Free Preview

This is a preview of some of our exclusive, member only content. If you enjoy this article, please consider becoming a member.

Lower reinsurance costs aren’t the most talked-about benefit of introducing AI and automation tools to the desks of property/casualty insurance underwriters, but with chatter about harder reinsurance markets heating up, it might be one to consider more seriously.

Will Ross, chief executive officer and co-founder of Federato, didn’t present his thoughts on reinsurance in exactly that way. But toward the end of an interview centered on multi-eyed underwriters trying to search through third-party data, internal underwriting rules and goals, catastrophe model and aggregation results—and more—Ross highlighted that benefit when asked how carriers gauge the benefits of investing in Federato’s RiskOps, an AI-based platform that provides underwriters with a unified view of all the disparate information they need to select and price risks.

“What we have found is the single biggest return on investment that senior leaders in an organization care about right now [falls] in the category of portfolio management….When you can articulate to your reinsurers that you [the carrier] are going to do what you say you’re going to do” to manage a portfolio, and “that you’re investing in technology to make that actually happen, that is worth credit at the reinsurance table today.

“We have seen that across the board at our large carrier customers and our small MGA customers,” Ross said.

“Insurance operates on a deal-by-deal basis but experiences the results on a portfolio basis.”

Will Ross, Federato

“People are often paying for our product in a single reinsurance negotiation…I think, at the end of the day, [that’s] why the choosers choose us,” he said.

Ross refers to carrier executives who sign the checks to pay for emerging technology as ‘choosers,’ to distinguish them from the ‘users’—people within an underwriting ecosystem ranging from underwriter, their underwriting assistant, that underwriting assistant manager, the back office support function that is offshore—for whom RiskOps was purposefully built.

Ross, who is not an underwriter but instead has a background in machine learning, started to think about underwriters’ needs when he was doing some research focused on wildfire models. And he got a true understanding when Federato conducted 1,200 hours of interviews with underwriters to design the platform to fit their needs.

The Backstory: A Platform Is Born

“I started my career very involved in natural language processing and natural language classification. And then through time I became very interested in climate and atmospheric modeling.” The vision for the company came as Ross and co-founder William Steenbergen, a computational mathematical engineer, started to collaborate on some research at Stanford around wildfire spread modeling. In developing the research, they started to get to know the insurance industry, ultimately doing small projects for some MGAs in particularly catastrophe-exposed parts of the country. “We thought there would be an opportunity to help the insurance industry by building better cat models.”

As they worked on this business plan, they stumbled on an insight into the insurance industry that became a more interesting problem to solve—that what the cat model said an organization should do, and what the organization actually did, typically wasn’t lining up.

“What we realized was that the person who was making the actual decision of what risk came into a portfolio or not and how those risks were priced—the underwriter—was very, very poorly equipped” at the time he or she was making a risk selection decision. Not only didn’t the underwriters have all the information they needed about an individual risk, but they especially did not have the “information that put that risk in the context of a broader portfolio,” Ross said.

“The truth is that cat modeling is done in a silo.”

Megan Bock Zarnoch, Federato

Megan Bock Zarnoch, Federato’s chief operating officer, who has almost two decades of hands-on experience as an underwriter and has served in underwriting leadership roles at major commercial carriers, explains the disconnect. “The truth is that cat modeling is done in a silo” by a data and analytics team or an actuarial team or specific global cat modeling team. “That work is done separately, and the output of that is provided to underwriting leadership—the chief underwriting officer or the head of the business, and the VPs who lead various regions, “on a retrospective basis, essentially quarterly.”

“They’re literally publishing a separate report that says here’s how much capacity you have consumed, here is what would be the modeled expectation for expected annual loss, and here’s how that tracks relative to what we would expect the premium need to be to cover that model,” she said. “What’s not happening is that’s not being sent or communicated to the individual underwriters who are doing each of those transactions on a day-to-day basis. “They’re blind to that,” she said. Instead, the individual underwriters are making decisions based on a high-level directive, such as “increase cat pricing by 10 percent,” which is not specific to the individual policy the underwriter is reviewing.

Ross sees similar problems when underwriters try to meet other portfolio management goals at the individual risk level. An example might involve the instruction, “On renewals, we are trying to take a certain amount of rate on a certain class of business while maintaining a certain level of retention.” For the underwriter, it’s hard to know where the company is “tracking at a portfolio level right now so that I can make the best decision for the risk that’s in front of me at this point in time.”

Essentially RiskOps is closing the gap and meshing the views, he explained. “That portfolio retroactive side is the ‘risk.’ That’s the risk department in an insurer today…Operations really refers to the core—the underwriting operations [department] that is looking deal by deal.”

“Insurance operates on a deal-by-deal basis but experiences results on a portfolio basis,” he said

With RiskOps, when an underwriter is working on a certain account at the desk level, he or she can see information about exposure accumulation, and about where the company wants to be in terms of retention and rate level. “And in context of every single other decision that has also been made by any underwriter in your organization,” Bock Zarnoch points out. “Because the optimal decision on accumulation, for example, is going to vary based on what all your other teammates did…Did they already bind an account that is taking up that additional capacity?”

Underwriting La Cañada High School

The Federato executives and others have described how the proliferation of third-party data sources and predictive models have underwriters drowning in data and mired in non-core manual tasks, eyeing three or more computer monitors and accessing more than a dozen Web-based applications to bind a single account. (Related article: “The Road to 2032: Big Changes Ahead for Commercial Underwriters“)

“What Federato is trying to do is actually simplify that and [automatically] serve up all that information to the underwriters under a single pane of glass,” instead of hunting down 20 different sources of information, Bock Zarnoch said.

According to information on the Federato website, RiskOps also delivers some features common to automated underwriting workflow tools—items like account status, task deadlines, real-time third-party data insights, fitness to appetite, winnability, underwriting authority rules, among other things. But without the portfolio view, “I would argue RiskOps is the only platform available where you can underwrite La Cañada High School,” Ross said, referring to a school in Los Angeles county California located in what it is considered “a very high fire hazard severity zone” threatened by California’s largest 2009 wildfire, the Station Fire (according to Wikipedia). “It is the only platform we have seen that doesn’t involve an underwriter taking 50 different off ramps to 50 different systems to get the job done.”

To get an appreciation for the off ramps, Ross and his team conducted 1,200 hours of interviews with underwriters.

“Businesses don’t have problems; people have problems,” Ross said, calling this this “fundamental thesis” behind the design of RiskOps. “We believe as a company that if you can’t tell me who the user is that’s going to use a product, if you can’t identify their profile on LinkedIn—if they don’t have a consistent job title across hundreds of organizations, or a somewhat similar job title—there is a chance that person doesn’t exist. There’s a chance that the ‘problem’ you’re defining isn’t the problem.”

“We found that it was that core underwriter who is processing submissions, doing a great deal of manual work but trusted to make that ultimate decision of both risk appetite and final price,” he said. That involves more than one LinkedIn title, he affirmed. “Some organizations are going to have field underwriters, and they’re also going to have more home office underwriters who handle renewals, for instance. As you dig in, you start to realize how important the underwriting assistant [is], or assistant underwriter, depending on the organization. You start to realize that at certain organizations, underwriters are called territory managers.” And regional managers and AVPs and SVPs matter, too, he said. “You start to form an understanding of that whole ecosystem,” he said.

The idea was to isolate each individual and spend the time understanding how they work today, and then walking them through how it could look tomorrow and getting their feedback. “What do you think this is for?” is a typical question in that third step, he said. “If you can’t get through that three-stage process with every user within an ecosystem, I would argue you have not built a usable, workable product.”

“This was so core to what we did. I was absolutely intimately involved. I just did an hour of user interviewing with one of our clients first thing this morning,” he reported on the same day he spoke with Carrier Management.

Involved in the meeting were leaders, underwriters, underwriting assistants, underwriting assistant managers and members of the offshore back-office support function, he said, referring to a thesis of Deep Collaboration advanced by one of Federato’s investors, Emergence Capital.

“What deep collaboration really focuses on is the job to be done and a recognition that the best SaaS companies today are facilitating interaction between multiple stakeholders,” or users, he said. While the underwriter is the core user, enabling each user in the ecosystem to touch the process ensures the carrier can underwrite La Cañada high school. “To do that, they have to be able to touch these other people and collaborate with these other people,” Ross said. “It even extends out to the actuaries. If you need a loss pick request, an actuary is the one who’s going to respond…Our user base is much broader than just the underwriter,” even though the underwriter is very much the primary user.

“What we are ultimately selling is this portfolio and deep collaboration aspect of the platform that creates so much more value for [carriers]. Ultimately, that’s why we win deals,” he said.

Related article: The Inside Story: How QBE NA Is Digitally Transforming Its Underwriting Operations
Addressing the ROI question with a discussion of potential reinsurance cost savings, Ross said one claim Federato does not make is that using RiskOps will improve a carrier’s loss ratio. “Loss ratios are inherently volatile. And I would argue that any organization that claims that they can prove how their software decreased the loss ratio is not credible because you would need a hundred years of losses to make a statistically valid representation of that claim.”

“We don’t make that claim—not because we don’t believe it to be true. I think our customers believe it to be true, but it is not provable.”

A Different Pricing Model

Ross believes Federato is breaking new ground by changing the way its technology is delivered.

“One of the things that we have observed in the industry is that if you think about your average policy administration system implementation, it’s potentially hundreds of millions of dollars” and takes somewhere in the five-to-10-year range to complete. “The majority of those dollars are going to services—SI [systems integration] implementers, people who are incentivized to charge you hourly over prolonged periods of time to make a 20-30 percent margin.

According to Ross, Federato is flipping that model. “The raw speed of our implementation process [is] eight weeks, [then] four weeks user acceptance testing, and you go live. We have done that at large carriers and we have done that at tiny little MGAs. It does not change.”

“We charge a flat SaaS fee, and we do not charge for any implementation services. We do not charge for any customization work done through the life of the contract. It’s $0. And we have really had to educate the market. People keep coming back to us, ‘Wait, just to be clear, if I want to do X, there’s not a charge for that?'”

A brochure on Federato’s website states that its solution is sold through a subscription license, and pricing is either user-based or tied to direct written premium.

“We just believe that the business model we have out there is so much better aligning our incentives to actually get the carrier live and provide value versus yet another upcharge, yet another delay, and more billable hours for us,” Ross said. “We really believe that the way we deliver insurance technology is as much an innovation as the product itself.”

Topics Carriers InsurTech Wildfire Tech Data Driven Artificial Intelligence Underwriting Reinsurance Insurance Wholesale Market