Why We Invested in Insurdata: The Future of Data-Driven (Re)insurance

We’re thrilled to announce our partnership with Insurdata, a Denver and London based team with deep background in risk management building a product that provides property insurers and reinsurers more precise exposure data for catastrophe (cat) models. Welcome Jason, Jeremy, and the Insurdata Team to the Prototype Capital family!

There are some of the fundamental truths we came in with:

1. Data quality issues lead to a massive difference between actual and modeled losses in the (re)insurance industry. According to an RMS 2008 study, “losses from the 2004/5 hurricane seasons showed that as much as 45% of the gap between modelled loss and actual loss came from data quality issues.” As they say in data pipelining, garbage in, garbage out.
2. Exposure data is one of the only directly controllable parts of building a catastrophe model. Currently, insurance exposure data from a single source is used to compute geographic identifiers (mainly using poorly formatted address data to reference interpolated lat/lon points), sometimes leading to (re)insurers analyzing a building that’s many kilometers from the building in question. Information is also not shared within different companies and so uncertainty goes unchecked. Plus, even though a building might flood at an entirely different level to the building next door, they are often assessed and modeled using the same incorrect or generic attributes, resulting in the same risk and premium for, say, flooding.
3. Higher risk leads to higher losses for insurance companies. As a result of incorrect exposure data, (re)insurers are taking on more risk and volatility than necessary and therefore experience actual losses that are materially different from estimates.
4. Improved exposure can lead to lower premiums. If somehow property (re)insurers can improve risk modeling, they can reduce their losses and thus they can reduce premiums and thus they can attract more customers.
5. Better data pipelining tools and computer vision / perception can help improve the precision of these exposure data leading to the aforementioned cascading effect.

We believe Insurdata offers the best solutions to these trends and realizes these industry truths. Insurdata is building a sort of “Bloomberg terminal for exposure data.” When a property insurer or reinsurer uses Insurdata to get their exposure data, instead of relying on a single source of data for geographic identification for that particular (re)insurer, Insurdata has built data pipelining tools to achieve better estimates with multiple streams of data and between multiple organizations. Insurdata will also use a variety of machine perception tools to get very precise points of entry of water (i.e., using image recognition to find what are the exact nonlinear lengths and widths of a building, what is the exact point of entry for the water on the ground, what kind of material is the building made of).

Because of a combination of tools, Insurdata has developed a robust and precise exposure tool which has already attracted large global insurance companies like SCOR and Canopius.

Ultimately though, our partnership and investment into Insurdata was mainly predicated on the team. The CEO, Jason Futers, has been in the (re)insurance industry for over 25 years with a majority of that time at Risk Management Solutions, a company that provides catastrophe model to over 93% of international commercial lenders; at RMS he served a variety of roles including CEO of RMS Japan and Head of Emerging Risk Innovation. Who better to improve how risk management is done than someone who has lived and breathed it for 25 years?

We’re excited to be working with a team that’s applying a modern technology to one of the oldest industries out there, hopefully helping to reduce risk for massive insurance companies and as a result, helping reduce premiums for consumers and small businesses around the world and helping to provide property insurance for the masses.

Let’s build the future, together.

Team Prototype Capital