In this article series, we will explore how new behavioural data is generated and early signs of how this data could be the next wave of predictive power for issues like fraud, and maybe even claims costs. Later this month we’ll be hosting a live webinar where we will discuss these trends and review a software solution to capture this kind of data. To register, click here.
In our article last week, we talked about the subtle difference between causality and association and shared some insights about how a new software product can help insurers and lenders capture behavioural data from applicants. This data shows a company the specific behavioural actions exhibited by an applicant during their web session, which can be especially helpful for detecting fraud.
So how does this work?
A simple example might be an applicant for private motor insurance who first enters a value of ‘3’ in a question about violations or convictions in the last five years. Before submitting their data, they go back to the same question, hover, look at the levels offered (the most was ‘4+’) and change their answer to ‘1’ before hitting ‘submit’. Insurers would love to have this data, and with ForMotiv, this data is captured and made available in real-time. This not only facilitates post-sales investigations; data is collected in real-time, so the insurer can choose a different treatment for the customer, or even decline the application.
But remember the association discussion from last week?
What’s really interesting about ForMotiv data is its predictive power as new associative variables. Consider the speed of keystroke typing – data has shown that individuals who exhibit a much slower (or in some cases much faster) speed of typing than is typical for a certain field are more likely to be fraud-prone. A field that an applicant would normally know, such as their national ID number, is generally entered at a common speed. But one company found that a small number of applicants enter this field much faster and they are much more likely to be fraud-prone. This data itself is not fraudulent, but the highly accelerated typing is associated with fraud propensity.
In addition to entering, changing, hovering, typing speed, total time to enter and other speed metrics, ForMotiv captures more data elements for every page a user encounters. For some insurers and lenders, this means that they are receiving thousands of new variables for their applicants. Summarised scores of abnormality and outlier propensity can be calculated, and customers can use these built-in summary scores immediately, whilst bespoke models are being created.
Learn about ForMotiv and the new partnership created with FTI Consulting here.
Next week: what’s so special about my mobile phone number?
Read part one of this series here.