Fraud is unluckily part of the insurance business and it is quite commonly accepted by « normal » people to cheat insurances. Fraud is also a complex and multifaceted problem. Nevertheless, modern insurance can use technologies to do so much more than before:

It starts by having a « suspicion score » for each client from the first day that a policy is created. The starting point of that KPI is based on two key factors, the data collected by the mobile app when the insurance is being purchased as well as external data present in the system. Concerning the data collected by the mobile app, we can already detect behaviours (when was the app used, at what time of the day, from which geolocation, did the customer connected multiple times, did he play the custom settings or not, for how long, etc). The second point is external data since we can also understand where does the person lives, why type of town or neighbourhood, average incomes, etc.

When a claim will be raised, the « suspicion score » in combination with the type of claim, claimed amount, previous client claims and last but not least overall past fraud analyses will be combined together (e.g. the suspicious disappearance of expensive jewellery has a higher potential for being fraudulent than a stolen smartphone or laptop). It is important to target the right claims, at the right time, with the right tools. Luckily, predictive modelling and advanced analytics are coming into play as essential tools for fighting insurance fraud. These tools can be automated, preventing the need for hands-on manual analysis. Our nore advanced investigations will be focused on the items that have the greatest potential for cost avoidance and successful identifications.

Other approaches will be used as well such as social media listening to see if the claim fits the social media profile. We will also be using video and livestreaming in raising claims which is difficult for the average person to fake a video, especially when the device’s location access is turned on. Using video also means that you can have the person recorded to explain its claim which automatically make people more reluctant to cheat (it feels easier to cheat with a filling a form than doing it live in front of a camera). Last but not least, in future we can use behavioural analysis to check videos (studying body movements, facial expressions and eye movements).

In conclusion, technology, data and scoring engines will enable us to decrease the fraud compared to our competitors and we will get better and better at doing it. This will allow us to save consequent anounts of money by preventing and detecting fraud which will allow us to improve margins and be better than our competitors, resulting in cheaper policies and more clients going forward.