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.
Insurers have long known, and have statistically proven, that a person with prior motor insurance losses is more likely to have a claim in the near future than another person with no prior losses. This feels appropriate and makes intuitive sense.
Society believes a prior loss may have involved poor driving ability or lousy decision-making and that behaviour will carry forward into the future. But the social understanding and the statistical reality are actually very far apart.
A crucial unseen link
What society is not aware of is the reliable link between claims that were not the driver’s fault and any future claims, regardless of fault. There is even a link between a prior claim related to a natural disaster loss and a future claim for collision or theft. At this point, one may ask how there could possibly be a link between the two events, and therein lies the challenge. We tend to see any ‘link’ between a behaviour and a claim as being related to causality, but this is rarely the case. There is an association with the two events: people with one behaviour are more likely to be in a group of people with another event, even if there is no causal connection.
Predictive modelling – piecing the data together
This is particularly true of behavioural data. Some of the biggest breakthroughs in predictive modelling have arisen when new behavioural data has been introduced. The most common example is credit data, where a person’s behaviour in applying for loans and credit cards and timely payment of balances result in a credit score which can help predict who will make a claim on the motor insurance policy – even though there is clearly no causal relationship between the two. Early users of credit data were shocked to see such a strong and uniform correlation between poor credit and high loss experience.
Web behaviour – spotting the connections
Today, there is a new type of behavioural data on the horizon: in-session web behaviour. ForMotiv, a Digital Behavioural Intelligence platform, is making it possible for companies to capture customer journey and behavioural data, looking at how an applicant interacts with a company’s website. Did the applicant first choose ‘2 accidents’ and then change to ‘No accidents’ before hitting submit? Did they hesitate when answering the question about income? Did the individual initially fill in the mother’s maiden name question and then erase it?
Insurers and lenders in the US are using ForMotiv to capture this data and they are finding that it helps identify times when erroneous or fraudulent data is submitted, resulting in fewer mis-priced policies or credit offers, and ultimately lower costs for the rest of us. Now they are going global and FTI Consulting is helping them do that.
Learn about ForMotiv and the new partnership created with FTI Consulting here.
Read part two of this series here.