Case Study: How Clever Messaging Rooted in Psychology Reduced Insurance Fraud by 36%
If the opportunity arises, people might exaggerate their circumstances to get themselves a better deal. This is costing the UK insurance industry approximately one billion per year. How can you make sure people are honest when claiming their insurance?
In this article, you’ll find out:
- How you can apply nudging to reduce opportunistic fraud; and
- How Dectech achieved an average 36% decrease in fraud with clever messaging.
InsideBE is the largest behavioral economics and consumer psychology hub for marketers, sales people, and business professionals alike.
How to reduce dishonesty in insurance claims
The need to tackle opportunistic insurance fraud was flagged by the UK Government-established Insurance Fraud Taskforce. And this is not a small need. It is estimated that in the UK alone, opportunistic insurance fraud costs the industry around £1 billion every single year. That’s a 1 with 9 zeros.
For many insurers, it is a multi-million-pound problem and one which causes the premiums of honest customers to be higher than they otherwise would be. In turn, it leads to greater customer dissatisfaction and an erosion of trust between customers and insurers. All around, it’s bad for business.
It is estimated that in the UK alone, opportunistic insurance fraud costs the industry around £1 billion every single year.
To deal with this problem, the Taskforce recruited behavioral experts from Dectech to understand if psychology could be used in insurers’ communications with customers to reduce opportunistic fraud.
The questions at the forefront of this research were:
- Can we incorporate messaging into communications to reduce instances of opportunistic insurance fraud, and how effective would this be?
- Can these messages and interventions be developed without negatively impacting consumers’ perceptions of the insurance provider?
Dectech, through the use of multiple interventions, managed to reduce opportunistic insurance fraud by an average of 36%. Let’s dive into how they did it.
What is opportunistic fraud?
Opportunistic insurance fraud occurs when otherwise law-abiding citizens submit false or exaggerated information on an insurance application or claim forms to get a better deal for themselves. With insurance, this means paying a lower premium or receiving a higher payout (more money).
This type of fraud can take quite a few different forms and can have quite a few different motivations - it’s a complex behavior. It’s unsurprising then that Dectech rolled out and tested a large number of interventions.
Testing for fraud
Testing for fraud is not easy. It’s a behavior people do not want to admit to, and it’s quite easy to see when people are trying to figure out if you’re lying. Dectech deployed their Behaviourlab, an innovative research methodology that replicates the real-world decision environment faced by consumers in an online experiment.
The Behaviourlab is an immersive experience: setting the scene and presenting realistic online forms, persuading participants to behave as they would in real life, ensuring that the results obtained do actually replicate outside of the lab.
Testing for fraud is not easy. Thus, Dectech deployed their Behaviourlab, a methodology that replicates the real-world decision environment faced by consumers in an online experiment.
So, what did this look like? Well, it involved recreating real-life motor insurance applications and claim processes, including questions to which a customer may respond dishonestly in order to get a better deal. In this case, the ‘contentious’ question was whether they had been caught speeding before.
So, what was the actual intervention? Dectech decided to test 18 different ideas to see if people could be nudged to answer the contentious question honestly, for both insurance applications and claims.
Dectech ran the study in three parts:
- They used an elaborate method (Unmatched Count Technique) to determine how often people really speed.
- Next, they designed a control condition to test the extent to which people were dishonest when asked about speeding directly as part of a motor insurance application or claim.
- And finally, Dectech tested 18 different messages, for both applications and claims, to see which would best prevent dishonesty.
And behavioral science did it again. An average 36% reduction in fraud was achieved across the different conditions.
Dectech decided to test 18 different ideas to see if people could be nudged to answer the contentious question honestly. The results were astonishing. An average 36% reduction in fraud was achieved across the different conditions.
Behavioral insights used
Each of the 18 test conditions, for both applications and claims, was a message that popped up immediately before people filled in the answer to the contentious question: “Have you ever been caught speeding??” All of the messages used were grounded in 5 key principles of behavioral science:
- Social Norms: people tend to behave according to social norms and will follow the behavior of others.
- Self-consistency: people seek internal consistency in the way they behave.
- Priming: the choices people make can be unconsciously influenced by the immediate context, even when not apparently relevant.
- Framing: people can be influenced by the way information is presented.
- Reciprocity: people tend to reciprocate good deeds from others.
To elaborate on how this would work, these are two examples of the messages used:
Framing message: “Insurance Fraud is a serious crime which can result in serious consequences for fraudsters who may find their future job prospects impacted, find it harder to obtain insurance and other vital financial services, obtain a criminal record and even face the prospect of imprisonment.”
Social Norms message: “Did you know? 95% of insurance customers fill in their forms fairly and accurately, and make honest insurance claims - be one of them.”
An average 36% reduction in fraud
Almost all of the 18 interventions had a statistically significant impact on dishonesty. The interventions worked equally well across both insurance applications (average impact of 36%) and insurance claims (average impact of 37%).
However, the performance of the individual interventions varied, with the very best (the social norm message from above) having an impact in excess of 70% of dishonesty swayed.
Dectech was curious to see if these interventions could have a cumulative effect: rather than testing one in isolation, what would happen if people saw multiple messages?
To test this, the second round of experiments was run, mirroring the form and approach of the first experiments, but also testing the impact of showing two or three messages, one after the other. The results showed an even greater impact, with an average of 57% of dishonesty swayed when two messages were shown, and 70% swayed when three messages were shown. This worked equally well for both insurance applications and insurance claims. Double win!
If you design solutions with a clear understanding of what’s going on in your customers’ minds and how their subconscious works, tiny changes can have a huge impact.
In addition to opportunistic fraud having been reduced, it is also important to mention that neither the single interventions in the first study nor the cumulative interventions in the second study had a negative impact on perceptions the consumers had of the insurer. So insurers can apply these insights without having to worry about UX or reputational damage!
This case study is a beautiful example of how complicated problems don’t always require complicated solutions. If you design these solutions with a clear understanding of what’s going on in your customers’ minds and how their subconscious works, tiny changes can have a huge impact.
Key Takeaways:
- Combine solutions that prove effective. If results show that some of the solutions you designed worked well, try to combine them. They might work even better. There is of course the caveat that the number of interventions needs to balance the effectiveness of adding interventions with the need to maintain their subtle nudge-like effect.
- Solutions don’t need to make rational sense to work. The fact that 95% of people don’t cheat shouldn’t, in a rational world, discourage others from cheating. But it did. It is not necessarily intuitively obvious which nudges will be most effective, so they should be tested.
- Timing is crucial. Sometimes more important than the message is the timing when you hit your customer. It was essential that Dectech displayed their messages right before people answered the crucial questions. Had it been at the beginning or at the end of the form, the effect might have been much smaller.
- Targeting the right people/decision. The nudges worked because some customers are genuinely on the fence about whether or not to be honest, so a small nudge in the right direction can be very effective
