Reading Time: 10 minutes

Around the time this article is published, a whopping 7000 people will descend on Las Vegas to discuss the future of insurance. For three solid days, people from all walks of life who have joined the insurtech wave will introduce themselves, network, and fight to get heard in a sea of innovation. It’s all happening at the industry’s leading conference Insuretech Connect.

Back in 2006, I co-founded an insurance software company when insurance technology was synonymous with modernizing back-office systems written in assembler code.

I’m not kidding.

Insurance core systems were about replacing these legacy systems — characterized by a sea of green screens monitors — with more modern, web versions.

The word insurance was a dirty word in most venture capital circles. I remember participating in one of those speed-dating investor events in 2007.

You know the type I mean with the “two-minute elevator pitch.”

“Hi, John. What’s your elevator pitch?” said a New York-based VC.

(There weren’t many VCs in New York at the time.)

I cleared my throat.

“We aim to transform insurance underwriting,” I said confidently.

His eyes rolled back into his head so fast; I thought he was having a seizure.

“Oof. We just sold one of those. We were so happy to get out — tough market. Good luck to you,” he said sympathetically after listening to my short pitch.

(It was a reinsurance software company they sold to CSC — the former guerilla in the space that has been supplanted by Guidewire — a company founded by a team of insurance novices, just like we were.)

“Rotate,” shouted the master of ceremonies for the event.

I was demoralized.

I went on to pitch over 30 VCs around the country before Flybridge Capital finally gave us a chance. A decade later, we joined the Guidewire juggernaut.

It’s incredible how the world has changed since.

Today, the industry has a new name — Insurtech — and has over $4.5B invested globally from traditional venture firms and carriers alike.

Insurtech goes beyond software providers and includes full-stack insurance companies in P&C, life, and health. It includes data analytics companies, drones, and artificial intelligence platforms. It also encompasses tech platforms with massive data sets at the center.

It’s a whole new world of innovation.

I now live in the background, quietly providing advice from my own experience to the incredible leaders building new ventures in the industry.

I listen and empathetically coach some fascinating people — usually over a meal.

(I have even shared my learnings here.)

One leader I visit with periodically is Carey Anne Nadeau, founder, and CEO of Open Data Nation (ODN).

ODN is an innovative startup that repurposes public data to help save lives on city roads, change the nature of risk underwriting, reduce discrimination, and improve financial efficiency.

Her journey to date is a fascinating one that I’ve often wanted to share.

So, after enjoying a ramen noodle lunch recently, I asked her to indulge us in telling her story.

Our conversation went so well; we decided to make this a two-part series.

In part one, we ask Carey Anne about her founding story, some of her innovative marketing techniques and the power of AI in insurtech.

(My colleague Cybele Ramirez interviewed Carey Anne.)


It’s like a map of the roadway network that lights up red, yellow, or green based on the dangerous nature of the roads. (Image: Carey Anne Nadeau)

CR: Let’s start by sharing a bit about yourself and how you founded ODN?

Carey Anne: Oh, sure. So, I’m the founder and CEO of ODN. I began the company about five years ago now, as a graduate student at MIT. I saw a market opportunity with the increasing proliferation of data sources from state and local governments. I saw this massive data revolution, and I realized that I could take my background in geospatial statistics and urban planning, bring them together and develop a measure of, what we call, the “world of risk.” Essentially meaning, instead of qualifying individuals based on data that they can provide or individual data we can collect about them, we measure the world, measuring where it’s safe to drive or where it’s safe to open a business. We’re attaching that to individual customers so that we can further qualify the risks that they pose to the world, but also that the world poses to them.

CR: What is the problem you solve?

Carey Anne: The risk measure that we create is a measure of how dangerous the roads are, for example, where a traffic crash is likely to happen in the next 30 days. It’s a new measure of risk, something that the insurance industry hasn’t been able to monitor, track, or compare with our customers previously. What it enables them to do is provide risk management insights on the personal lines and commercial lines to deliver suggestions to their customers about how they might be able to change the route to modify their risk exposure. You know, a minivan full of kids driving home from soccer practice, do we want to take route A, that’s the fastest or route B that might be a little bit slower but significantly reduces our exposure to traffic crashes? On the commercial fleet side, it works as a routing technology so that commercial vehicles can route around unsafe intersections and not be at the wrong place at the wrong time.

Our ambition ultimately, and what we’re working towards right now, is to file territory rates with regulators. What this will enable us to do is further refine and modernize the risk rating factors that go into the price you get for auto insurance.

It’s like a map of the roadway network that lights up red, yellow, or green based on the dangerous nature of the roads. (Image: istockphoto)

CR: So, why city data?

Carey Anne: City data is an underutilized data asset, and it is something that I was deeply familiar with. I felt like there was a lot of potential and a lot of power in it. First and foremost, it’s collected by very qualified local experts, right? The Department of Transportation officials have degrees in transportation planning and engineering. They know their data very well. So, we felt if we were going to predict traffic crashes, it was important to qualify any predictive variable we built into our models with those people who really understood what was going on at the local level. From there, I think the other really big benefit of localized data is that it’s very timely and geographically specific.

What I mean by that is, it is collected at a point in time at a specific location. If you have a traffic crash, that crash happened on Tuesday at four o’clock. It happened at this intersection. We can precisely put you there. The reason why that’s so powerful is that it doesn’t rely on preexisting geographies or other political boundaries to qualify where things happened (that can mask or aggregate to a level where it’s not useful).

The information is much more detailed at a lower geographic level. That enables us to develop a hazard map, much like a flood plain map or a hurricane map, in the environmental hazard lines where every road segment gets a risk score. It’s like a map of the roadway network that lights up red, yellow, or green based on the dangerous nature of the roads. It allows us to be that geographically precise. If you’re traveling through this roadway the next 30 days, we know what risks you will likely be exposed to. Whereas, you know, other sorts of historical data sets really focus on collecting data at a zip code level. The number of crashes that might have happened in your zip code is X, but that doesn’t get down to what the risks of the roads are that you’re driving on.

Geospatial statistics is new to the insurance industry for different lines. For environmental hazards we’ve been using geospatial statistics, we know where floodplains are, we know where it’s likely to have heavy rain and sun hail and hurricanes and tornadoes. Great. But what about geospatial statistics for homeowner’s policies? For commercial business policies? For auto policies where we can train dilate data about the risks around the physical establishment or around where we’re driving to the individual? This is the game changer.


CR: So what would you say, then, is the biggest challenge in selling this product to insurance companies?

Carey Anne: Well, so many challenges. I mean, the insurance industry is like a family, right? I feel like the girlfriend that the favorite son brought home.

They’re interested to see behind the curtain. They want to know what you’re doing and want to get to know you. But it takes time to build trust and relationship and really be part of the family. So I think one of the great challenges for us when we were first starting out was building enough trusted relationships to be able to test artificial intelligence solutions with carriers, with reinsurers, with folks that have data about claims. Why that’s so important is that if we do aim to create a risk rating, we definitely need information about claims to be able to optimize our models for loss experiences or filing claims. We need to be able to justify that the math works. Unfortunately, when you’re in a new sort of non-incumbent in the space, your access to that data is limited. When we were able to pierce that threshold and able to start to measure some of the impact and demonstrate in real numbers for the actuarial and underwriting teams that it was worth their time, that is where we really started to see folks build trust, build interest, begin to welcome us to the table a little bit more regularly.


I’m a data nerd, so of course, I get excited about these things, but really we’re asking AI developers to try to build models with their eyes closed. (Image:istockphoto)

CR: Is AI a holy grail for Insurance?

Carey Anne: In the insurance industry, we have a lot of privacy concerns around sharing data about claims and about customers. We’ve institutionalized a group of folks or group of companies that we feel like should have access to this data, regulatorily, and we made them the privileged few folks who can actually estimate and measure the impact of new data on losses and, in this case, crashes. So what that does is, it requires an insurance technology company, or risk technology company, selling data into the insurance industry and trying to build AI models, a lot of iteration in silos rather than iteration and optimization through a traditional sort of data science pipeline of all the data. If I have all of the data, I can adjust or select variables; I can modify to improve the performance.

If I only have a few, I’m taking a shot in the dark and hoping that I hit the bullseye and sometimes you can, but even then, there’s probably an improvement that you can make to that model to make it perform better and optimize the performance a little bit more. But if you keep data in silos the way that the insurance industry today is, it’s really stalling the innovation that we can see. That’s a problem because there are a lot of companies who can’t survive the many years it would take to really pierce the threshold of getting data and being able to play a little bit, be giving you a little bit of liberty, a little bit of leash to let the AI models breathe and grow and optimize.

I’m a data nerd, so of course, I get excited about these things, but really we’re asking AI developers to try to build models with their eyes closed. Not just because these companies are new companies trying to work with established players, but because that’s ultimately what’s going to get you to the model that you need in the model you want to deliver to your customers.

So I mean, I see both perspectives, right? We’re at this fulcrum that says we invested so much and it still hasn’t worked. What I’m saying is we need to double down. We need to learn what we can from our first experience, which is that we need more data to be released and more trust to build it. The amount of research dollars that CEOs and insurance spend today is a bismal. It’s almost reckless how little they spend on innovation where I think they can spend more. The executive at Softbank is leading this charge and setting the tone that the next hundred years, the most important thing you can do is double down on AI, open up your teams, open up your data, and really start to let it breathe a little.


“You know, a minivan full of kids driving home from soccer practice, do we want to take route A, that’s the fastest or route B that might be a little bit slower but significantly reduces our exposure to traffic crashes?”


CR: What innovative marketing approaches are working for your branding efforts?

Carey Anne: So I founded and host a podcast, it’s called The Golf Course. It’s distributed by a really amazing group called The Insurance Nerds, who are a network of folks interested in getting more content about insurance tax, insurance technology, insurance innovation. The Golf Course is intended to be a modern forum for business conversations and insurance, whereas, you know, traditionally, conversations happened in a very exclusive place, invite-only. Well, we’re really flipping the table on its head. We’re saying, look, we want to make this an open space for more folks who wouldn’t necessarily get that invite to the golf course, but who have ideas from outside of the industry, or who are inside of the industry, but are leading transformational change.

So one of the big challenges of podcasts for this industry specifically is that it’s quite difficult to get folks who work for established carriers, who work for established institutions to speak openly, simply because they are required to get approval, maybe from their organization or they’re risk-averse, don’t want to say the wrong thing, don’t want to risk their job for a podcast, which I totally understand and appreciate. So in the spirit of being a CEO and trying to find rules that everyone lives by, that we all don’t have to follow necessarily, I decided to innovate the podcast a bit. We do regular sit down podcasts, but we also do this thing called a dinner party, and we invite ten insurance experts around the table typically around a conference. We anonymize the time so someone won’t give an introduction to themselves, we’ll clip out their name and their association, maybe leave a little bit about what their expertise is and what they’re at the table to talk about, then open the forum with drinks and dinner to really allow people to speak openly about the issues that they know to be true based on their qualified experience. They can debate, have dialogue, disagree and confront one another, which are things you don’t see in most insurance spaces But we’ve invited folks, and I made a safe space for folks to feel like they can say what’s wrong, offer solutions, and really be open to being part of what the future of insurance looks like, rather than feeling like they don’t have a voice and they can’t participate.


Visit part two of our conversation with Carey Anne, where we discuss women and diversity in Insurtech, her CEO hacks, her favorite podcast episode and more.