away with it and what can be done.
The problem: fraudsters change parts of their identity – slightly change their name, pick a new address or a different car to fraud from. So the next time they show up they don’t look like the guy that stole money last time. Almost but in traditional rules based analytics almost doesn’t cut it.
But most fraudsters don’t change out everything – they’ll keep their phone number or they’ll access online sites from their real home, or use the same car while fiddling with their name and their garaging address So these unique markets are critical to connect the disguised fraudster and relate him back to the person that’s already been identified.
This is incredibly hard to do quickly and well with traditional relational databases. So we use Graf databases that map every data element as a network of connections: thus a phone number will be connected to its owner, so you find the same phone number and vin connected to a different person and you have strong evidence that the person is submitting false information in order to get by your screen.
They have the wargame edge – the ability to ‘case’ a carrier’s rating methods and rules by submitting many quote requests with slightly different information to see where the discontinuities in the rating system are.
They can morph their details and confound traditional business rules.
Fraudsters can take their time investigating and plannign but carriers must quote and decide in a matter of minutes.