The Dejargonizer

Auto Insurance Is Stuck In The Past. Here's How AI Can Fix It

August 31, 2023 Amir Mizroch Season 2 Episode 4
Auto Insurance Is Stuck In The Past. Here's How AI Can Fix It
The Dejargonizer
More Info
The Dejargonizer
Auto Insurance Is Stuck In The Past. Here's How AI Can Fix It
Aug 31, 2023 Season 2 Episode 4
Amir Mizroch

How can data and AI make auto insurance fairer and roads safer? We talk with Yonatan Matus of insurtech startup Fairmatic about using mobile sensors to measure driving risk and incentivize fleets to drive more carefully.

Yonatan explains how traditional auto insurance companies use proxies like demographics and credit scores to determine premiums. This can lead to unfair overcharging of safe drivers. It also fails to incentivize risky drivers to improve. 

Fairmatic aims to change this by using mobile sensors and AI to directly measure driving behavior. Their tech analyzes factors like speeding, hard braking, and phone use to score commercial fleet drivers on safety. 

Fleets that work with Fairmatic get insights into their riskiest drivers. They're incentivized via usage-based insurance premiums to coach these drivers and improve safety. In turn, Fairmatic's insured fleets have reduced their crash rate by 25% on average.

We discuss:

- How commercial auto insurance losses have totaled $22 billion over the past decade

- The role commercial fleets play in road safety, driving over 50% of all miles

- Examples of safe vs risky driving behaviors that Fairmatic tracks

- The privacy implications and driver acceptance of fleet telematics

- How Fairmatic's data-driven model positions them for autonomous vehicles

- Why usage-based insurance can accelerate AV adoption by assuring regulators

Yonatan argues that fairer, data-driven insurance models are inevitable. He envisions Fairmatic will become the largest US commercial auto insurer within 10 years. Their usage-based pricing could make roads safer for everyone by incentivizing millions of fleet drivers to improve.

Support the Show.

Listen
Apple Podcasts, Spotify, Google Podcasts, Audible, or anywhere you get podcasts.

Connect
LinkedIn
Twitter
Newsletter

Email: dejargonizerpod@gmail.com

Become a supporter of the show!
Starting at $3/month
Support
Show Notes Transcript

How can data and AI make auto insurance fairer and roads safer? We talk with Yonatan Matus of insurtech startup Fairmatic about using mobile sensors to measure driving risk and incentivize fleets to drive more carefully.

Yonatan explains how traditional auto insurance companies use proxies like demographics and credit scores to determine premiums. This can lead to unfair overcharging of safe drivers. It also fails to incentivize risky drivers to improve. 

Fairmatic aims to change this by using mobile sensors and AI to directly measure driving behavior. Their tech analyzes factors like speeding, hard braking, and phone use to score commercial fleet drivers on safety. 

Fleets that work with Fairmatic get insights into their riskiest drivers. They're incentivized via usage-based insurance premiums to coach these drivers and improve safety. In turn, Fairmatic's insured fleets have reduced their crash rate by 25% on average.

We discuss:

- How commercial auto insurance losses have totaled $22 billion over the past decade

- The role commercial fleets play in road safety, driving over 50% of all miles

- Examples of safe vs risky driving behaviors that Fairmatic tracks

- The privacy implications and driver acceptance of fleet telematics

- How Fairmatic's data-driven model positions them for autonomous vehicles

- Why usage-based insurance can accelerate AV adoption by assuring regulators

Yonatan argues that fairer, data-driven insurance models are inevitable. He envisions Fairmatic will become the largest US commercial auto insurer within 10 years. Their usage-based pricing could make roads safer for everyone by incentivizing millions of fleet drivers to improve.

Support the Show.

Listen
Apple Podcasts, Spotify, Google Podcasts, Audible, or anywhere you get podcasts.

Connect
LinkedIn
Twitter
Newsletter

Email: dejargonizerpod@gmail.com

Amir: 0:14

Hey, this is Amir Mizroch. Welcome to The Dejargonizer. On this episode, we're going on a road trip into the world of auto insurance. Why auto insurance? Well, auto insurance is probably the most familiar type of insurance for most folks. Before you own a house, you typically own a vehicle. And so by law, you need to have it insured. All of us end up paying auto insurance for decades. The sums can add up over our lives to, a lot. But how much do we know about how the auto insurance business works? Why do people who drive carefully pay the same amount of insurance as people who drive carelessly? What if the auto insurance model incentivize us to drive more carefully? Making our roads safer and lowering our insurance costs. Our guide into this important and fascinating topic is Yonetahan Matus, the CEO of Fairmatic. A startup that uses a driver's mobile phones to collect and crunch data about driving performance. It then calculates their order insurance payments accordingly. That's quite a step forward from the way most auto insurance companies still work today, which is by classifying drivers by things like there is zip codes, demographics, and income levels. I'll let Yonatan explain.

Yonatan Matus: 1:43

In the US drivers end up paying significantly more than they should because the insurers can't differentiate between good and bad drivers. And because insurers largely can't measure driver behavior directly, they need to use proxy signals to guess who's a good driver and who's a bad driver, and price them accordingly. And the result is that they're making a lot of mistakes. So you can have a very safe driver that's overpaying and a very unsafe driver that's underpaying. just isn't cool. It's not fair, but there's also an aspect of it, which gives the wrong incentive for drivers because if there was a way for drivers that are significantly safer to pay less, and drivers that are significantly worse pay more, people will have a good reason to be safer on the road.

Amir: 2:35

Can I just ask you to give us a quick differentiation of, I mean, it's, it might seem completely obvious, right? What is the difference between a, a safe driver and an unsafe driver? obviously speed is one thing.

Yonatan Matus: 2:46

Yeah, I mean, look, at the end of the day, what we care about is what is the likelihood that someone will be in a crash within, let's say a year or a million miles, and there are behaviors that lead to that. So for instance, you said speeding, distracted phone use, hard breaks, sharp turns, tailgating. There, there's a variety of different behaviors that are risky. And when you conjoin all of these together and you look at any specific driver, or in the case of Fairmatic and what we do, any fleet, you can see which driver or groups of drivers are more likely statistically to be to be in a crash. And that model is then built into what actuaries refer to as a risk model or an actuary model. In traditional insurance, because you don't have access to these driving behavior that I mentioned, you need to fill in the gaps with all sorts of other signals like. Demographics, how old are you and where do you live? Where do you park your car? Are you a graduate of high school or a PhD? what's your average annual income? Are you married? These are all things that statistically might correlate to actual driving behavior, for the majority of policies driven, we're talking about hundreds of billions of dollars. These are secondary signal, and so I'll give you an example: A young driver who just graduated from college and decides to buy a red Corvette will be assumed to be three to four times more risky than same driver a few years later, or the same driver if they bought a minivan

Amir: 4:28

a red Corvette, means they want to speed, they want to have fun, right? He's not just gonna buy, and that raises the likelihood of risk. Is that what's happening?

Yonatan Matus: 4:37

there is a correlation, but not a causation here. Insurance companies looked at thousands and thousands of claims, and they saw that most young adults that are male and drive red converts, in their twenties have a higher likelihood of being in a crash. But obviously, let's say eight out of 10 behave that way, there's two that are actually very safe and actually paying much more than they should, just because they fall into this demographic bucket. And if the insurance company has actual information about how they behave, they could price those two fairly, but right now they are paying a lot more because they're similar on the outside to someone

Amir: 5:17

if it's just a question of being charged fairly, but overall it kind of works out because only a, let's say a small percentage is being charged unfairly, is it really a big deal? Is it, or, or what's the real hair on fire problem there?

Yonatan Matus: 5:33

So I think it's a big problem. We're talking about a product that regulators require you to buy, and it might work for the insurance company because on average they turn a profit, but for the individuals it doesn't work. more than that, I gave you the example of the red Corvette and the young adult to make it really intuitive. But insurance companies often use signals that are weird. Like for instance, people with low credit scores end up paying much more than people with perfect credit scores. Now, I might have been a rich kid, born into a rich upper class family, never had to deal with student debts and so on. I might be a reckless driver, but I'd pay less than an African American kid who grew up in slums, who is very diligent, hardworking, and risk averse person who's driving really safely, but just because of his financial upbringing and historical situation would end up paying much more. And this point, you start asking questions, is insurance becoming something that hurts minorities and hurts disadvantaged groups in society, and that's an even bigger problem than the young adult driving a red Corvette.

Amir: 6:46

It sounds like you're saying that there's, a system that is calibrated, to give auto insurance companies, profits based on data that is, not directly related to how the person's driving over time, but more about where they're from, where they studied, where their parents were from, things like that. This has, financial impact one, but it has a much bigger impact on perpetuating, I guess, uh, structures. Does that sound right?

Yonatan Matus: 7:17

Yeah, so, If you were an airline that wanted to charge based on how much space an individual takes on their seat because you want to optimize how you place people on, the plane and seats that have more space are more expensive. Um, one way to go about that is to measure how tall a person is and that will be objective, and that's the equivalent of actually measuring driver behavior. But another way is to look at statistical tables, historical tables and say, oh, people in the Nordics people from this or that country, from Holland, tend to be taller. And so we're just not gonna measure people and we're just gonna say, Dutch people are going to pay more because they're Dutch. And that's effectively what's happening in different pockets of auto The

Amir: 8:06

Dutch are not gonna like that at all. I wanna now zoom in on fleets. We're talking delivery of all sorts trucks, post, food

Yonatan Matus: 8:15

people around. taxis, non-emergency medical transport. half of the miles driven on the road globally, regardless of geography, are driven by professional fleets. So if you think about the world economy, so much of our GDP. depends on moving goods and things and services from one point to the other. And that activity by law is required to have commercial auto insurance coverage. And so to the degree that you don't want to be in a crash with a truck find out that you're on your own, you want to have an insurance company that will help cover your medical cost or vehicle replacement cost, it's quite useful. The reason why I'm interested in commercial fleet insurance is that due to the large footprint of commercial miles driven, it's an opportunity to make roads safer. If you can make the way fleets behave on the road safer, you can make overall road safe for us and for our kids.

Amir: 9:20

At least half of the miles driven. it affects, I guess everything in terms of, um, price of insurance, the cost of goods, everything is through fleets.

Yonatan Matus: 9:28

There's all sorts of different niches there that move people or goods, but it can also be people that are service provider. Your plumber, when they get into their F-150 of equipment in the back and they drive to your house to help you unclog your toilet. They are doing commercial driving right now and they're covered by commercial

Amir: 9:48

What is an F1 50?

Yonatan Matus: 9:50

F150, for those of us who not living in the US it's the big, uh, Ford pickup truck that typically construction workers and plumbers and others use to haul their

Amir: 10:00

Even within the fleets the insurance, which is mandatory is, not real time. Where would you say that the commercial fleet auto insurance industry is on a scale of one to 10, that it's, starting from zero, which is pen and paper, all the way to a few of the kind of newer things that you guys are working on in, in your r and d.

Yonatan Matus: 10:22

look, I have a ton of respect for the industry for our competitors and partners, but overall commercial fleet insurance is behind and there's a lot to improve there, even the insurers themselves are not making money. you have a massive industry, tens of billions of dollars, over the last decade, lost 22 billion. it's not just that there are fleets that are safe, that are paying for fleets that are unsafe, or fleets that invest a lot in proper operations and, and safety and so on, and don't get any credit for it. It's also that the insurers themselves are kind of in the dark and aren't able to turn a profit, with exception of a handful. And so you have this system that is unsustainable and the result of that is that insurance companies, in order to catch up with previous years losses, end up increasing prices every year by about 10%, and it's been like that for a decade.

Amir: 11:14

How much, um, does your average, fleet pay for auto insurance every year?

Yonatan Matus: 11:20

There are so many sub-segments in commercial auto, because when you move pizza, It's very different from moving a person to the hospital in terms of the risk exposure you have, and moreover, when you move a person to the hospital in California versus in Texas, the laws and how litigious state is, might mean that can be a very different business.

Amir: 11:44

So I just wanna understand this. If I'm an ambulance driver and gone and picked up a person I need to rush them the emergency room. Let's say it takes 15 minutes, 20 minutes. If I'm a pizza delivery I need to get this pizza or a few pizzas dropped off before they're canceled, before it gets cold before or whatever is there a risk lever here?

Yonatan Matus: 12:14

The minute you have a passenger in the back, you have the risk of that passenger getting hurt and suing you. When you have pizza in the back, the most exposure you have is the cost of the pizza or the unhappy customer. And then fixing the car, whatever damage you cause but when there's another person in the back, especially if that person has a medical condition, and so they might be fragile for other reasons, if you get into a crash, the cost of fixing the situation can be significantly higher.

Amir: 12:47

Better get pizza. okay. were on a road trip here through auto insurance, what are the technological challenges that Fairmatic has had to overcome? you're not a hardware company. you sell, software how does it plug in? How does it generate the data? Where do you get the data from? What do you do with it? What are the challenges towards that?

Yonatan Matus: 13:08

Fairmatic is an insurance provider, but unlike other insurance provider, we're also building tech products and we bundle them together. the tech products that we build are doing the direct measurement of driver behavior. By measuring driver behavior directly, we can tell the fleet owner, Hey, Joe is a great driver, you want to encourage Joe to continue doing what they're doing. So, But Nancy is reckless. She's on the phone all the time. She's speeding, she's taking side streets instead of going on the highway, whatever it is, you really wanna coach her. And we give our fleets the tools, coaching advice and so on. And we also give them financial incentive to go and fix Nancy's behavior. And improve her driving such that she looks more and more like Joe in terms of driving risk. the result, because we're using technology and we give financial incentive for those fleets to do what's right, result is that reckless driving drops significantly. And as a result of that, the fleets that are insured by Fairmatic save a lot of money. Because we only charge them what's fair. It's in our name, Fairmatic. So we charge'em what's fair for how they drive.

Amir: 14:25

Joe he speeds less. He doesn't break as hard, doesn't use the phone too much. Nancy on the other hand, speeds more, uses the, brakes more, is more distracted on her phone. How, are you measuring those things and what, what other things are you measuring?

Yonatan Matus: 14:48

Stop sign compliance, how sharp the turns are, which roads Nancy or Joe choose to take and when, do they keep enough distance and so on. the way we measure that is through pretty powerful technology that uses the one supercomputer that everyone has, and that supercomputer is the smartphone. It has, you know, GPS and accelerometer and gyroscope variety of really powerful sensors, and it can run AI algorithms on the fly in real time and understand what's happening on the road, and then score that driving behavior in real time. using that score, we can tell that Joe is safer than Nancy and by how much, or if Nancy reacts well to coaching and improves, we can tell her boss that she's now as good as Joe or even better. On the insurance side, the same data then dynamically changes the pricing that the fleet pays per month for Joe. rather than paying once per year for Joe and Nancy and everyone else on the fleet, a fee that doesn't change and doesn't improve over time, now the fleet has an opportunity to improve Nancy and get all of the drivers to be more like Joe and therefore save a lot of money.

Amir: 16:01

Okay, I'm gonna throw you a little bit of a curve ball, but I'm sure you completely have practiced this. What if safe driving? Bumps up against their duties for the day, and they're not getting in time too much traffic, whatever it is. And, and they need to get their job done. And by the way, there is a shortage of drivers, truck drivers, haul drivers, all sorts of drivers, really. how are you looking at that kind of balancing on the one hand, fewer people having to do more work, spend more hours in the car, that gets them tired. They're cutting potentially corners because their pay and their bonuses, are dependent on doing their job. How does that fit in with safe driving?

Yonatan Matus: 16:40

Our mission is to make roads safe. What we care about is safety. But in some cases there are fleets that choose to optimize only for other things. you know what? If there's a fleet that strategically decides they're gonna be unsafe, but they'll get to every delivery five minutes before their ETA, good for them. I don't want to be their insurer, and I don't think it's a good fleet to have in my city, but it's a free country. I believe that they'll eventually go out of business because they'll be in so many incidents at the end of the day, operational efficiency. Isn't as correlated with driving very aggressively. It's correlated with taking better routes, better customer service, staying within the speed limit not picking up your phone and going on Twitter, doesn't mean that you're gonna be less efficient.

Amir: 17:23

Okay, there are two things I wanna do before we get to the end. This idea of being tracked, the way you drive, where you go, do you stop at home? Do you go visit people? Whatever your life on the road is in your office, big brother. How are people reacting to this? How are drivers reacting to this?

Yonatan Matus: 17:44

So one of the reasons that I chose to focus on commercial driving commercial auto is that. In the consumer side, analyzing driver behavior really opens up a lot of privacy issues and concerns. But on the commercial side, fleets and drivers have been monitoring driving for decades, it's a part of your professional duty, it's not a new concept.

Amir: 18:07

What do you mean by professional duty?

Yonatan Matus: 18:08

If you're a driver and your job is a driver then if you're reckless and you're destroying your company vehicle and their brand and racking up traffic tickets and violations, you're probably not in the right job. But if you are just commuting to the office and back then you're just doing what you gotta do and if you're aggressive on the road, sure it'll be nice if you behave better. I would say it's important for you to behave better, but no one can really force you to do that. It's not a part of your professional duty if you're a dentist, to also be a safe driver. the concept of safe driving as something that professional drivers should commit to is endemic to that profession. And you know, there are 7 million long haul truck drivers in the US, it's the number one most common job title in the US. We're talking about a lot of people, you want those people driving many, many miles be as safe as possible.

Amir: 19:07

Okay, so that's a way of saying that most of the drivers in the fleets actually accept this as, as a good thing. It's actually better for them.

Yonatan Matus: 19:16

They accept it as a necessary part of doing their job.

Amir: 19:20

Okay. In the last couple of minutes, I wanna take us, into the next chapter, The future We've started by looking at auto insurance, we saw how a lot of that is based on data that is irrelevant or not as relevant as, um, live actual driving behavior. Take us into, the next couple of years. How long until you think there is a visible change

Yonatan Matus: 19:49

When you say visible change, do you mean in terms of actual driving behavior on the road or the way insurers treat this opportunity? Or

Amir: 19:56

I guess both.

Yonatan Matus: 19:57

so commercial fleet insurance in the US at least, has seen a lot of losses, and this is not due to lack of trying to fix the problem. Insurers have tried all sorts of different things all sorts of solution for understanding driver behavior. It's called telematics insurers aren't set up well enough to understand that information, to ingest it and to use it for fair pricing. I believe that in the next decade will still not be well set up to deal with this. In terms of actual driving behavior change, that there is going to be, and I already see that change, and I believe that the catalyst for the change will be companies like Fairmatic that actually give fleets a reward when they are safe. You know, we're growing very rapidly and so whether you're looking at the biggest player the most promising startups, there's going to be an acceleration or growth in fleets that choose to be rewarded for safety and a shrinkage of fleets that would rather just pay dumb prices or flat prices. And I believe that in the next decade, Fairmatic is going to become the largest commercial auto insurance company in the US, hopefully will also have a very meaningful impact on how people drive on the road when fleets work with us their driver safety improved by 25% And so if we can become a large enough insurance company to have millions of drivers have the right incentive to do the right thing on the road, we will meaningfully make roads safe.

Amir: 21:24

Amazing. What about autonomous fleets I, you knew this question was coming, but it seems to me like within the next five years, 10 years, there's gonna be a lot more driverless trucks, cars, pizza, delivery, ambulances. Maybe, maybe not. how does your business deal with that? How do you think about that?

Yonatan Matus: 21:43

Yeah, that's a great question and one that's very popular with investors and, candidate employees and so on. I think it's inevitable that autonomous vehicles will come, but smarter people than me said that the future is not gonna be equally distributed. And so, It's gonna take a long, long time until we get to ubiquity. Until then, you're gonna have pockets. You might have Palo Alto or somewhere in Arizona where there's a lot of autonomous vehicles, but it's gonna take a long, long time for this thing to really cover every every place in the, in the world, um, and replace existing vehicles. And we're talking decades. During those decades, we have a big business to build. But even if you say, Hey, it's gonna be five years, your projections are off, it's just gonna be five years. The kind of insurance company that will know how to handle a world drivers aren't there is not the company that asks you, are you married? Where do you park your car? And how big is your shoe? Or which college you went to? It's gonna be a company that knows how to use rich sensor data. It knows how to build APIs and to have one AI talk to another AI and that company is Fairmatic. in my opinion, it's premature for us to go and build a lot of that technology cause the market is not there yet. But when that happens, we're gonna be in a great position to capitalize on the

Amir: 22:54

It's so interesting because you'll have the data that shows what is considered safe and efficient, driving. from what I'm hearing from autonomous car companies, is that the computers aren't the problem. it's the humans other cars. take us into the future. 10, 15 years. how does Fairmatic fit into this kind of autonomous world?

Yonatan Matus: 23:15

Fairmatic will have both a big portion of the actual fleet drivers out there, and that information will be streaming live and will help educate us on ambient risk of any autonomous vehicle going on a specific road because of, as you mentioned, the most dangerous thing on the road is humans in vehicles. and so we will have a really good understanding of that. and on the autonomous vehicle side, we will have SDK software development key that will run on those vehicles and will analyze the same kind of sensor input that those vehicles see in real time will be able to analyze how decisions that autonomous vehicle is making conform with safe or unsafe environmental risks, the companies that operate these fleets will pay us by the second or by the mile, based on actual exposure we will analyze it in real time, just like we're doing with human drivers.

Amir: 24:06

And if we tie this right to the end, why should people who are not interested in fleets and who would just have their car and just pay their own car insurance. What's important for them in this? What's at stake?

Yonatan Matus: 24:25

One of the biggest challenges for autonomous vehicles to start operating in different communities is how regulators perceive the risk of those vehicles driving around. And so if you have really good insurance coverage that is not hand-wavy and is actually looking at the actual risk based on sensor data and based on massive amounts of historical data, then regulators can feel more comfortable that the communities that these autonomous vehicles operate in will be safe. And the financial liabilities of taking risks are well handled, and so we will help bring autonomous vehicles to ubiquity to communities that will otherwise not accept this technology. And for the average consumer, that's a very good thing. you'll get your goods and your services delivered to you more cheaply and uh, faster.

Amir: 25:13

Great. listen, I, I feel like I've just scratched the surface of I've learned a tremendous amount already. Yonatan Matus, Fairmatic, thank you very much for coming on.

Yonatan Matus: 25:23

Thank you so much for having me.