Wouter Joosse, product manager at FRISS, considers the global insurance industry, and is struck by how progressive the sector is in the Netherlands. Innovative, digital and automated. He still thinks, though, that more can be gained. According to Joosse, insurers can use their role to increase fraud awareness among ordinary claims handlers while at the same time expanding their capabilities. For example, by using authorized media such as photo and video materials in the handling process.
Insurers who work with the fraud detection software of FRISS receive a FRISS Score on their claim reports. Green means there is nothing wrong. Orange means that unusual factors have been found, and red means raise the alarm. Joosse estimates that about 95 percent of claims will get a green flag: pay out as soon as possible. Red is also clear. But what do you do with orange? FRISS is trying to convince insurers to integrate a plugin developed by its partner eviid into the FRISS Score. eviid automatically assesses the authenticity of visual material.
Authorized media as an option based on the claims report
The software makes it unnecessary to include a photo with every claim. “We thought we’d announce this option after receipt of the claims report,” says Joosse. “When we receive a claim with an orange score, we can automatically send a signal to eviid requesting the end user to take a photo or video of the damage. Claims handlers can upload the photo to the insurance company themselves, but it is preferable to use the eviid app. This is a closed system. There’s no way that the photos will be influenced in this system.”
The fraud assessor
How much proven fraud does FRISS actually remove from the claim flow? According to Joosse, it’s possible to give a general indication. “We’re using the system to create a fraud assessor. There is a certain percentage of irregularities, some of which we highlight for the claims handler. After some questions for verification purposes, we are happy if around half of these are actually investigated. And then we’re satisfied if half of these are closed as proven fraud.”
Joosse stresses that proven fraud is not the only quality indicator. “For example, we also measure the claimed value versus the paid value of the claims we flag. And the time savings in claim processing.”
Which characteristics does the system check the photos for?
“The photos automatically provide location information, plus the date and time the photo was taken. We are currently focusing on confirming or disproving a claim based on that metadata. In the longer term, we also want to be able to interpret the photo automatically. This means assessing whether a photo submitted as glass damage actually shows glass damage. That’s the future.”
The system works the same way for videos. Joosse also prefers that insured parties send in a video of their claim rather than a photo. “A video gives much more context to a claim than one or two photos. And don’t just zoom in on the damage. Sweeping from left to right gives the handler more context than just a photo of that wet spot on the ceiling.”
Broader deployment than just fraud detection
FRISS is currently in talks with a number insurers to add eviid‘s technology to their services. Joosse has noticed that companies prefer to use the authorized media technology more widely than just for fraud detection. For example, having an insured send verified photos and videos could make a large number of the physical appraisals superfluous. “That’s what talks are often about. Should the project be guided by expertise, the claim or will the investigation department use the product? From a fraud perspective, I don’t care who provides the photos. I just want them.”
How do you prevent the insured from feeling as though they are being treated like potential fraudsters?
“That’s something insurers are often afraid of. I think you should give the insured a choice: either they provide a video and their claim will be dealt with very quickly. Or an expert can be sent, which will extend the claims process. That’s a narrative that the end customer will understand.”
You should give the insured a choice: either they provide a video and their claim will be dealt with very quickly. Or an expert can be sent, which will extend the claims process.
According to Joosse, retrieving photos and videos has a preventive effect, because fraudsters abandon their claims prematurely as soon as they are asked to provide visual material.
When it comes to prevention, insurers also have a lot under their own control. “At FRISS we can see that claims handling speed is a factor in the fraud rate of a claim. In many cases, there is nothing wrong with a claim. But because an insurer responds slowly or repeatedly asks for additional information, you notice that people start stacking claims and exaggerating the amounts because they think they have to put a lot of effort into making the claim. If an insurer responds to claims more quickly, the amount of fraud declines. Many claims can also be handled automatically.”
Automated payments as well?
“Absolutely. If the claim is covered then insurers should focus on the quickest possible handling. I think it’s more important for insurers to say: I want certainty on certain points. Am I sure I’m not paying fraudster? Am I sure I’m not paying out previous claims? Am I sure the amount claimed is correct? If you can answer yes to all these questions then you should pay out immediately. Regardless of the amount claimed. With FRISS we can add a lot of speed and certainty to that process.”
Suspicion of fraud is one. Proving fraud is a time-consuming and costly process. How do you prevent your departments from being overloaded with claim reports?
“In the first place: if an insured is being fraudulent and you have evidence, you need to investigate. But before that, there are a lot of steps in the claims handling process when I think: the claim could have been verified at this point as well. It is better to confront the insured earlier with the facts. You don’t need a fraud investigator for that. A claims handler can also ask: ‘we can see several types of damage that are similar. Please explain this.’ If the insured doesn’t have a valid story then you don’t pay out.
You could also suggest paying an exceptionally low amount. If the insured claims 500 euros, but you only cover 100 euros. That’s the sort of bargaining that should take place.”
As an insurer, do you want to know everything?
“That’s a moral question. That’s the conflict we often have between a fraud department and a claims department. The claims department will say: ‘I want to deal with the claim as quickly and efficiently as possible. I don’t need to know things that aren’t relevant.’ While an investigator will say, ‘I want to see everything.’ It’s a mix. During the process of implementing FRISS, we get a lot of historical data showing fraudulent claims that could have been detected. Then we can fine-tune the system to the capacity available. But at that moment you also indicate to that insurer: these are the claims that you have missed. We regularly encounter blind spots in the historical data. We ask insurers if they would like to see this data? Because if they’re going to see it, it will mean half a percent more detection and it will cost them, for example, 2 FTE extra. It’s a constant weighing up: what do you ignore and what do you detect? We ignore a lot of anomalies, but those are conscious choices.”
Again: should you investigate fraud in detail if it is quite unimportant?
“We wrestle with this issue more often. Some insurers even use internal calculation models. If a claims department identifies a fraudulent claim, then this costs money. I think you can also manage the claims process in this respect. When a claims handler thinks: ‘something is obviously going on here: this isn’t true, but I can’t substantiate it.’ Then you could also propose paying an exceptionally low amount. If the insured claims 500 euros, you could only cover 100 euros. That’s the sort of bargaining that should take place. Are you going to start another investigation for this 100 euros? If that insured persists and will increase the amount of his claim because you are bargaining, then of course you should involve the fraud department.”
According to Joosse, insurers can use their role to increase fraud awareness among ordinary claims handlers while at the same time expanding their capabilities. “That’s a theme we apply to every project. Hence the videos. This enables a claims handler to do more than just hand over a claim to an expert. In any case, it should not be a shared responsibility over departments, because then a lot falls between the cracks.”
Do you actually have to re-educate insurers about their claims litigation process?
“We sometimes have discussions about this with insurers. Are they going to try FRISS with one or two employees, or are they already in some kind of claims 2.0 project anyway? We have noticed that many initiatives are already underway at insurers. It is certainly true that our product is many times more efficient when we are involved. It also depends a bit on who the buyer of our product is. If that’s the investigation department, then claims handlers are only informed that they are not allowed to deal with a claim any further. We also have insurers where the claims department is given insight into the detection results and can follow these up itself. We prefer the latter. My main goal is to include authorized media such as photos and videos in the claims process. That’s the message from FRISS I keep spreading.”
Original article based on an interview with AM:‘s Bart van de Laak published in AM: Insurance Innovation Magazine.