As of today, 10-15% of claims within an insurance company are fraudulent. This represents a huge amount of money – over US$40 billion a year in the US alone – and it is increasing. Investigators do their best to spot them, but the process is time consuming and costly; most detection happens at the end of the process, after money has been paid out.
The need, then, is clear. Investigators have to be able to spot problems earlier in the chain. But there are obstacles at every turn: competition from increasingly sophisticated professional fraudsters, as well as people who are simply massaging their domestic claims. And all this as the sheer amount of data they have to wade through multiplies by the minute.
Fraud is an easy enough thing to understand, but the key to success is in spotting it in time to be able to do something about it – and the answer to that lies in instant access to big data, and the ability to mine it for essential information, saving money and helping companies to run more efficiently.
That’s what we’ve done by combining fraud management software with the HANA platform, creating a solution that makes an immediate and tangible difference. We recently used it to run a simulation over 185 million policies – and got the results back in less than ten seconds
Take the case of claims coming in, say, for road accidents in an industrial area, at night, with no witnesses. Any investigator would want a closer look at them. But if they change the location to the countryside, for example, we can now combine rules with a predictive algorithm to spot those patterns too, right from the moment of the incoming claim notification – and, thanks to the speed of SAP HANA, we can do this in real-time using not just a sample, but total stored data. This means that potential issues are identified way before any payment is made.
And all kinds of data can be collected, from payments to policies to claims. Even customer click-behavior, so we can see how long people spend reading the terms and conditions when they’re online – a useful warning for people looking to massage a claim.
It means we spot some claims that would easily go unnoticed too, or which are long-winded and tedious to research – such as fraudster rings where people work together to create a series of frauds. All identifiable with an easy user-interface that can be integrated seamlessly into any existing business process.
There’s no way that a human can be taken out of the equation, so we have to support them as well as possible or we risk credibility and customer satisfaction. To me, the most important thing is that by managing big data effectively, saving companies money and time. We’re also freeing up experts to be experts, making the final decision with all the information they need at their fingertips but none of the fuss. That’s a future we can all look forward to.
Fraud – SAP official Overview Video: http://www.youtube.com/watch?v=L_NiH6D28Z0