Leveraging big data to mitigate risk | Zurich Insurance

Kris Shalda is the Senior Vice President and Head of Claims Customer Relationship Management for Zurich North America, where she is responsible for leading a team that manages relationships and service needs of our customers and brokers in partnership with the Business Units, Distribution and Technical Claims. She has been in this role since April 2022.

Shalda joined Zurich in 1997 as a Claims Specialist in Mass Litigation and progressed into roles of increasing responsibility including Chief of Staff and Claims Chief Operations Officer. Just prior to her return to Claims, Shalda led Zurich’s Business Transformation and Strategic Initiatives team where she oversaw the Claiming our Future initiative, led the design and implementation of the operating model transformation, and had responsibility for a resilience program in response to the pandemic’s impact on our business, operations and employees.

Larry serves as Regional Claims Vice President and is responsible for leading the development and delivery of customer-centric claims services that differentiate Zurich and demonstrate a strong commitment to customer satisfaction. In this role, key accountabilities involve developing and providing technology and claim insights to assist Zurich’s National and Middle Market customers in optimizing their total cost of risk.

HOST:

As part of the Zurich North America Communications team, Stephani Gordon finds and shares stories by asking questions that connect people with ideas to pique curiosity, broaden awareness and create communities. Fondly considered a compassionate interrogator, she has coached executive communications for the CEOs of Zurich North America and Zurich Canada, lead C-suite video productions and connected employees with corporate strategy through storytelling and engagement. In addition to hosting this podcast, she unabashedly admits to spending too much time on TikTok in the guise of “anthropological study.”

 

EPISODE TRANSCRIPT:

PLEASE NOTE: This is an edited podcast transcript, capturing speakers with natural speech patterns that may include incomplete sentences and/or asides, grammatical errors, verbal shorthand and some statements that may be less clear in print.

STEPHANI GORDON: Hi, I'm Stephani Gordon. Welcome to the Future of Risk [podcast] presented by Zurich North America. We explore the changing risk and resilience landscape and share insights on the challenges facing businesses to help you meet tomorrow prepared. So, let's start with Business 101. Better data can often drive better business results. If you're a business looking to minimize the negative impact of claims, you probably won't find a better source of insight than your insurance company.

Insurance companies were the original “big data.” In the U.S., Zurich has been accumulating loss information for more than 110 years, because that's the foundation of how we predict future loss and identify new risks. We harness the power of data and provide insights to customers to help them navigate. So, having said that, I want to welcome two guests to our episode today. Kris Shalda is the Senior Vice President of Claims Relationship Management for Zurich North America and Larry Zeiner, is a Claims Vice President and Eastern Region Manager for the Claims Relationship Management for Zurich. So, welcome to both of you.

KRIS SHALDA: Thank you, Stephani.

LARRY ZEINER: Thank you.

GORDON: So, Kris, why don't we start with you? Why don't we talk about the importance of using claims data to improve a company's business. How do your customers use the data that you give them about their claims?

SHALDA: There are a lot of uses for the way in which our customers can use the data. It certainly can help improve their outcomes and their losses. It can drive changes in behavior from their employees or the way in which they operate. It can identify ways in which they might need different coverages to make sure that they're insuring their risk appropriately. Also, give some awareness around trends that we're seeing from their loss perspectives.

GORDON: Great. So, I think ‘claims fraud.’ We are talking about trends and hopefully it's not a trend that a customer is necessarily seeing that could be identified, but that's another significant expense or can be for a business. Can you ever detect potential fraud by just looking through claims data?

SHALDA: We can and actually fraud is a huge risk. [It’s] something that we've identified as a significant potential risk in the near future and long term. So, we can begin to watch patterns that we see in losses when they're reported [and] how they're reported. For example, if losses are reported 60 [to] 90 days after the loss, it may trigger you to start investigating whether or not there's a potential fraud in there.

Everybody's talking about artificial intelligence now and certainly the use of artificial intelligence in the fraud arena has grown significantly. I always like to tell the story that when I first started off in the industry and was handling claims, we usually identified medical fraud by seeing whiteout on a medical report and a name change in a completely different font on the medical record. Well, now with AI, anybody can pull a medical record and easily change it or create it. They can pull x-rays and all different types of medical information to create an injury that may not actually be real.

GORDON: It's crazy to think imagery… AI generated imagery as well. People can create things that do look so real and I can see how that's going to be a huge issue as we try to help our Adjusters and Claims professionals, our Fraud Investigators [to] keep up with that dynamic changing use of AI. Are there current trends in fraud that you make customers aware of? You mentioned AI, I think you said, “it's an increasing risk,” which is disconcerting. But what's driving that?

SHALDA: There's a lot of factors that drive it. I think the overall sentiment in the American population is ‘what's the difference if I get a little extra, even if I'm not fully owed it?’ So, that sort of perspective on how people view corporations and the way it impacts their perspective on committing fraud. And it might be something as simple as sending in multiple bills when they know they've already been paid.

So, we do utilize various models and predictors to try to identify where that behavior can happen. I mean, even from the perspective of not necessarily being fraud, but something that we do have to keep an eye on is an injured worker not wanting to return to work when they're ready to return to work. So, how we use our data to help predict the injury, the type of treatment and the type of ways that we can get them back to work quicker is important because every dollar matters to our customers. And that's also lost time from an employee perspective at their company.

GORDON: And those matters of fraud, actually, that impacts a customer's claims loss costs, right?

SHALDA: Absolutely. Fraud is a big impact to a customer's loss costs.

GORDON: So, then Kris, let's talk a little bit about how you guys share data and insights with customers because just “raw data” in the aggregate can be absolutely overwhelming. So, how do you get that in a boil down in a way that's meaningful for customers?

SHALDA: There are a lot of ways in which we can do that. I mean, we have self-service tools and capabilities that allow our customers to actually look at their losses and see the outcomes. And ways in which we can identify how their losses compare to others in their industries [and] trends year over year, as well. I think it's important for us to make sure that we educate our customers on what their data is telling us.

It's one of the areas in which Larry, his team and others on our team spend a majority of their time-sharing data information with their customers. By the way, not just sharing the data, but helping them think about what are some of the ways we can improve their outcomes, whether it be engaging with others in the organization, or spending more time with them from a safety and protocol perspective.

GORDON: So, Larry, I think that you recently finished a project that I think Kris was alluding to in terms of improving the data insights that we can give customers. Can you share a little bit of that work?

ZEINER: Yes, thank you, Stephani. The core enhancements that we've made include importing the customer's exposure data, and a creation of a universe of benchmarks against like customers in our Zurich portfolio. In making these enhancements, it allows us and the customer to better understand and act on their cost of risk opportunities or to reinforce superior performance. This can be done over the life of the policy or simply until the last claim is closed.

GORDON: So, you're talking about “benchmarking,” “industry benchmarking” that's been added at this point, right?

ZEINER: We've studied loss outcomes and loss types and developed 23 core benchmarks which heavily influence the customer's cost of risk. This particular enhancement allows the customer to see their performance against a universe of Zurich customers in the same industry. We do this through matching the customer's standard industry code with like customer codes. In doing this, we allow the customer to place in context where opportunity may exist to enhance their cost of risk.

GORDON: What kind of things can you see now that you didn't see before you were bringing in all these other components?

ZEINER: Great question. Often, we see changes in customer's costs. Without the customer's exposure and benchmark data, it's difficult to determine material changes in their business giving rise to greater or different exposures. For example, a customer acquires an entity whose business is different than their core business. Exposure and benchmarks allow us to better understand those differences and place in context their expected performance and opportunities to continually enhance their cost of risk.

GORDON: So, Larry, I think it goes without saying, but there's value in looking across the holistic organizational data, right? You mentioned incorporating payroll into the conversation. How does something like that impact the way a customer considers its coverage or its exposure?

ZEINER: Holistic data is, I think, a challenge in our personal life. If we think about it, this is true in all data sets that we experience. Our world is awash in data and it's becoming increasingly difficult to find meaningful insight specific to insurance frequency exposure and severity are often treated independent of one another. Our core enhancements integrate and deepen the insights associated with exposure, frequency, and severity, allowing the customer to understand the best path to improving their cost of risk and to making sound investments that will correlate to those improvements.

GORDON: So, they're better able to pinpoint and even predict some of what they could be preparing for in the future. Right?

ZEINER: Correct. Let me share an example. We have a customer that was experiencing a rise in their general liability costs. Using the exposure data and benchmarks, it led us to an opportunity on “foreign substance” claims or simply people slipping and falling on foreign substances. Through the combined work of our risk engineers, the customer's safety and risk management team and our defense lawyers, we were able to tighten up the company's operational procedures dealing with foreign substances. The net impact of that reduced the frequency of claims as well as the severity of claims. So, without that holistic view, it wouldn't have been possible for us to fully understand that opportunity.

GORDON: I think that's a great example. Thank you. I do want to follow up with a question about risk engineering insights but first I have to ask, what is a “foreign substance” in terms of a slip and fall hazard? I can't imagine what that is.

ZEINER: So, imagine yourself in a store and on the floor, there is water or oil… or food or some other substance that doesn't typically belong on that floor and if you can think about it in a retail setting, there's lots of things that get dropped on the floor that exposure is significant for their patrons… for the customer's patrons. And as they enter the store, oftentimes there is an accident that follows. So, it's incumbent upon the customer to have tight procedures around spill cleanup, notice and helping their patrons avoid that particular exposure.

GORDON: Got it and that makes complete sense. It's just foreign substance sounded like, I thought, “Ooh, this is something exotic. What's exotic on the floor?” But sure, you're right. Water, fruit, other things also don't belong on the floor, fair enough. (laugh). So, I wanted to come back. You mentioned risk engineering insights as being one of the data streams that you just talked about. What kind of data is that? Is it purely reactive after a claim incident, or is any of it predictive?

ZEINER: So, Stephani, as you know, we have predictive models that operate behind our data. Using these tools, for example, it allows us to understand that infancy of claim, the propensity of a large case. So, if we think about a case just being newly reported to Zurich, it allows us to forecast out how that case will grow and how that case will become expensive. On the prevention side, on the risk engineering side, it allows us to understand the increase in the number of adverse claims at the report of loss. That then prompts the engagement of our risk engineers to work with our customer on safety and prevention actions.

GORDON: Got it. Kris, I think you have another example that shows how data can drive your decisions in a different potential direction.

SHALDA: Yes, one that's really outside of the insurance realm that I like to share with people is: we all have credit cards, or at least probably 90% of the population has a credit card. And the way credit card companies predict your behavior really helps prevent fraud. And so, I'm sure we've all been subject to a large transaction going through on our credit card. I know I personally have experienced this and getting the warning, “we have not approved this transaction because we don't think it's you.” Most of the time it is me buying something quite large, but I appreciate the fact that the credit card company is using data about my spending habits to help prevent future fraud. And I think oftentimes we don't think about that actually is a data-driven decision trying to predict an outcome.

GORDON: That's, I love that. That's a fantastic example actually. And it made me think when my son was in high school, he was a student musician and we had to go buy a tuxedo and going to a men's clothing store and buying a tuxedo was clearly not on my profile (laugh) and the bank flagged it. They're like, “We don't think this is you. This is not typical behavior of you.” So that's a fantastic example. What are some of the more educated decisions you've seen customers make as a result of getting this data? And I know I think it's relatively new that you've been combining all these data sets. So, I don't know if you have a proof-of-concept story yet, but have you seen any behavior changes as a result of this more holistic company view?

ZEINER: Stephani, yes, but in practice we've been using this data. We didn't have the advanced tools that we now have. So, our advanced tools allow us to simplify the gathering and analysis. But as a matter of practice, Zurich has been doing this for many decades. It's just been a much more laborious effort to get us here. But let me talk about more future states. So, our tool allows us to create a platform for customer specific actions. We don't see or have a one size fits all as no two customers are alike.

Examples that may come to mind include how a customer responds to an incident, their techniques, their actions. How customers invest in controls through people or process or technology, or as simple as helping the customer understand the importance of reporting a loss timely. So, the platform allows us to build custom actions depending on the needs and exposures and the opportunities for those customers to invest in actions that will improve their cost of risk in the future.

GORDON: Just out of curiosity, does your model let you change some of the parameters to see what potential outcomes could be if the customers made certain changes in their risk profile?

ZEINER: [That is] an awesome question. Yes, we are continually working towards that future state. We can do it today; we have the ability to blackboard where we can add and subtract that would create sort of a different path and a different outcome for our customer. Our future state, these enhancements set us up for that in a better way in the future. So, that is the outcome we're working towards. So, in essence, putting on the customer's desktop, the opportunity to experiment. If I invest $10 here in these actions, will this create a return of say $30 down the road? That's our future state, but we needed to take this step first to help us get there.

GORDON: Sure, that makes sense. That's really interesting that that will also be an incredibly powerful predictive tool when we get there. [My] last question, earlier you mentioned something about your building and benchmarking capabilities. Can you talk a little bit about what that looks like and why benchmarking matters?

ZEINER: We often have a question that comes from our customers and we ask the question of ourselves as well. How is this customer's performance stacking up against customers in the Zurich portfolio? And that question leads us to specific opportunities and specific solutions that I mentioned earlier.

So, in building in these benchmark capabilities, it provides us real time instant access allowing us to essentially open up the customer's performance and decide where those opportunities might exist, so that when we sit down with our customers, we have a very educated approach to here's the improvement opportunity or here's the gap, or conversely, here is where our customer is excelling and what can we do to ensure that excellence continues.

GORDON: Stop me if I'm wrong but what I think I'm hearing is, let's say customer scores “X, Y, Z,” a score of 10 and maybe they improve that score by 10 points the following year and that might be really great for them. But if you look at them in the bigger data set compared to their peers, you might say, “Yeah, that 10-point increase is great, but you're still 10 points behind the norm of your peer group.” Or you could say, “Yeah, that 10-point improvement is great, but you were already 20 points ahead of your peers in this particular dynamic.” Is that fair?

ZEINER: That is true. Where we see it to be most powerful and where we're often asked in this area to support our customers is they need to seek investment to make changes in their operations, to improve their outcomes. And like all businesses there is a need to fight for that capital. And so, the benchmarking allows us to put in context what that opportunity, first and foremost, is and what the magnitude of that is. And so, as that risk manager or executive within our customer's operation is pitching their story, it's a foundation to build on the need out there. That's where we see the greatest power.

GORDON: Got it. Got it. Makes good sense.

SHALDA: Stephani, can I add something there?

GORDON: Absolutely, Kris.

SHALDA: Real quick. I think Larry makes an important point. We're all looking for what is the return on our investment? And hopefully through some of the changes and updates that we've made to our benchmarking capabilities and the self-service capabilities we're giving our customers, it makes that story easier for them to tell. Because we all know, even as consumers [that] we may take an investment in making sure that our children know how to drive well before we put them behind a wheel of a car because we're trying to prevent a loss. So, it might take a little bit more of an investment from us in the long term with hopefully good outcomes.

GORDON: Thank you, Kris, that's a very relatable analogy, so appreciate you sharing that. And with that, I want to thank both of you for taking the time to come. Congratulations on all that you've, accomplished and all the new insights that you're going to be able to share with customers and brokers, improve their business. Like you said, their investments, their decisions, and ultimately hopefully their loss outcomes. It's been a pleasure talking to you guys, so thank you very much.

SHALDA: Thank you having us.

ZEINER: Stephani, thank you.

GORDON: And to our listeners, thanks for joining us today and we hope you'll tune in for another edition of Zurich North America's Future of Risk podcast. Thanks.