Predictive Analytics

INTRO

On episode 2 of the InsurTech Geek Podcast, talking Predictive Analytics in All Things Innovation with Skip Brechtel. EVP and CIO of CCMSI.

The InsurTech Geek Podcast powered by JBKnowledge is all about technology that is transforming and disrupting the insurance world. We will be interviewing guests and doing deep dives with our own research and development team in technology that we see changing the industry. We are taking you on a journey through insurance tech, so enjoy the ride and geek out!

INTERVIEW

JAMES: Alright! Greetings everybody. We are going to have a great discussion this week. Episode 2 of the InsurTech Geek Podcast. It is the early days, but we are going in with the big hitters. I pulled the guy from the batting line up that should be in that cleanup spot. He is the big hitter. Mr. Skip, Brechtel, EVP, and CIO of CCMSI. Skip, how are you doing today?

SKIP: James, I am doing awesome. Great day, Friday, getting ready to have a long weekend. So, looking forward to our conversation today.

JAMES: It is going to be a good one. We just saw each other. We have seen each other a good bit in the last couple of months. You and I have been working together for years now, about 11 years, but we got to see each other at the RIMS Conference where we met some amazing people, saw some great technology. We got to co–present to a neat forum talking about predictive analytics and that is where we talked about having this discussion on the air so that everybody out there in podcasts listener-land, could talk about, and hear about what is going on at one of the leading TPA’s in the country with predictive analytics, and of course, a lot more than that. But before we jump into that discussion about the current and future tech that we are implementing, I love giving background on our shows guests to our listeners. They can get to know you a little bit better. Now you have got a unique accent. I believe you call it the Irish Channel. You are from the great city of New Orleans, correct?

SKIP: That is correct. I grew up in the inner city of New Orleans. I attended a high school that was a block and a half, from my house. Walked there daily. And then, I attended the Loyola University of New Orleans. I was fortunate enough to get a scholarship for sports. Played D1 college basketball when they let little short people play the game.

JAMES: Exactly. So, someone who was 5”10 would be allowed to play guard.

SKIP: That is correct. So, even played a little forward now and then, sort of crazy. The game has changed dramatically if you would.

JAMES: Yeah. But you loved sports and athletics. I know your family pretty well. They do as well. You had a fun time. Of course, the great thing is it pays for college and helps you get through a phenomenal education. What did you study there in college?

SKIP: I got a degree in business administration, with a degree in accounting. I had a minor in computer science which was very interesting because it was all Kumbell and Fortran and so, embark more on the accounting end of the career because the old computer systems were punch cards at the time if you remember. Well, you will not remember, you are significantly younger.

JAMES: Well, I do remember. So, I Fortran, was one of my first programming languages. I started writing code in the early ’90s. About 91. We learned Fortran 77 and Fortran 90. So, we did some of the foundational, I call them foundational languages. You have to learn to understand where computer science has gone to now. I mean, it is a completely different program. Now when you look at computer science grads coming out of school now than what you and I went through. I as well got a degree in accounting and got a second degree in information systems. So, you and I share that. We have a passion for business and accounting, but also a passion for technology. And it is played a big role, your love for business accounting technology has played a steering role in your entire career, hasn’t it?

SKIP: It has James, I started with Chevron, standardizing all accounts out of college. I was on an executive training program with them. I went out to the home office in San Francisco, to help with the annual consolidation of all of the financial statements for reporting. Everything was manual on paper. This is 1976 so we had walls of spreadsheets from 50 different international companies. It was crazy how we used to do things back then. Then, technically, you got into the early IBM systems. System 32 and some of these others, but again, a lot of that was to do invoicing, billing, not the degree of technology that we can do now, in so many different areas.

JAMES: Yeah, it is dramatic. It is almost hard to remember a time in business when we did not have computers. My father, who you know well, built, a large Teflon company, he started in 1979. He sold it in 2004, and it was not until the last 2 years of a 25-year business that he put a computer on his desk. Believe it or not. I cannot even wrap my brain around this. For 22 years, he built one of the largest producers of Teflon in the country without a computer and no email. At a 1-800 number, use the fax machine. I mean, big innovation at the time. Fax machine. He said they were able to cut order times down by days by using fax machines. They were early adopters on faxing because of how much time it saved, and he loved it. How much time it sped up the sales cycle! And so, it is interesting, and it is hard to remember a time when we did things in paper. What attracted you to the insurance business, because certainly, you did not study that in college?

SKIP: No, when I left Chevron, I went into the oil field service industry. They asked me to go abroad and I decided that was not my cup of tea. We had just had our first child. So, I went with the oil field service company. Originally as a controller and then eventually I became vice president and general manager, and one of my jobs was the procurement of insurance, so we had a marine exposures lodge, trucking exposures, property. Cyber was not even a consideration back then, but again, I started with the procurement, became good friends with the insurance brokers that I was with and low and behold, 15 years later, I wound up joining his firm as the COO of his insurance broker and he had a small TPA where I became president of the third-party administrator, which I at the time had no understanding about what a TPA was. And that was in the early ’90s if you would.

JAMES: So that is big, and those of you out there in listener land who are in the insurance space know that is a major transition to go from buying insurance to running a TPA. For the uninitiated, what is one of the major tasks of a third-party administrator?

SKIP: Well, again, in the worker’s comp and property and casualty, we are approached by several different types of clients. Large individual, self-insured, fortune 500 companies, insurance companies that we handle claims for, captives. So, you deal with a broad spectrum of clients that each has specific needs, specific requirements, and we try to make sure that we do a customized approach versus a cookie cart approach. A school fund in Michigan is different than a large waste management company that has operations in 40 states throughout the country.

JAMES: Which is different than an auto dealers group.

SKIP: Right.

JAMES: And you are administering claims, so when people slip, trip, or fall, or there is an accident or there is a major liability event, you are the first phone call, right? I mean, you are the ones on the front line receiving that report of injury, or the report of the incident and grabbing hold of that claim and trying to get people healthy, get them back to work, get everything paid out. You are the frontline of the insurance business and the claim side, right?

SKIP: That is correct. And we are a little bit different than a number of our competitors because we grew, in the late ’70s of off self-insurance and group funds. So, we underwrite and do policy issuance on over 200 000 000 in premium, where we take the submissions in from brokers, do underwriting, policy issuance, loss control, and then when there is a claim that happens, we handle the claims. That was where we started. That is where CCMSI, the little sector that it grew out of, but now we are handling large deductible programs for clients with many different coverages, etcetera. So, it is very, as I said, knowing it from the policy issuance end to the claims and, it takes us through the complete cycle.

JAMES: Yeah. And that is unique. Now, there is more TPA’s that do that, but it was not the time where you are acting as a carrier without taking the risk on.

SKIP: Correct.

JAMES: You are doing all of the processes, whether it is binding the policy, underwriting the policy, working with the broker when you are doing all of the steps, the difference is, you do not carry the paper, right?

SKIP: That is correct. And, as technologies advance, you can get significant enhancements in all of those areas. Underwriting and policy issuances, all the way through the claims app.

JAMES: How has the TPA business changed? Let us try and set technology aside maybe from customer expectations, what they expect of a TPA. How has that changed from when you first took over in the early ’90s, running this TPA to now, 2019? What has changed there?

SKIP: Well, I would say in the old days, in the early ’90s, we were looking at primarily as a claims handlers. We would be the ones to handle claims for the clients and their focus was to ensure that we did a great job of doing that. As times have changed, and now I consider us, a data–driven company as much as a claims company. Clients are significantly more interested in their data. Using their data to help drive better results for their programs.

JAMES: You are at some level a data aggregator because of how many different clients you touch in the granularity of the data you have access to. I think that is evidenced by the number of entities like States and carriers who are now placing significantly higher demands on you to produce data them right?

SKIP: Well, not only the States, you are correct, but we were also, all of the various carriers have different data and data requirements. So, we have to make sure our applications and data can handle the carrier–specific needs as well as the state–specific needs.

JAMES: Yeah. And so, you have seen, what was, I have got to ask this question. We are going to jump into predictive analytics in the second. What was the most interesting project or interesting things you were asked to administer and, just your career in the claims administration business, which now spans what, almost 30 years?

SKIP: Yeah, almost 30 years.

JAMES: I mean, you are a pro. I mean, I have learned more from you about the insurance business than arguably anybody else I have ever worked with because you are such a phenomenal teacher. But you have had some weird stuff and some interesting stuff you have had to administrate. What are some of the more interesting projects you had to work on?

SKIP: Well, let us see. Going back in time, we have had some large government programs that were spread out through a complete state and it was interesting, just going across the state of Louisiana way back when visiting the local parishes and seeing the different needs, etcetera. And then Louisiana where you have four or five different dialects of language within the same state.

JAMES: Yeah.

SKIP: From way down South Louisiana, New Orleans you got a different type. So, just trying to go across the state and understand the different components of the program and how people thought about them was very, very unique.

JAMES: Yeah. Louisiana is a fascinating state. I mean, I was born and raised there. As were you. You stayed there. I went to Texas but leaving and going to Texas gave me even more perspective on the unique nature of just dealing with people. And in the TPA business, you deal with a lot of people, all the claimants. You have to be able to understand them. You have to able to communicate with them. It is a big challenge. Did you ever get involved in any hurricane claims?

SKIP: Well, the probably other unique one was the, not hurricane claims, but the BP Oil Spill.

JAMES: There you go.

SKIP: Thee BP oil spill happened, and the state of Louisiana hired us to be the data aggregator. Different companies were handling the claims. But the state was trying to get their arms around all of the claims and all of the industries. So, we worked with a large national firm to help build the models for each industry, for the loss of business shoes and business income. So, that was an interesting project because we dealt with the shrimping industry. We dealt with the hotel–motel industry, building models to go back and help establish their claims for the loss of business shoes generated from the oil spill.

JAMES: And that was a pretty expensive claim, wasn’t it?

SKIP: Let us put it like this. I think they are still paying out at this time, almost 10 years later.

JAMES: Oh, my goodness.

SKIP: Yeah, it was somewhere around 3.1 9 million barrels of oil that spilled into the Gulf. Did technology play a role in your ability to deal with the Deepwater Horizon oil spill?

SKIP: We were one of the few people that could take data from multiple sources, aggregated together, and then provide the state of Louisiana back with the information they needed by industry where they could assess what the losses were for each segment of the industry and each individual insured.

JAMES: Yeah. So, the answer is yes. Data aggregation, the ability to pull the data together. Now let us kind of jump to today because you have entered this, you started out administering claims originally TPAs administered claims on paper. Then, we went to managing them digitally. Then we went to digital workflow software. Then you took, then you went paperless. That was a big thing that you all early on just to dump the paper and start routing digital documents. Then you digitize the workflow. Where you are at now though to me seems like completely next–level technology cause you are not just processing data anymore. You are using these decades of information you have been storing and capturing to start making much better decisions that are founded in fact data and history rather than gut feel. And so, talk to me about predictive analytics and what it is to you and the role you think is going to play going forward.

SKIP: Well, the insurance industry is sort of late to this game, quite honestly. Lots of other industries, automated cars, we can go into so many different areas, but the insurance industry has had tons and tons of data that they collect from claims, etcetera, but to a large degree we were not utilizing that data to the degree it could be used to help in providing better predictions on the probability of certain things happening or jumping on a claim because it had the characteristics that would indicate it was going to be a high–cost claim. So, from my perspective, back in 2010, we did an internal analysis on a large national gaming company that we did that had had, thousands of claims a year. And we were just trying to determine the effect, that comorbidity might be having on the ultimate cost of claims. Needless to say, the machine learning that we have now, even in 2010, was not existing to the level. So, we went through file notes, etcetera, and the term if a claimant adds 2 or more comorbidities, and all the characteristics of the claim were pretty similar, the cost of that claim was about 10 times more.

JAMES: For the uninitiated out there, just what are the common comorbidity factors?

SKIP: Well, we were talking about obesity. We are talking about diabetes; we are talking about smoking. We are talking about cancer, heart issues, arthritis conditions, etcetera. There is a list of about 8 or 10 of the major comorbidity factors that go into the analysis.

JAMES: And logically speaking, you would think, well, yeah, that makes sense, right? But you do not have data and evidence to prove exactly what type of impact a comorbidity factor can have on a claim. And so that is what has been interesting for me as an outside observer to what you are doing and involved technically in implementation, that when you start peeling back the history of the data in these claims. What helps is that you guys have done such a good job of tracking data historically, because you can say, hey, here are all the claims that had good outcomes, here are the claims that had bad outcomes, here are expensive claims, here is an inexpensive claim, and here are all the factors that led up to them.

You have to have that data so you can teach your model, and once your model learns what is good and bad and what is a good outcome about an outcome and expensive and inexpensive, then it can start identifying. What I like to say, Skip, and you have heard me say this before, causality versus correlation. Just because someone happened to be obese does not necessarily mean that caused their claim to be expensive. It might just be a correlation, but it might not be causality cause causality is not that truly caused the outcome. And that to me is what you are trying to drive towards is, here is the correlation, but here is the real causality. The result is what? That you can set reserves better administer the claim, better, get the person back to work faster?

SKIP: Well, yeah. What you are trying to do is identify problematic claims or potential problematic claims, high–risk claims, high–cost claims. You are trying to identify them early on. So, our adjusters, supervisors, and clients can jointly be more proactive in that claim to hopefully drive a better outcome for the injured claimant, which in turn should drive a better financial outcome from a claim cost perspective. And I will give you a perfect example. And this goes way back when we were doing the analysis. There was a restaurant worker that was obese. So about 5”8, 350 pounds. It was just a little slip and fall with a sprained ankle. Well, that person, 32 years old, wound up having diabetes, having other comorbidity factors, and ultimately a minor slip and fall that resulted in a sprained ankle, wound up being an amputation because of complications from diabetes, which was not work–related, but ultimately that claim came in over $500,000. So, you can easily, in a situation like that, understand, how factors not necessarily related to the causation, but are impacted by the medical condition of the claimant has a dramatic impact on the claim.

JAMES: Yeah. And medicals can be a big bugaboo.

SKIP: Correct.

JAMES: Because the human body is so complicated. You are not sure when an incident occurs, exactly what outside factors that were preexisting conditions, will impact the outcome of that claim. No 2 claims are alike, and no 2 claimants are alike. So, you have to constantly look at all the criteria. Has all this resulted in you guys collecting a lot more information?

SKIP: Well we did. So, from that comorbidity study back in 2010 and 2013, we created a tool called a claim risk assessment. And again remember, originally this information was going in paper, then you started having filed notes, but you did not have data fields where you could easily data mine. And you were not in an area at that time that mined data notes effectively. So, we developed a tool called claim risk assessment where we added about 70 new data fields, distance to work, pain threshold. I am just giving you a couple here, the probability that you think you will return to work. And we developed an internal algorithm that would score claims, and primarily we were doing this with just indemnity claims and rating those claims is high, moderate, or low–risk claims. So, our adjusters would, again, this would normally be done during the recorded statement, which would happen in the first 10 to 15 days of a claim, which would give us an early indication of some of the drivers of the claim that we thought would help our adjusters do a better job and proactive claims management.

JAMES: But that was all human assumptions. You did a study that was powered by humans, that was runned by humans, and then at the end of the day, you produced a formula that was baked in your software, and that was based on the results of this comorbidity study. The game kind of changes when you move to predictive analytics because now you are feeding this engine, the data, and then it is now saying, hey, there is some stuff you never thought about that is driving your claim costs.

SKIP: So again, I think we all in the technology space have been trying to keep up with the rapid increase in the use of machine learning. A lot of people call it artificial intelligence, but it is machine learning where the systems we utilize, can learn so much from the data and be proactive in providing you analysis of that data. We also started looking at some other alternatives in 2016, 2017. Early 2018, we signed an agreement with a company called Gradient AI. At the time, they were owned by Milliman. They had a 26 million claim, a work comp dataset that we could tap into, provide our data, build models off of it and, provide just some outstanding analysis. Both from the probability of future treatment, that will be needed on the claim, the comorbidities that are impacting that claim, and the probability of this claim having a high dollar threshold. So, we started life with them in early 2019 where all of our work has now come into a product that we are extremely excited about.

JAMES: Has it resulted in any surprises where some claims are identified as risky claims that under your traditional human analysis just would have been missed?

SKIP: Well, one of the things that I think surprises us is how quickly they start putting or signing a nee incurred value, 30 days into a claim. Now they do this both for medical only and indemnity claims because they feel like before assigning a value, it needs 30 days of information to set that value. But what surprised us, and we have looked at the analytics, is that, as good as our adjusters are, it appears that the artificial intelligence generates during that, at least from day 30 through, maybe day 150 with the adjuster catching up. Normally at about 180 days into a claim, the adjusters and artificial intelligence are more in sync with their expectations on the ultimate cost of the claim. But, again, as we talked about earlier, getting on problematic claims very early, normally drives better outcomes. So, if the machine of the software can assist our adjusters in identifying these problematic claims earlier, we feel we will get better outcomes and better financial results.

JAMES: Yeah. It is amazing to me when you look at claim volumes and how hard it is to just stay on top of all the data in insurance. Just the number of data fields, number of claims you have to manage, and you guys do a great job of keeping claim counts per adjuster in line. Even with that, all the work you do, to keep an appropriate claim account and give your adjuster’s time to manage a claim, it is so difficult to catch everything. And that is to me, one of the biggest things here is that you finally have some robotic assistance, not just processing the data and presenting a screen you can click on, but saying, hey, you might want to check this out because I think that it is going to have some problems. And that is where this comes into play for me, is that you have got this area of machine learning, which is a subset of AI. It is a specific form of AI that is going to deliver, I think leaps and bounds of volume. The other area that fascinated me, Skip, is all the plain text images and videos. So, we call that, unstructured data. Data that was not in fields that you can now tap into using a predictive analytics engine. You can tap into years of these diary entries and start to peel useful information out of them.

SKIP: James, I could not agree with you more. We push every data field in our system, every file note because now the technology can read the file notes, pick out keywords, phrases, everything. Cause the adjusters to put a ton of information in there. We push every medical bill, every PBM bill. But where the future is going, quite honestly, we are already discussing the possibility of using voice in the analysis as well, where when you do the recorded statement, or even at the initial call–in of a 1-800 report of injury, it is amazing the technology can now pick out key factors from a person’s voice that might be providing false statements, that we all know has an impact in the cost of a claim.

JAMES: Sure. Yeah. If they are lying to you, you might have a problem. It is interesting. So, truth analysis. Also, another subset of machine learning and AI where you are trying to detect intent tone. If you ever watched the TV series “Lie to me?” It was all about using machines to identify facial features and voice features when someone is lying to you. So certainly, you are trying to detect that, but also just read using a machine to read through 2 decades of diary entries and 2 decades of notes that were impossible for a human to sit and read through and say, hey, we have got some commonality here. We have got a real issue. Whenever an adjuster starts talking about this on a claim, we got a problem. That to me is probably one of the most powerful areas I have seen you utilize here. Have you seen a lot of people in the insurance space doing this? Or do you consider this a pretty early adoption?

SKIP: I can tell you in our discussions with Gradient, we are the first TPA that I am aware that is doing this. I think a lot of our competitors in the industry in general, are trying to get their arms around this. I think they are all in agreement that machine learning and AI, can assist, but we have not seen much from our competition that is delivering the kind of information that we are going to be delivering here.

JAMES: Do you think all of this is going to lead to the ability to start taking simpler claims and auto adjudicating them? Where you can just say, hey, look, we have got a whole bunch of claims that are pretty simple that we can just automatically payout?

SKIP: Well, let us put it like this. Right now, we are looking at this to be an additional tool for our adjusters and our supervisors. So basically, utilized to see if there is a major variance between the estimate that the AI is predicting versus what our adjusters have. But let us be real about this. As the machine gets more information and, the analytics and the AI will continue getting better. In the future, I do see that there could be the potential for, especially in the medical–only space, doing a medical–only claim, almost on the auto adjudication basis. We are not there yet, but it could happen. It will happen.

JAMES: Yeah. Well, it will happen. We are already seeing it in the residential property business. We are seeing it in personal lines auto where there are a certain class and category of claims that just get automatically paid out now. I am very careful about all of my talks and all of my podcasts that I do. We are not talking about human replacement. We are talking about human augmentation here. We were talking about giving people the ability to do higher thinking tasks. We are not talking about replacing a sea of adjusters. We are talking about taking those same adjusters and making them way more effective. And we are talking about taking those same adjusters and allowing them to potentially manage, the high volume, nasty claims and spend a lot of time on those to manage them. Those things that need human interaction and offloading the menial work on the machines that can handle that for them so they can focus their time and efforts on getting people back to work faster and taking care of the customer. And so, I just want to make it clear for all of our listeners, we are not talking about human replacement. We are talking about human augmentation.

SKIP: I would agree. And I would state, in the old days, James, the carriers did so much training of adjusters, etcetera. The carriers are not handling as many claims as they did. A lot of them are sorting outsource claims. And the adjuster pool, it is very difficult to find quality experienced adjusters out there. So, I do not see this as a replacement, as you said, but it will assist organizations to maintain quality adjusters from a staffing perspective.

JAMES: Yeah. It is going to be awesome. I think it would be awesome for the quality of life of some of these workers that you just have millions of workers in the insurance business that are kind of bogged down with tasks that can be automated so they can free their day up to think, and to use their God–given talents to reduce the cost of risk and add value to the client. And so, it is an exciting future. For me now, this cannot be the only tech you are excited about. I know that you have a lot of different technology that gets you geeked out. So, what else out there that you have been seeing over the last few years do you think is a game–changer?

SKIP: Well, I am going to wrap up the artificial intelligence for the adjusters with the next phase, which is recommending interventions back to the adjuster, like is this an appropriate time to do an IME? Is this the appropriate time that you get a field case manager, a TCM, and what is the financial benefit of doing that? That is coming as well. So not just doing predictions, but just recommended treatment interventions or claim handle interventions back to the adjuster. That is coming shortly after, probably within the next 6 months or so. A couple of the other areas, that so many clients now are capturing extremely large video surveillance, it is the storage of information that we are talking about, utilizing data. Well, we have got to store the data and having the ability to store all of these large video files, and so the cloud has been a phenomenal new source for housing a lot of this data at much less expensive costs, that will allow both the client and the TPA to retrieve that information extremely quickly, at a low cost.

JAMES: Awesome stuff. Let us talk about the future of insurance. Cause we are starting to see technology companies that are starting to carry risk, instead of just the technology company being a vendor to the insurance business. You are starting to see the technology company takeover risk operations and carry risk. Do you think that this is a harbinger of some significant change in the fundamental makeup of the insurance business?

SKIP: Well, I see new companies, new carriers entering the space. A lot of them seem to be very, very data driven. They are using the data, not only for claims but to do a large portion of the underwriting, to identify what they feel is the best and most profitable risk. So, yes, I see the industry changing a good bit, and that is why we are now, as I started early saying, we are data–driven companies as much as anything else.

JAMES: For me, it is feasible to see a world in which you have a series of companies that are simply incapable of leaping to being data–driven organizations that carry or manage risk. And you have a series of companies that are started as fundamentally started as data and technology organizations, that happen to carry or manage risk as one of their functions. And I think you are going to see, I am going to say that you are, I think we are going to see some pretty significant change in the makeup of the market. I think that is why you see a lot of the big insurance carriers have started technology venture funds, and they are starting to invest in the very companies that are going to disrupt their business models. Because they are almost acknowledging, it will be challenging for them to change it from within. And so, they are investing in the very companies that are doing it so they can at least have equity ownership and of course one day merge those companies into their operations. You are seeing, big players in the insurance space become InsureTech venture players.

And I have seen venture funds from some of the largest guys that we see at RIMS every year. Or putting together InsureTech venture funds and they are saying, hey, if we cannot drive the innovation internally fast enough, maybe we can invest in the next people that will, so we can still play a relevant role going forward in this data–driven process. The other area I am seeing like I am a drone pilot, I love flying, I love flying planes, I love flying drones and Veri-fly, is one of those interesting providers that just sell you insurance by the minute. So, you are starting to see this interesting fractionalized market where I can log in on my drone app and buy 60 minutes of coverage in a specific geography to a specific altitude for a specific purpose.

And when I got involved in insurance, almost 20 years ago, I would not have imagined a world in which I could buy 60 minutes of insurance. But you can now. You can buy it yourself. It binds immediately. It issues the policy. It uses your GPS on your phone to know where you are. It knows what kind of drone you have because it is connected to your drone. It is incredible how intelligent this is. And if you go through some of the modern underwriting processes for residential and renter’s insurance, they are pulling on 20 different public data sources. You are not answering any questions hardly on underwriting because they are answering all of the questions on your house for you. And so, it, to me, Skip it, it seems like a world in which guys like you are going to operate well because you are used to running a data–driven organization, but some just will not be able to leap.

SKIP: Well, again, we have always embraced technology to help us be better as a company and as an organization. We are all about our people. But if we can make our people better and perform better by giving them the tools to do their job, and technology is extremely important in our space to do that, then we need to do it. I mean, Telehealth is coming to the comp space. So, just think about this. You have got video, or you can have a video coming back and communication coming back from the claimant and the Telehealth provider. Again, just think about the ability to drop that into an artificial intelligence tool that will assist and analyzing that claimant’s attitude, they claimants desire to return to work, etcetera. It is just going to continue, by leaps and bounds because of the enhanced capabilities of technology.

JAMES: Yeah. There was an interesting article today, Skip about Google duplex. Google Duplex is Google’s voice tool that allows their machine to call restaurants and salons and set appointments for people. And I am a big believer in the consumerization of IT, where that consumer tech creeps into business tech and it starts in consumer land, and then it moves into businesses. And so, this particular chatbot is a voice chatbot that can call people and talk to them, even has little nuances of human speech, like an aha and, hm. And, they have made it so realistic that many people say on the receiving end, they cannot tell if it is a person or a machine.

And, there is an article that recently came out that Google was using about using humans for about 25% of those calls. So, if a call kind of went sideways and the machine could not deal with it, it would bring humans in to finish the call, but about 75% of these calls can be handled entirely by the machine. And, and so a lot of those same types of technologies will we use exactly like what you just mentioned. That is giving you the ability to analyze what they are saying and how they are saying and provide recommendations and trigger key phrases. I mean, it is going to be an interesting next 10 years in the insurance space. And I think you would agree the pace of change is certainly accelerating over what we saw from 1990 to 2000.

SKIP: Well, it is rapidly changing. I think almost all organizations are trying to figure out how they can use technology to get better. Some are doing it better than others, and those would be the companies that take the lead in the industry.

JAMES: Let us wrap up our final comments on benchmarking. Because it was in the past, you did not have a great way of benchmarking how you are doing in the insurance business. You have NIC that has the $0 million first dollar, data. But not a great data set on a large deductible self-insured retention. And so, benchmarking has become a hot topic for you, so you can identify how your customers are doing versus the data universe, not just how they did it. Because what you see a lot of times Skip is, customers and you look at how you did compare to yourself the previous year because that is the easiest comparison to make. You already have all the data.

SKIP: Benchmarking internally tends to be easy. I mean, you can see what my results were this year versus last year, what was my cost per claim, etcetera. But that is great benchmarking yourself, but how are you benchmarking against your peers. Are you performing as well as or better than your peers? It is critical to try and get that data and get data that you were talking about self-insured. NCCI is what I call more of a guaranteed cost, smaller client, but large deductible clients and large self-insured retentions, those clients tend to be much more sophisticated. They are taking a large portion of that risk themselves. So, ensuring that their programs are driving the results and keeping them as competitive or even more competitive than the people they compete with, is critical. So, I think benchmarking, it is great to have the AI to help you drive results and you know you are getting better results internally. But again, other people are pushing to get better as well. So how are you staying as competitive or you beating your peers?

JAMES: Yeah, and again when you benchmark against yourself, sometimes you can give yourself the most improved awards. I was in the Corps Cadets in the day and we had an award for the outfit that was the most improved, and no one called it that award. Everyone in the Corps called it the you sucked last year award. And to me, that is what benchmarking against yourself is like, what if you just sucked last year? Like what if you were just really bad, and now you are just mostly bad? And that is concerning. And that is what self-benchmarking can drive is this irrational belief, not rooted in the reality that you are doing well compared to everyone, not just compared to yourself.

And so, we worked on a pretty awesome project called a comp Mark that allowed for TPS to benchmark themselves against the data universe. And now that project and there is a little too early to announce what that project’s becoming, but that project is going to become something else, something independent for people to build a benchmark, claim performance against a data universe without worrying about their information going public. It is a confidential secure system. And so, that is what is exciting to me, and this was, you were one of the brainchildren behind this. Where it gave your ability to say, hey, among hospitals in Illinois, how are we doing? Are we doing well or not? And it gives you a truly objective scorecard.

SKIP: Well, if you are a hospital in the Chicago land area versus a hospital in the Southern part of the state, that is two different universes. That is just like, claims in California are significantly more expensive than claims in Indiana. Different laws, different statutes, different regulations. So, when you are benchmarking, just to say, well, how did I compare across the country, you needed to look at it on a state by state basis because each jurisdiction, it is so different. For someone that might have operations in certain areas of the Midwest, their claim results, just because of the area they are operating in, is going to be less than California or less than New York, etcetera. So, you have got a benchmark back to your jurisdiction as well. Very, very critical.

JAMES: Yeah. And what’s interesting, Skip is that the process of benchmarking teaches you about local and data nuances that impact the cost of claims and the cost of risk. Just being able to see, hey, show me, the average cost per claim by state, by this, industry code. It allows you to become a much better, adjuster, much better risk professional. Cause you start to understand not just a gut feeling that things are more expensive in California because let us be honest, everything is more expensive in California. The true data backing to prove that, yeah, claims cost more there. O I started to be able to peel through these years of benchmarking data, it taught me a lot about what drives the cost of risk and how important data and benchmarking are.

SKIP: I could not concur more.

JAMES: Well, Skip, you are, in my opinion, a true visionary and the insurance and technology space and I appreciate the time you took to talk with us today. Any closing comments, or remarks?

SKIP: No man, we need to keep jointly geeking out and trying to figure out better ways to drive better results. And I am one of the old guys in the industry that is still is having a great time trying to come up with new ideas and thoughts. I hope our company and our clients get better results. It is always challenging, but very, very, fun.

JAMES: Yeah. And if you do not know Skip, that little number called his age is maybe going up a little bit, but he is still a, he is still a healthy 21 on the inside. And, do not ever forget that he has more energy than almost anybody I know. And has a passion for innovative technology. But the biggest thing that you have a passion for, that is worth noting, Skip is your passion for people and investing in people. You have invested in me heavily and I appreciate it. You have invested in a lot of other people who owe a lot to you. So, we all appreciate it, and thanks for your time today on the podcast.

SKIP: What a great time is was and everyone has a great day. Take care, everyone!

JAMES: That was our interview with Skip Brechtel, CIO, and EVP at CCMSI talking about predictive analytics, data, and risk.

This has been the InsurTech Geek Podcast powered by JBKnowledge. It is all about technology that is transforming and disrupting the insurance world. I have been your host, James Benham. Thank you for joining us this week. Look forward to talking with you soon.

We are taking you on a journey through insurance tech. So, enjoy the ride and geek out. See you next time.