How a Data-Driven Approach Boosts the ROI of Risk Control

Pete Miller (00:38)

Thanks for joining us for the Predict & Prevent podcast! In today’s episode, we explore how a global insurer uses data and tech to help clients prevent losses. My guest is Dorothy Doyle, Senior Vice President and General Manager of Risk Control Services at Liberty Mutual Insurance Company.

Dorothy joined Liberty Mutual in 2021, after executive roles in loss control at E.X.L. Service and The Hartford.  Dorothy discusses the evolving role of risk control with increased access to real-time data. She explains the importance of testing new technologies and solutions to ensure effectiveness and ROI. Dorothy emphasizes that having a strong safety culture is crucial for successful risk control. We’ll also explore technologies like machine learning, collaborative robotics, and wearable sensors, that are transforming the way Liberty Mutual and its customers approach risk control.

Pete Miller (01:37)

Dorothy, please tell us a little bit about your role and how Liberty Mutual is helping customers reduce risk.

Dorothy Doyle (01:43)

Sure, so I have the privilege of leading Liberty Mutual’s Risk Control Services Organization. So this is roughly 300 people. We get out of bed every morning to collaborate with our clients, help them understand where their risks are, whether it’s through losses or exposures, and then to develop strategic data-driven multi -year service plans or plans to reduce that exposure or those losses. Ultimately, our team is targeting reductions in total cost of risk or reductions in exposures that lead to total cost of risk.

Pete Miller (02:21)

So can you give us an example of how your team’s risk assessment has helped a client reduce its risk profile?

Dorothy Doyle (02:28)

I mean, where to start? Absolutely. So as an example, we have a large client. It is a retailer and we worked with that client leveraging some in-house proprietary tools. Maybe we can go into those a little bit later, but specifically a tool that helps do an ergonomics assessment on the fly. So we use that tool to evaluate their operations in the front end of their stores, so the bagging department. And just for context, 40 % of their losses were coming from manual material handling in the bagging department. So lifting, pushing, pulling, bringing something up, twisting to load something or unload something from a cart, reaching below the cart for heavier items, et cetera. And using this ergonomics tool, we worked with them to understand the risk profile of how their cashiers were moving and then ultimately redesigned the workflow of their front end of their cashiers experience. We were able to recommend that they add a bagger, which is a fairly basic recommendation, but not something that they had the sponsorship internally to do.

And when they did that, they ended up increasing productivity by 9%, which was not something that we were factoring into our business case. And they also ended up seeing about a $500,000 reduction in losses at each of their locations, and they have thousands of locations.

Pete Miller (04:08)

That’s a big number. That’s awesome. In light of risk control’s historical role in preventing losses, like risk control is not new, but can you tell me how, can you share with us how you see it evolving today?

Dorothy Doyle (04:24)

So, you know, I think risk control, safety, you know, helping businesses be strong and resilient to your point is not new. And ultimately, where we do our best work is where we come in as a consultant or a thought partner, and we understand the risks or the losses that our client is facing.

We understand their culture. We understand their long-term strategic plan. And we work with them to identify potential pre-loss interventions, identify the stakeholders that they need to make any of the changes that we’re going to recommend. We pilot, we test, we evaluate any technology. We determine what do we think from that pilot an ultimate outcome will be. We seek funding, we help them seek funding, and then we help them launch more broadly. And so, you know, the reason I say all of that is that that remains, I think, very much the same.

So when we say how is risk control changing, I think that we have more access now than we have in the past to more real time indicators. Right. So we see things like computer vision or you know, haptics or wearables, right? That indicate when someone is, you know, perhaps lifting in a certain way that, you know, is less than healthy or, you know, someone is, you know, perhaps in an area of a warehouse that is not safe for them to be in, right? So we have this, blossoming of technology that helps us identify risks sooner, but the ultimate process around what do you then do with that, I think is very similar to what it always has been.

Pete Miller (06:18)

So maybe a better set of tools, but sort of the factors are not dissimilar. Would that be a correct statement?

Dorothy Doyle (06:28)

I think that’s correct. You know, the tools are amazing, right? So you see machine learning and AI optimizing operations, predicting, you know, incident risk. You know, we see our clients using collaborative robotics to automate certain tasks, dangerous, dirty, dull tasks. You know, we would call that co -bots. You know, we see video -based computer vision, machine learning, again monitoring for other employees in places in a warehouse that might be unsafe or pieces of an operation that might be unsafe. You know, I could go on and on. So we see these things and the important connection is not that we have the technology that can do it. I mean, that’s amazing, but you have to connect that technology to then a process whereby the operations of that client can benefit from that technology. So, so we saw a thing. We saw a risk happen. So what? How quickly did we act on it? How effectively did we impact the safety culture and environment, right, to take advantage of that new, more quickly available data?

Pete Miller (07:41)

So let’s just talk about data. You mentioned that you have a data -driven focus for preventing losses. So can you tell just a little bit more on what that looks like and how Liberty Mutual is using that data?

Dorothy Doyle (07:53)

Yeah, so there could be a lot of angles to this. And so I’ll share a few. We, as a company, have invested very heavily in data scientists. And so using predictive analytics, identifying unique patterns in claims data that may unlock hidden insights, using operational data on behalf of our clients in conjunction with our claims data, you know, data that’s regularly or, you know, easily available within the industry, can help our clients make pretty important business decisions. And I’ll give you an example of this.

So we had a client who was looking at safety features as an example. So they were looking to upgrade their fleet. And they were looking at what is the benefit of a safety feature over the life of a vehicle. And our data scientist was able to take that client’s vehicle data run it up against Liberty Mutual’s experience on the auto line of business, not just through our commercial experience, but also through our very large personal lines experience. And then was ultimately able to calibrate and say the value of each of these safety features that you’re trying to decide if you’re going to purchase, you know, collectively, if you purchase vehicles with these safety features over the life of the vehicle, that will result in a $3 ,000 save per vehicle over the life of the vehicle. So that’s an example.

We also, we get really excited about forward-looking metrics. So, you know, we have claims data and that’s great, but when you can align operational data with that claims data to say, we know when the claim happened, that’s not the beginning. The beginning is, what were those near misses that occurred before and what in that operational data might be statistically significant when you align it with the claims data to say like this is predictive of claims.

And so like a good one that everyone in the industry would recognize is, we had a client who was experiencing more slips and falls. It was a restaurant client. They were experiencing more slips and falls at one location than another. And so we, they were generous enough to share all sorts of operational data with us. And what we found is that they were cleaning their floors at a different frequency, using a different solution than other restaurants in that, in that chain. And so again, like just really thinking about the power of being able to align client specific operational data with claims data to then say like, can we get more predictive about what might happen here?

Pete Miller (10:45)

That’s fascinating. Sorry. I mean, that’s fascinating. Cleaning the floor. You could get the data to make that relationship.

Dorothy Doyle (10:53)

Yeah, I mean, depending on what the clients capture, you know, we usually ask them, they say like, well, what do you want? We’re like, well, what do you, what do you have? What do you capture? You know, you’d be amazed at what might, what might be relevant here. And, you know, I will say that, that we did Liberty Mutual did win the Business Insurance innovation award in 2023 for predictive analytic consulting as a result of some of that work.

Pete Miller (11:16)

Can you talk about the importance of using a pilot to test technological innovation before pushing out these solutions more widely? And then maybe some of the experiences you’ve had behind that approach.

Dorothy Doyle (11:30)

Sure, absolutely. Part of our consulting process is to identify what our clients are trying to accomplish. So what are they trying to solve? How any intended technology or solution will help them achieve that result and then create small pilots, innovation proof of concepts, where we work with them to develop a statistically sound pilot testing plan and determine the potential ROI of that solution before testing begins. And so that may seem like a lot of words, but what I really mean by that is, you can think you’re piloting something and there can be all kinds of variables that impact your success. And for our clients that we’re working with, we could be ultimately talking about recommending things that are millions and millions of dollars of investment. And so we want to make sure that our pilots are statistically sound.

When we have had success with pilots or when we have seen that process to be effective, we are typically picking a smaller problem or a smaller operation. So as an example, we have a client that we helped them set up a pilot for whether exoskeletons would be useful in helping people stay at work, so reducing frequency of lifting, pushing, pulling type injuries. And whether exoskeletons would be useful with people returning to work.

This was a very large client. They had high aspirations for exoskeletons, you know, in a variety of different roles. And we again, decided to start with, you know, one easily controllable role. We made sure that we had sufficient participation. So they had over 100 people that ultimately decided that they would participate. And then we measured and monitored all kinds of things from you know, experience with the distance from the location that they would be doing the work, to the location the exoskeletons were stored. You know, experience people had feelings that they had around using the exoskeletons throughout the pilot, etc.

And so I bring this one up because as an example, exoskeletons are very expensive. And again, there was a lot of high hopes for many different tasks, but some of what we found in this pilot was, it’s going to sound basic when I say it, but depending on where you store them and depending on how far people have to walk to get them and then put them back and how many of their peers they need to walk by that may have a comment on them using this exoskeleton really matters. And so instead of a broad, you know, launch of, hey, this is an amazing new technology and it can be very helpful at the task, at the point of the task. Instead of, you know, sort of going wholeheartedly into that, we were able to say, like, we did this pilot. We know it’s helpful at the point of the task, but here’s all the other things you need to think about for this ultimately to be successful. Cause if you’re storing it 50 yards away, people aren’t going to use it.

Pete Miller (15:09)

That’s kind of a risk mitigation exercise in the pilot for a risk mitigation solution. Really, right, right?

Dorothy Doyle (15:15)

Yeah, absolutely. I mean, it’s a large investment and you want to, you know, before you make that investment, you want to hash out what are all the hurdles and pain points. Because again, the technology could be amazing, but the hurdles and pain points to getting people to use that technology and then do something different once they do use the technology can be all the difference.

Pete Miller (15:37)

Always people, right? So you talked about pilots. Can you give us some examples where the technology innovation was a success and you evaluated everything and the pilot was a success and you’re able to move it beyond the pilot phase?

Dorothy Doyle (15:52)

We had a client that developed a new piece of equipment with new technology in it to reduce significant injuries that they were experiencing with their workers lifting and loading very heavy objects. And prior to this, they were using something that was very manual. And they had in their mind, we want to create something over time. Nothing existed in the market. They didn’t, they couldn’t buy anything, but we want to create something over time that is accessible to our teams, that is ready and waiting and available when they need it. And that can significantly help reduce lifting injuries. You know, think, you know, a 500-pound item we’re trying to lift with multiple humans from the ground to you know, a four feet, four foot platform or something. Right.

And so we piloted with them, over the course of, I will call it six years. We worked with our client to say, here’s the initial risk. Here’s the initial prototype. Let’s use the prototype. Let’s try it. Let’s see what people are doing with it. Let’s expand it. You know, can we use it in other areas beyond what we thought we would initially find success in? Right. And so ultimately, over the course of six years, significantly improved what the tech was able to do and how accessible it was. And so, you know, at the end of that six-year period, again, it was probably a $25 million investment that the company made to this technology, which would ultimately reduce their leading cause of very large losses.

But that occurred, you know, again, after many, many tests and learns. And, you know, the final thing I’ll say on this story, right, is that our client, the person we were working with knew the importance of making a business case and knew the importance of having people experience the win of something themselves, right? So like, even though we have this amazing technology and it worked, the $25 million investment is significant. And so this person set up a scenario with the leadership of their company to say, here’s this task that your workers are doing. I would like to invite you to try it.

And so the C -suite was there sort of trying this task themselves and realizing like, you know, no matter how many of us there are lifting this very heavy thing, 500 pounds, you know, four feet up in the air is going to be really difficult. Right. And so then by setting up that scenario and then pulling out into the open, this new technology and saying, here’s what we’re asking you to invest in. You know, they created a real emotional connection to help make that business case which is also really critical.

Pete Miller (18:53)

So you talked a little bit about a successful pilot and how you’re getting adoption. Are there other things or other factors that is helping Liberty Mutual get adoptions by clients for risk control and prevention?

Dorothy Doyle (19:11)

One of the more important components to all of these success stories is the willingness and openness of our clients to engage in this kind of activity with us. So what we would call that is safety culture, right? So, you know, we have clients that may have access to all of this and never do anything with it. And we have clients that come to the table hungry every day that really want to understand how they can employ pre-loss strategies to help reduce total cost of risk. So I would say coming to the table with that kind of understanding that When you create a safer environment, you’re creating a more productive environment. You’re impacting your bottom line, et cetera. I think that’s really huge. And so what I will say is in our conversations with our clients, we try to build that. So we focus on the connections between claims and risk financing and insurance placement and pre-loss risk control working together to help impact you know, risk management objectives, but ultimately the bottom line for our clients.

Pete Miller (20:29)

Where do you see technology and AI? AI is obviously a big buzzword now and you hear it everywhere. Where do you see technology and AI adding the most value in risk control?

Dorothy Doyle (20:41)

There’s so much that is going on right now and it’s happening so fast. It’s very exciting. So, you know, I’ll say many things and you can stop me and say that’s enough if you’d like. But, you know, I would say that using machine learning, using AI to optimize operations or staff utilization can be very impactful. And what I mean is, you know, we’ve done studies that show that, and this is not a surprise, right? But we’ve done studies that show that the higher the overtime, the more the risk of loss increases. And so just really thinking about how do we optimize staff, not just for that operational in the moment, but for that over the whole business how do we optimize staff to reduce costs over the entire business, including the cost of someone having an injury?

You know, we see, I mentioned this earlier, but we see a lot of great opportunity for collaborative robotics. So how do you use robots to automate dangerous, dirty, you know, dull tasks? How do you use robotics to monitor for site safety? And then what are the increased risks that you create with those co-bots that you then have to be thinking about and mitigating. We love video-based computer vision, learning, monitoring, whether that’s sort of more the basic telematics, or whether it’s more I’m going to monitor, again, in my warehouse, I may have vehicles that are driving around in my warehouse and I want to make sure that my pedestrians in my warehouse are you know in the areas that they need to be and I want to make sure that my drivers of these vehicles are stopping at stop signs right. You can use computer vision to be monitoring for those you know near misses.

You know, we love the idea of wearable sensing technologies right so for heat stress, which is only going to become more and more of a challenge for the industry for lone workers for, you know, to provide haptic feedback when someone is, you know, potentially lifting something in a way that could be healthier. You know, we, as I mentioned, love predictive analytics and all the things that you can do with operational data combined with claims data.

You know, we spoke a bit about this earlier, but I’ll just say like fatigue sensing. So wearables with fatigue sensing, you know, for drivers, for folks that work in emergency services can be pretty, pretty exciting. Active listening, last one I promise, but you know, it’s an exciting place, right? So we’ve seen, we’ve seen clients using drones for pickup or delivery of small things. So for example, in large hospital systems, we may have some clients that are using drones to pick up prescriptions and deliver prescriptions back and forth from the lab, which actually reduces fleet risk. It creates some other risks that you have to learn about, but reduces fleet risks.

So I think I could keep going honestly, that’s the limit, right?

Pete Miller (24:14)
It’s amazing how many things are available. Is there one that you’re like, wow, that’s really going to be impactful, maybe more than the others? Or does it depend on different sort of exposures?

Dorothy Doyle (24:29)

I think it depends on the exposures. What I will say is that, you know, I’ve talked a lot in this conversation about lifting, pushing, pulling injuries. Liberty Mutual puts out a workplace safety index every year. It ends up being one of our most downloadable or downloaded, excuse me, topics of research. And we know that lifting, pushing, pulling injuries are one of the highest causes of loss across the entire economy, right, from an insurance perspective.

And so, you know, you’ve heard me talk a lot about wearables or robotics or, you know, really thinking about like, how do we identify in the moment when someone is doing something that is not healthy And then how do we think about engineering that option out of their day to day? Right. So, I think the technology that I do get most excited about is probably technology that can help with our largest loss driver.

Pete Miller (25:41)

Makes sense. So you talked about implementing and incorporating technology and AI to obviously, but have you seen a, because we’ve heard a lot of on other episodes of, you know, there’s a, there’s, it’s always, it’s got to be implemented. So is there a process that you’ve seen that’s effective when you’re trying to incorporate, you know, technology and AI to reduce risk?

Dorothy Doyle (26:07)

I mean, there is, I fear I might be repeating something that I said earlier, but you know, what I will say is that identifying the loss drivers, like really understanding what are you trying to do with this technology, right? Because a lot of technology vendors or, you know, amazing inventors, right, will create something and it is bright and shiny and it does do amazing things, but really trying to understand like what specific business problem as, you know, our client has that they think this technology will influence, right? So identifying the lost drivers, identifying what we really think this technology is going to impact, helping them develop that statistically sound pilot or test for will this technology be effective? Introducing that technology in small bite-sized form, bite -sized, you know, chunks and learning from that experience, changing, modifying the approach based on what you see, again, very basic, right?

But then using that experience to show, here’s the return on investment that we believe we can get from this to achieve the funding to then implement that more broadly. That is a huge hurdle, right? And then from that broad launch, continuing to refine over time. I’ll say that we always incorporate change management principles into the project planning. So again, you can have a technology that works. And if it’s 50 yards away from the point that you need to use it at, that may not work. Or if the culture is such that peer pressure would dissuade you from using one of those technologies. That’s something we need to work into the plan to address. And then finally, I’ll say we always involve the people who will be using and impacting, using the technology or impacted by the technology in the study phases and the pilot phases and the testing and the vetting and in the rollout, those closest to the work.

Pete Miller (28:15)

Dorothy, can you tell me the impact of client relationships in your process?

Dorothy Doyle (28:20)

One thing I will call out for our organization, we are very fortunate to have a very engaged client advisory board. So we leverage insights from our clients and what they are working on. We thought partner with them. We have a constant pipeline of emerging trends and emerging issues that we are fortunate enough to be able to partner with them on and dive deep on. And so, you know, the other thing I’ll just say is that as an effective risk control organization, you know, excited to get up every day and help people solve their problems, right? Help businesses stay safe and sound, you know, really learning and teaching and teaching and learning from our clients at the same time is I think critical to the success that we’ve had.

Pete Miller (29:21)

I also like that you’ve, more than most people I’ve talked to, you distill it down to return on investment. Because a lot of people are like, yeah, that’s really good, it’ll save a lot of money. How do you know that? Like intuitively maybe, but actually some things that appear intuitive don’t end up being actually real.

Dorothy Doyle (29:41)

Yeah, we really focus on that, which probably, well, in some ways lengthens the process. But in other ways ensures that the process will be repeated. Take the time to do it right and make sure that you are making a good business decision. Because ultimately, I mean, so, you know, we get out of bed every morning. We want to keep people safe. We want people to go home whole. And ultimately, in order to get a decision approved, it has to be a sound business decision. And so, you know, if you’re not making that case for a return on investment, then you’re doing your workers a disservice.

Pete Miller (30:24)

Dorothy, thank you so much. I really appreciate your time. I certainly learned a lot and just fascinating work you’re doing. So thank you very much. I appreciate it.

Dorothy Doyle (30:33)

Thank you for having me, Pete. It was an honor to be here.