From car ownership to transportation as a service
Lyft was founded in 2012 and is one of the largest transportation networks in the US and Canada. As the world shifts away from car ownership to transportation as a service, Lyft is at the forefront of this massive societal change. The transportation network brings together, among others, rideshare bikes, scooters, car rentals and food delivery, all in one app.
Lyft is based in San Francisco, California, and operates across the US and in Canada. It is the second largest ridesharing company in the US after Uber. Curtis Scott is the VP risk and customer platform at Lyft. He is passionate about creating a future where gig workers enjoy secure and independent work. He is convinced that platforms and insurers should partner together to make this a reality. Before joining Lyft, Curtis was an early executive in Uber’s legal and insurance teams.
In our experience, Curtis is one of the few people with a true vision of insurance and the platform economy. We had the pleasure to sit down with him and ask him a few questions on how platforms are changing the data that’s available today and what you can do from an underwriting perspective.
Can you tell us a little bit about yourself?
I started my career as a lawyer, but there are a few too many of those in the United States. That’s why I went back to school and got into product work and I spent 12 years in the insurance industry building out different consumer insurance products before getting into technology. What I really like about the tech side is that you kind of solve problems from the ground up. You build something that does not exist from a whiteboard perspective. So at Lyft I have done everything from data collection and curation, and then over the last six years at both Uber and Lyft, I’ve stood up some of the largest telematics programs in the world. So: how a car is being driven, how fast it breaks turning, and also the largest usage-based insurance programs in the world. So, it’s been a fun journey.
What does ride-sharing have to do with insurance?
Insurance is important to the gig economy. It’s our number one expense for the bottom line of our company. And it’s also something you have to have to get the car on the road legally. Otherwise, we can’t have growth, we can’t expand without insurance. So Lyft’s mission is to improve people’s lives with the world’s best transportation. And literally, that’s what Lyft does: we do the first rideshare.
So, how are you getting around in a car from A to Z? We’re the largest operator of shared bikes in the world. It’s a really large growing business that we’re excited about. We also do scooters and transit.
You can book a train ride and also rent a car to go on a longer trip. So, really we get from A to Z throughout cities. And I think, you know, Lyft has really invested in understanding how to make every ride safe and how to ensure every ride and make sure that it’s a great experience. And so, as I said earlier, insurance is one of our largest costs. So, we have to both manage that as a business. But it’s also part of our narrative of having a good ride. And we want every ride to be safe. So, ideally nothing bad happens. But if something does, insurance is there to help make it better. So, we put the two things together, safety and insurance. And, you know, the goal is obviously never to have an accident. And if we can avoid all the accidents, that would be the ideal outcome. But things do happen when you’re doing millions of rides a day.
How do you tackle this?
First, it looks a little bit like an insurance company, so I have a team of engineers, actuaries, and product managers that work on these problems day in, and day out. It’s not a small thing. This is not just something that’s transactional. I’m not simply just buying insurance. I’m building the entire risk platform to try to, one: identify where risks are, two: avoid those risks, to try not to have the accident. And then three: price and service the accidents that do happen in a sensible manner. So, I think platforms have really changed insurance over the last few years and that there’s so much more data available than we ever had in the past. Insurance used to be very asymmetrical. The insurance company had all the underwriting data. You went to a broker, they gave you a couple of different prices. You generally pick the lowest price and that was the transaction. We’re in a different world today, though. That doesn’t really make sense anymore. If we look just at, you know, Lyft, for instance, it’s amazing how much useful data this phone can create. If you think of it, on every trip we have not one phone, the driver’s, but also the riders that get in the car and there’s an accelerometer and gyroscope. And we can get all kinds of interesting data points from this that can be used to predict safety.
It can also be used to underwrite insurance. So, first: what is the driver doing? How long have they been on the platform? What type of vehicle are they using? How have they been rated? I look at other things like the environment. You know, what stage of the trip are they in? Has it rained today? Is their sun in the angle of the eye of the driver as they’re going around? Wildly predictable stuff that we didn’t use to use. And then lastly, the phone itself: is the driver touching the phone, are they squeezing it, tapping it, is it making noises? All of these things we are now using to underwrite the auto insurance risk on the platform. We’re talking about dozens of high-quality data points that you didn’t use to have. If we were to go back just 15 years, it would be (in the US at least): what’s your credit score, and where is the car garaged at night? And usually, people lie about both of those two things. So, this is a much more accurate way and this is a model that we’ve had in production now for five years and we keep refreshing it three times a year to get better and better outcomes. And really it’s a life cycle of, like I said, safety. Where is the risk, and how can we avoid it? And then the accidents that do happen, how can we promptly take care of them at the right price?
What are some lessons that you’ve learned along the way?
Personal versus commercial is a very dated way of looking at the world. I remember when I first started buying insurance at Uber for Auto, I would go to a lot of insurance companies and the president of Commercial Lines and the president of Personal Lines didn’t really know each other. And I had kind of a weird problem. I’ve got a car being driven around by a person, so it’s like personal auto, but I’ve got scale and large risk, so it’s like a commercial, and I kind of need the best of both worlds to make it work. And it took a little while for the insurance industry to change and recognize that, but I credit it. You know, having stood up partnerships now with over 50 insurers in 86 countries, I am amazed at how well it can adapt. I think insurance sometimes gets a bad reputation for being stuffy or old. And I think it’s actually anything but that. It’s the original big data industry. And I think what you find at the end of the day is insurers like to take care of people and like to solve problems, but they need data to understand it. And what we’ve really done at Lyft and prior to that at Uber was to create these partnerships, to explain the wealth of data that are on our platform. And really, we’re now taking what I would call data science and the tech world and marrying it to actuarial science in the insurance world. So, really, how do you know something’s happening? What’s the quality? What’s the statistical significance of risk and pricing?
Where do you see the future?
I think insurance will get much more granular because of this data. People should buy insurance for the time and duration they need at the right price. And we can finally do that for the first time in its existence because of all of the great data that’s available. And it’s very antiquated to buy something for a year just in case, when you may only be using something for a few minutes. I think the insurance industry is changing a lot and really starting to recognize how assets are being utilized differently and that each person has a different risk profile. One thing that we’ve noticed with drivers is: you get better the more you drive. It’s muscle memory, it’s experience. But you also lose this improvement after you take a big break, you have to learn to drive well again. Also, the world is changing faster than ever. We have new risks that are emerging, new companies, new ideas, and we have to respond quicker. We have to insure these risks. The world needs insurance, but we also don’t have ten years to study something and come up with actuarial patterns as we did in the past. The world moves too fast these days, so we move quicker and I think we partner to do that. We have to explain our data. We need to partner with insurers, and I think the world is getting there.
Which insurance company will win in the future?
The company that’s willing to change is going to be the company that wins. There are whole new worlds, whole new underwriting cycles, whole new risks, climate changes, and cyber. The list goes on and on. But I think it’s an exciting time. It’s an exciting renaissance, in my opinion, right now in insurance. And it’s because we’re looking at data again. We’re coming up with new problems, and new solutions in ways we haven’t in the past. One of the first things I built at Lyft was a telematics program. One: we collect data. We had to collect it in real-time to give insights. We had to also create the ability to then share that data in real-time with insurers so we could get credit for it. In terms of what is the pricing? What’s the rate? That same tech, over time, we found other ways to deploy it.
Here is an example of a Lyft driver who was robbed of their car at gunpoint in the United States. And we were able to stand up a safety program based on telematics that allows us in real-time to detect when something bad is happening. So, we saw that this trip has gone off course, where it was supposed to be. We then reach out to the driver to see if there is something wrong. The driver is able to tell us that his car has been stolen. We’re then able to (with our partner ADT) use the geolocation telematic data instantly as to where the car is. Police are notified and the car was recovered in 30 minutes. And this is all something that was done originally for an insurance reason and that we’ve turned over time into a safety product. So, again, you know, one outcome is you pay for the stolen car. Another outcome is you recover the stolen car and try to make that person’s day better. And I was glad in this case we were able to do it. But I think it’s an interesting application, right, of like, you know, someone might say, hey, a boring pricing insurance product has a real-world safety benefit. So, exciting times in insurance!