The risk selection differentiator

Fresh from launching an E&S habitational program alongside its established admitted offerings, Honeycomb is looking to translate its low loss ratio built on data-driven risk selection to other areas, according to founder Itai Ben-Zaken.

What characteristics of the market opportunity in habitational business appealed to Honeycomb?

We’ve had a consistent strategy from the very beginning, setting out to solve a deep challenge in an industry where many insurers have decided to categorically say “no” to certain classes of risks or regions, leaving customers with little to no options. We wanted to take a different approach. We wanted to figure out what is the real root cause of refusal – what is the underlying problem?

We asked ourselves, is California, for example, a place where you cannot write business at all? The answer was “of course not”. Yet, we see many other carriers leaving the state as things start to look bad for them on an overall level. I guess their perspective is that they prefer to just not deal with it, and it might make sense for them. We see it differently; we see it as an opportunity and as a market that could really benefit from our technology and offering.

That’s why we put a significant emphasis on risk selection. We developed a way to examine the millions of properties in California, or in any of the other states we operate in, and use AI to do high-granularity risk selection at the building level and say, “This one is great, I'd love to take it, and the other one might not fit my appetite”. This way we can offer great value, without any blanket limitations.

How do you differentiate through that risk selection and risk pricing?

We look at the individual roofs that we assess, all the data points that we collect about the landlords, the level of maintenance of the property, the remote inspection, and the AI technology that we put in place to assess each property individually. Most of the heavy lifting is done once you have the imagery of the roof and building, but putting all those data points together is what gives us a truly unique point of view and allows us to have a really great offering to customers.

I see us continuing to do exactly that. I t's been proven to be very effective for us in the multifamily property market, where we initially launched. We're running at a 30 percent loss ratio inception to date on an incurred basis, and on an ultimate basis, we expect to end up in the low to mid- 40s.

Deepening that advantage is the key to our continued growth – we don't want to stay where we are. There is so much more to innovate on in this market – from street-level photography analysis to remote-inspection photography analysis, there's a lot more work that can be done to create an even more sophisticated and accurate platform and continue to improve the levels of automation and consistency.

How transferable is the mousetrap you’ve built to other areas?

We’ve built a really strong end-to-end process, and right now, it’s focused on one main vertical: multifamily properties, or landlords and HOAs. But we’re already looking at other classes of business to expand into. We plan to do it gradually, staying close to our core, but there are some great complementary opportunities– like the E&S product we just launched.

Next up, we’ll likely introduce a few more of these complementary offerings, potentially excess liability. We're always thinking about the problem from the customer’s perspective. Our typical customer is a real- estate professional who might own a few apartment buildings, but they probably have other types of real estate too. So, if you think about lessor’s risk-only policies for retail or other types of commercial property, those are probably where we’ll expand next.

We’re also focused on deepening our existing offerings because, honestly, we’re just getting started. There are still plenty of states we’re not live in or have only recently entered

How easy will it be to translate the low loss ratios you’ve generated so far to adjacent products?

In excess liability, we believe it is transferable because so much of it has to do with the level of maintenance of the property, who's living in the property, who is the landlord, how reliable they are, and what reputation they have in the market. Excess liability has a lot of overlap with the underwriting we already do. We already underwrite quite deeply for liability with the current offering where we provide$1mn/$2mn or $2mn/$4mn limits. So, if we were to offer $5 mn excess liability limits, we’ve already done a lot of the groundwork.

There are other verticals and products where we’d need to add another substantial layer to our risk management product and processes. For example, if we start insuring business es where most of the general liability claims come from customer foot traffic– like people entering a store– that’s an area where we’d need to build on top of what we have. But when it comes to the building itself, roofs are a huge component. We’ve invested a lot in that area, and that’s where the industry is seeing major losses. We’re not though, because we’ve done a much better job with roof underwriting and managing exposure to severe convective storms (SCS) in particular. We have a pretty unique approach to that, and it should be easily transferable to any property with a roof.

Tell us more about your approach to SCS and why it’s been successful so far?

The beauty of our approach is that our “secret sauce” is not really a secret. What we are doing with SCS is really executing a lot of the best practices that make common sense, but that are very hard to execute when you're a legacy carrier dealing with multiple lines. Because we focus only on this area, we can dedicate 100 percent of our efforts to building the best system possible for this particular, and sometimes vertical-specific, problem.

It all starts with accurately assessing roofs. We’ve developed the ability to precisely assess the age of a roof, using aerial photography and our unique algorithms. It seems obvious, but it’s something the industry rarely does well.

Not all buildings are the same, and neither are their likely damages from hail. The hail risk varies significantly from building to building, which sounds like common sense but is hard to manage manually, especially with outdated systems. We focus on just this one thing, so we’ve built a system with the granularity needed to compute the hail risk for each individual building. And we built that level of granularity where everything is coded in one system, and the data stays in one place, so we can compute the hail footprint for every building.

We also adjust deductibles and pricing in a more advanced way. Many carriers apply a flat 2 percent deductible for hail damage, but we look at it differently. We say some buildings might only need 1.5 percent, and on these, we can be profitable even with a lower deductible, while others might need a 7 percent deductible. It’s not just about pricing differentiation but also about tailoring deductibles to the specific building.

The last piece is that we also control our radial aggregations very carefully. It’s not a new concept, but we have a system that, in real time, can tell us which assets are insured or quoted, and the moment a quote is bound, it’s automatically updated. At any given time, we know exactly how much hail risk we have in any radial or zone.

So, in essence, our secret sauce is really about execution – being able to do all of this, consistently, at a growing scale, in multiple states that have different characteristics, and with multiple underwriters.