One of the most effective uses of data and analytics in insurance is to prevent fraud both at the point of underwriting and from claims.
Jeroen Morrenhof founded FRISS in 2006 and today the company has over 180 implementations of its technology around the world. Identifying potential fraud creates savings direct to the bottom line - Jeroen believes FRISS has saved its clients over 1 billion dollars.
Join Matthew Grant as he finds out what motivated Jeroen to start the company and lessons from the last 14 years including a chance to learn about:
- the choice between bootstrapping and bringing in external funding
- understanding the "fraud funnel"
- keeping ahead of the latest approaches to fraud during Covid-19
- using third-party data sources and in-house experience
- why the company is returning again to the UK
If you like what you're hearing, please leave us a review on whichever platform you use, or contact Matthew Grant on Linkedin.
Underwriting and claims fraud revealed - Podcast 105 highlights
Matthew: Where did the idea for FRISS come from and why did you want to set the company up?
Jeroen: I was working with insurers as a management consultant and we talked about fraud a lot. Fraud undermines the trust relationship between insurers and customers, and it isn't a victimless crime. It costs customers about £700 a year in additional premiums. For me, that was just wrong, but when I talked to senior industry leaders, they said, "We agree with you, but there's not a lot we can do."
That led to the founding of FRISS 14 years ago and now we have 184 customers in 42 countries. Even more important is that we've saved those customers $1bn over the last year. That was always my main driver, to make sure we protect honest customers and bring the savings back to insurance.
Matthew: A billion dollars is an incredible number. How do you measure and define it?
Jeroen: We offer three solutions; fraud detection and claims, risk assessment and underwriting, and supporting the Special Investigation Unit (SRU) up to court when necessary. With claims fraud, the savings are related to not paying out claims or partially paying out because they get withdrawn, or because something is proved to be fraudulent in court.
With underwriting, it's difficult to know what kind of saving a customer will get. They know how much premium they will miss, but they don't know how much the losses would be. That's based on historical data and analysis and we can predict what future losses a customer will get.
Matthew: How do you ensure FRISS stays ahead in a competitive market?
Jeroen: We're 100% committed to this industry, and we're 100% committed to reducing, preventing and detecting fraud that's happening in insurance. Because of that clear focus, we're also set up to be a trusted advisor.
A lot of people in my team have the knowledge and background in the industry. It allows us to provide great technology with powerful AI, but also to be that trusted adviser that helps carriers to stay ahead of the curve. That is something that sets us apart.
Matthew: That's consistent with other successful businesses, who have good technology and the people to use it and advise clients.
Jeroen: We use a combination of human intelligence and artificial intelligence. We're not able to detect new fraud schemes automatically and sometimes the feel of an adjuster or writer is important. We feed all the referrals, all the insights into one process and one repository of cases. The AI technologies pick that process up and improve it over time.
Matthew: Do you provide alerts for customers to make them aware of new forms of fraud?
Jeroen: We can see clear changes in fraud patterns, which is why we invest a lot of money in our product to pick up on new schemes and deliver that information back to customers. We push new insights and software to customers every three weeks and new intelligence every quarter, looking at what we call the fraud funnel. What is the number of claims or applications a customer had? What percentage did we flag? How many led to an investigation? Which of them ended up with a saving?
The data, combined with benchmarking with other customers, provides continuous improvement. The best way to show how our customers appreciate that is we're currently running at a negative churn of 109%. It's a clear indication that our customers value the services that we deliver.
Matthew: The two main areas FRISS helps clients with are assessing potential fraud around underwriting and fraudulent claims and claims payments. Can you explain more about that split?
Jeroen: About 40% of our revenue is from underwriting, 50% is from claims and 10% is from our case management solution. The last one is a cheaper solution so it has lower revenue, but it's been used by most of our customers.
With underwriting, we're looking at fraudsters and other types of non-rating related risks like fast claimers, frequent claimers, or a high probability of churn. Those aren't necessarily in the actuarial models, but we can predict those behaviours based on what we know about the customer and what we can find in external data.
Matthew: What are the more common types of fraud?
Jeroen: We typically make the distinction between optimistic fraud and organised crime. Optimistic fraud is often linked to an event or claim, but then someone decides to take advantage of it. For instance, we had a claim which was almost settled and then all of a sudden, they added what we call phantom passengers. It was seven passengers in a Fiat 500 claiming injuries.
With organised crime, you see events being created. The car accidents, the trip and slips, crash for cash, that are well known in the industry, especially in the UK. We're also seeing fraudsters move into areas like travel and pet insurance, or commercial lines, where it's easier to hide behind a screen. A recent example was some fake accommodation and a booking website to make it look legit, but then the bookings were cancelled. A lot of claims were paid on that one, and it is low risk for the fraudster in terms of getting caught.
Matthew: Can you give some examples of the data sources FRISS uses?
Jeroen: We have over 200 integrations with external data sources that leverage claims history, fraud registers, and data about objects like vehicles, properties or companies. The internal data that insurers have is also unique. A lot of insurers underestimate the value of what they have, so that's something we help our customers with.
We still have challenges around the permissible use of social media, which can help identify suspicious connections between people, and we're improving how we use classified ads websites for identifying potentially stolen goods like cars and jewellery.
Another big benefit for our customers is that they can change those data sources. For example, a customer can switch between multiple credit score providers through our platform to make sure they get the best price every time.
Matthew: You mentioned using AI and you're also using a decision tool based on data, which sounds like two slightly different approaches?
Jeroen: We leverage internal customer data and external data integrations in real-time. If a claim comes in, we search the relevant data sources and combine it with internal data to produce what we call a FRISS score, which gives one holistic overview of the risk related to the application or claim.
The FRISS score tells us why the risk is high, the data behind it and all those insights are available to a customer so they can determine what the next course of action is.
Matthew: You've decided to bring FRISS back to the UK, which you tried to do before. What happened then?
Jeroen: The UK was the first country we wanted to expand to in 2011, but we made all the mistakes you can make when expanding. We didn't hire anyone local, the product didn't have any local fraud schemes or data sources at that time, and we also didn't have the financial backing as we were still in the bootstrapping phase.
If you combine all that with the fact that the UK is a highly competitive market, we were too big a jump for insurers to make the first time.
Matthew: What was your experience of bootstrap versus having funding?
Jeroen: We were bootstrapping for quite a long time and I'm proud of what we created on our limited budget, but it wasn't all good. There were some monthly struggles to make paycheques and I had to transfer earnings from my other company to overcome some liquidity bumps. It was a fun phase, but it held the company back from reaching its full potential.
Making the switch to raising funds opened up a lot of that potential. To be honest, I think an outsider would have taken external investment earlier. Talking to investors gave me a different perspective of the company and taught me things that I wasn't doing during the bootstrapping days.
Matthew: What's the difference between the types of insurance companies you typically work with?
Jeroen: We serve tier one to tier four customers. Typically, when we enter a new country, tier one's are more hesitant to work with a provider that isn't known locally. So, the first carriers we work with are either tier two or the insurtechs or innovators in that market who are willing to take some risks, rather than just going with an incumbent.
Looking at the UK market now, it's still leading the way on insurance fraud, but I also see some first-mover disadvantages. We can provide our FRISS score and build networks in real-time, so there's aren't any dependencies or overnight builds. We have technology that insurers will benefit from.
Matthew: You have an open platform, which has advantages even for more established companies.
Jeroen: We believe in an open platform for customers, but also for the broader insurtech ecosystem. More and more insurers are adding data scientists to their teams and they build great models with all the available data, but they never get them into production.
What we've set out to do is create an open platform that allows customers to not only to use our models, but also other models from insurers.
Matthew: How do you make time to keep up with what's happening elsewhere when you're focused on building a business?
Jeroen: I have a very active network of peers that I get insights from. We used to chat and meet a lot to share challenges when that was still possible. I also love listening to podcasts and reading books. They make me reflect on my own business and that sparks creativity.
I listened to an InsTech London podcast on my mountain bike yesterday about reinsurance. It triggered a completely new idea about how to package our services with one of our partners. I guess the brain of an entrepreneur never stops thinking.
Matthew: Is there anything you'd like to share with us that's coming up soon?
Jeroen: One thing is around Covid-19, and maybe it's a warning about fraud during an economic crisis. For people to commit fraud they have to rationalise that it’s ok to do it, maybe because they think a company is making lots of money. There also needs to be an opportunity and some kind of pressure.
Currently, the pressure is very high on people to make ends meet. We see more and more people losing their jobs and businesses that are in trouble, and new fraud schemes are evolving. We're happy to elaborate on that and share with our customers. I'm also happy for people to reach out to me if they would like to know more.
Matthew: What's the best to contact you if anybody wants to learn more?
Jeroen: Our website has case studies and insights on there because we believe in giving back as well. People can reach out to me through LinkedIn and we have Martyn Griffiths working in the UK market for us. Those are the best ways to get connected with us.
Continuing Professional Development - Learning Objectives
InsTech London is accredited by The Chartered Insurance Institute (CII). By listening to an InsTech London podcast, or reading the accompanying transcript, you can claim up to 0.5 CPD hours towards the CII member CPD scheme.
Complete the InsTech London Podcast Feedback Survey to claim your CPD time.