Creating the Network Effect in Insurance | Bill Bloom, Aviva Canada

by Juan de Castro, COO Cytora

This a shortened version of Making Risk Flow podcast, episode: “Creating the Network Effect in Insurance". In this episode of Making Risk Flow, Juan de Castro speaks with Bill Bloom, a board member at Aviva Canada, Bamboo Insurance, and Foundation Risk Partners. Over his 28-year career, Bill has led the strategic direction at global insurers and consultancies, including Accenture, The Hartford, and Aviva. He has also played a leading role in integrating technology, data, and analytics in insurance operations and claims management.During the course of their fascinating discussion, Juan and Bill discuss why the insurance industry has evolved slowly and the challenges it faces in both underwriting and claims. Bill also shares his insights on the need for digitization, standardisation, and the deployment of artificial intelligence to improve efficiency and customer service too.

Listen to the full episode here

Juan de Castro: Hello, my name is Juan de Castro and you're listening to Making Risk Flow. Every episode, I sit down with my industry-leading guests to demystify digital risk flows, share practical knowledge, and help you use them to unlock scalability in commercial insurance. 

Welcome everybody to another episode of Making Risk Flow. Some people often say that one of the reasons insurance hasn't evolved for many years is because most people in the industry have only worked in insurance and don't often have a broader perspective. But every now and then we find this unique profile of an executive who has done pretty much every role and has one of these one-of-a-kind backgrounds that combines a unique understanding of what challenges insurers face with a strong vision for the future of the industry and a really intimate knowledge of what technology can do for the industry. And today I'm joined by one of them. I'm referring to Bill Bloom, who joins us on the podcast. So thank you so much for joining me today, Bill.

Bill Bloom: Oh, you were talking about me. Thanks, Juan.

Juan de Castro: Tthe reason I say that is because you've advised insurers in a consulting capacity. You've led operations and technology for a number of insurers. In your current non-exec roles, you advise insurers and InsurTechs. And this is one of the reasons that I think it makes your thinking so much more elaborate. So let's start there, give us an overview of your background.

Bill Bloom: Okay, well, thanks so much for the introduction. So it's always interesting, why does somebody get started in this industry? Which, by the way, in a non-joking way, I do believe that property and casualty insurance is, if not the most important industry, one of the most important industries to our economy across the globe. And I didn't get involved in it for any other reason than, as you mentioned, I was in the services side of the business. I was a partner with Accenture, back then Andersen Consulting and then Accenture, in financial services. And I started focusing on insurance. And it's interesting. It's got a lot of big, complex problems. It's got a lot of data. So I kind of stuck with it. And then as you described, I've done a few other things over the years. I helped run an outsourcing company called EXL that started off as really a BPO company, doing a lot more with analytics and technology over the years. And they're doing great. That was a lot of fun. I also worked with two carriers as well in senior roles, both Travelers and Hartford. At both companies, I was responsible for technology and operations. And in my last couple of years at Hartford, I also oversaw claims. And I retired from Hartford two and a half years ago, and as you described, have been doing board work and advising companies since.

Juan de Castro: You're advising from large insurers to InsurTechs and also technology companies. So you mentioned you've led technology and operations in Hartford, Travelers, and you were quite focused on mid-market commercial space. Let's start with what were some of the challenges or pain points that you found where you were leading those teams, and how were you thinking about the future of those operations?

Bill Bloom: Right. So both of those companies have excellent middle-market books and most products but also know a lot about what other carriers were doing in the industry. And I guess because you asked what the pain points are, I think it's always good to start with context. So why are there pain points? So we have this, and what I'm saying is not going to be news to anybody, but sometimes it's good to just say it out loud. P&C insurance is about risk sharing. Period. That's why it's such an important industry. But the way this has been created on every single risk, there are typically one or more carriers involved in the process of quoting on their risk, and often in the process of insuring that risk, and one or more brokers involved. So you've got this many to many relationship, especially in the middle market, where there's many carriers, many brokers, and there's really no standards, so to speak. There are some standards, but no real standards. And so most brokers have workflows and data within their shop that are different from other brokers. Most carriers have workflows and systems that are different from other carriers. And you'll talk about new business, you wind up having this process that is different from broker to broker, different from carrier to carrier. And the data that's been sent around is not standard. So every carrier has a different standard for how they submit loss runs, for instance, which is really important to the underwriting process. Every broker has a different way of putting together submissions, which, of course, is really important to the carriers, as they take a look at the risks and try to process them. So it's a long way of providing some context to say, what's the problem? The problem is, and we'll just stick with new business. It takes too long from the time a broker sends out a submission to multiple carriers for those carriers to get back to the broker with a sense of whether they're interested in participating in the process. And then it certainly takes too long for them to quote the risk. It's also too costly because there are too many people involved, within the carrier. You often have issues around quality where eventually all the quality issues get solved, but it takes a long time, even after you write the risk sometimes. So it's a really interesting problem where there are lots of people and lots of companies involved, no standards. And it takes too long. It's too costly. And sometimes the quality suffers.

Juan de Castro: I think you've summarised really well one of the sources of inefficiency in the industry. But you could imagine it's in the best interest of all players, both all brokers and all insurers, to fix it. Because all it creates is, as I said, inefficiencies for the industry, and inefficiencies for the end customer, too. And if you think, what would it take to make progress? And I think the context you just described is very much what in B2C you would define as it does require a network effect. How do you kick off a change where you need a minimal critical mass to start benefiting? Because if just a single broker and a single insurer started changing things, there would not be enough value for the whole industry. So what are you thinking? What does it take to create that type of network effect in the industry?

Bill Bloom: Yes. So there are two ways of approaching it. And certainly, I think the way the industry has approached it is on an incremental basis. And I'm talking mostly on the carrier side. So carriers have done a few things. Number one, they offshore work. I used to work for an outsourcing company. And that helps a little. It certainly helps with the cost. Sometimes it helps with the timeliness. Sometimes it hurts the timeliness. Sometimes it helps with the quality. Sometimes it hurts the quality. But it definitely helps with the cost. So outsourcing is one thing we’ve all tried. Then RPA came along, and RPA is an interesting tool, but there's no practitioner that would look at RPA as anything other than incremental improvement, where you're taking an existing process and an existing system, and you're finding a way to do what a person does today without the person. So again, compresses time for sure, lowers the cost for sure, and quality should stay the same or improve, but it's incremental and then every carrier, every broker tries automation. So why aren't we making bigger improvements? It goes back to when the submission comes in, we all try to fit that submission, carriers, into our existing systems and processes and try to fix things along the way. So unless you fix the submission, then it's going to be hard to make a significant change. At least that's the way I see it.

Juan de Castro: Building on what you just said. So some of the examples you've given about offshoring, in the end how does an individual carrier make the most, in terms of efficiency and cost savings, out of a current, very inefficient process? And what is the way of starting a network effect, where each individual player doesn't have an incentive to change the whole market. You have an incentive to change your own part of the industry. And you've got players like Cytora and others who can drive that benefit for an individual carrier or for individual brokers. But in the end, I think in my view, the network effect will start when all those technologies that are solving individual pain points for individual brokers or carriers start being able to talk to each other. Because then is when you can start solving not just for the challenge of an individual insurer, but encouraging others in the industry to start adopting similar technology. Would that make sense?

Bill Bloom: It absolutely does. I think, and at least the way I used to approach it is, if you don't want to be incremental, what does substantive change look like? And so substantive change, you start to say, all right, instead of it taking a couple of weeks to get back to the broker, why can't we do it in two or three days? Give the underwriter everything he or she needs to make a decision on risk selection and maybe a couple of days to do some pricing. Why can't we do that? Why can't we look at what our operations team does? And we did this. And you can see that a lot of the work that they do is taking legacy data from one legacy system, entering it into another legacy system, interpreting it, entering into another. A lot of work that's really valuable to work in your company, but not valuable for the broker, not valuable for the insured. So I think the goal should be to make the operations team this big, where they're really an assistant to the underwriter, and then try to compress the risk selection piece to a day or two, the pricing to a day or two, maybe the negotiation with the broker to a day or two. And how do you do that? And again, I think the only way you do that is by looking at what comes in the funnel, which is, can you standardise data? Maybe one day, but I don't think so. Then how do you take the non-standard data that comes in, digitize it so that it goes through the carrier's, improved processes to achieve a much smaller cost base because operations get smaller and allow the underwriter to just do underwriting. Risk selection, pricing, negotiation.

Juan de Castro: I really like this because I think what you just said, almost acknowledges that a single carrier is not going to change the whole industry and the way they receive risk from brokers. So you need to, for the time being, still receive the same analogue information about the risk. It's about how do you then digitize that risk as the first step in the process so that you can streamline your internal processes and then deliver a step change in terms of efficiency, broker service, turnaround time, et cetera.

Bill Bloom: Right. But you have to recognize that everybody that's working is doing something valuable. That goes without saying. But are they doing something valuable for the broker? Are they doing something valuable for the insured? Or are they doing something valuable because your systems and processes are inefficient? So you have to want to eliminate those things that are not adding value to the broker or the insured.

Juan de Castro: But in the end, you have to completely redesign your internal processes. But almost, I think we are starting to answer, how do you create a network effect? Because actually, some carriers are already going through the transformation you're describing right now. And driving that complete, digitization of risks upfront, redesigning internal processes, driving a step change in terms of quote turnaround time, efficiency, et cetera. 

Part of what enables a network effect is the fear of missing out. I mean, we saw this with early Facebook. You wanted to be there because your friends were there. I think the fear of missing out in insurance is competitiveness. It's like if your peers are going through this transformation and driving a much faster quote turnaround, much more efficient, to stay competitive, you have to start doing the same too.

Bill Bloom: I think it's the market that dictates it as opposed to FOMO. Because here's the thing, insurance, not just the carriers I work for, there are a lot of really good carriers with a lot of good actuaries and underwriters that work there. And there are a lot of really talented agents and brokers. And so you've got this very efficient market. And you'll hear that. So you might be able to find a niche for a small period of time where you can be more profitable than the rest of the industry. But eventually, the rest of the industry figures it out. So I think it's the efficiency of the market that's going to help with the network effect. So if somebody is working at a carrier and they're able to reduce costs by 70% on middle market risks, and they're able to compress the timeframe from two weeks to two days to get back to a producer, and they start writing more business and it's more profitable because they've got more room in their pricing because they've reduced costs. So that'll last for a little while. But the market is efficient and others will figure out how to do it just to stay competitive.

Juan de Castro: Exactly. But it does put pressure on the whole market to drive the same efficiencies. So we've been talking about inefficiencies, thinking about at a very high level, what's the process to address some of this? How will the economics of an insurer or what is the business impact of this type of transformation? You've touched on reducing turnaround time, but what do you see as the business impact of this transformation?

Bill Bloom: So I think that it will be twofold. One is what we've been talking about. The second is actually a little more interesting, at least to me. So you become more efficient, you do better quality work, the industry is moving quicker, that happens. But digitizing the data upfront, and we are an industry that has a lot of data, a lot of historical data submissions that we quoted and didn't write, submissions we quoted and did write, loss runs, all this kind of stuff. Once you've digitized the submission or the risk upfront, you have much, much better data to run your models. And by now, there are a lot of carriers, if not, well, most have machine learning models and AI models. And the more data you have, the better you can train it. And so you get the efficiency, which is what we talked about. But now you are able to build models where you do a better job of marketing, talking with the right producers about the risks that you're most interested in. You do a much better job of risk selection and you do a much better job of pricing. And I think that's where now you go back, carriers that get in early on the digitization are doing a much better job with marketing, risk selection, and pricing because they'll have better models.

Juan de Castro: The efficiency obviously results in a better expense ratio. We're talking about better risk selection, marketing, pricing. It's about the loss ratio. And I think we had seen historically, that there was a trade-off between loss ratio and expense ratio. This was like a common belief in the industry, which is that if you spend too much analysing a risk, then your expense ratio would suffer. If you want to do it quicker, then your loss ratio would suffer. I think what you're saying is this is the way of solving for both.

Bill Bloom: Look, the expense ratio is important. But what we've been talking about so far, the reason you do it is not to reduce your expense ratio, although that works, it's to improve your growth. Everybody's looking for growth. On the loss ratio side, and I know you guys are just starting to work with customers and claims, the loss ratio, is much more important than the expense ratio. Because if you're able to impact the loss ratio by a little, the effect on your profitability is much, much more significant. So you really, number one, you have to be careful about expense reductions and claims, but you get excited about how you can impact losses in claims because that's a much, much bigger nut.

Juan de Castro: Yeah. And to some extent, you said an improvement in expense ratio does accelerate growth. An improvement in loss ratio also opens a broader range of risks you can underwrite because you can get into slightly higher loss ratio areas where you can still be profitable if your expense ratio is optimal.

Bill Bloom: Well, and if you're a public company, now you're going for the big picture, your valuation changes if you're able to show that your combines are not just improving, but are better than your biggest competitor, your most important competitors on a consistent basis. Valuation will see an uptick.

Juan de Castro: Yeah. So we've talked quite a lot about the underwriting workflows. You mentioned that one of the areas you led in some of your previous roles was claims, and you've also touched on claims. So are these challenges obviously different, but similar in nature?

Bill Bloom: I don't know if it's similar. I think I'd be interested to know what you think because you're actually still working in it. I get to advise, which is a lot more fun than being accountable. But claims is more complex than underwriting. And I apologise if anybody from underwriting is watching this, but with claims, you've got typically multiple claimants, typically multiple carriers, and you've got more third parties involved, whether they're property restoration firms or auto body shop firms or doctors or independent adjusters or lawyers. Then the thing that really makes claims complex is we have lots of long-tail claims, like in workers comp, you can have claims that are on your books for decades and different doctors and different providers getting involved. So you look at the inefficiency of that, and it just makes for a very complex problem to solve. But what's similar is we've all been trying to solve it incrementally by taking little pieces of it. I think that at some point, solving it up front by digitizing what comes into the claims area is going to help whatever carriers do that, either on their own or with a third party, be much, much more successful on the claim side.

Juan de Castro: Yeah, they're definitely similar. And I think, to your point about it's conceptually similar, but very different, right? I think one of the biggest differences in claims is in typically new business, you have like a single start of the process, which is when the broker is sending the documentation about the submission, right? The documentation about the risk, when to write the risk. And at least most of it comes in a single communication. Whereas in claims, often when people think about digitizing claims, they think about the first notification of loss, which is often almost the simplest of the points. Because the complexity starts, to your point, with multiple claimants, multiple carriers, and multiple parties. The complexities, there are often 5, 10, 15 other communications that an insurer receives throughout the lifetime of that claim. So the notice of hearings, police reports, medical records. It is about how you digitize the intake, which is not just what the broker or your insurer has defined, but any other third party.

Bill Bloom: You're exactly right. And just to make that a little more real, the first notice of loss, actually the carrier controls that, either because they're the ones asking the questions on a phone call, or they've created a digital or an app or whatever. So you control that. That's not easy, but not as hard. You mentioned you're getting police reports in. Well, gosh, I mean, there have to be a hundred thousand police departments in the US alone. And I'm sure that some of them use the same systems and same reports, but I'm sure lots of them don't. And so how do you digitize some, you get this much data, some, you get this much data and how do you find what you need? And then you multiply that again by all the different participants in that claim. It gets very complex but well worth trying to solve.

Juan de Castro: Definitely. It's even more extreme in claims because often those notifications that you receive throughout the life of the claim, there's a legal requirement that they are submitted to the carrier in actual paper. And then they even need to be scanned up front to be able to start with the digitization. So I think to your point, it is a different level of complexity, less control over who you would receive information from. I mean, just a few minutes ago, we were dreaming about this network effect in the insurance industry where brokers and insurers could exchange data more efficiently. I mean, in claims, I think we're decades away from that because it just multiplied by a thousand, the actors involved.

Bill Bloom: That's right. So in claims, the goal is you have to continue to make incremental improvements. But that network effect that you talked about, that's just a little intimidating to try to solve as an individual anything. So you have to shoot for something in the middle.

Juan de Castro: Definitely. So the same way you talked about efficiency, you talked about better risk selection, better loss ratio, the underwriting side of things. What are the North Star metrics you want to impact on the claims side of things?

Bill Bloom: So very similar to operations, you're looking for timeliness. So how quickly you can start working on the claim, you can adjudicate the claim, you can close the claim, just like on the underwriting side where you're trying to get back to the broker quickly. So maybe there are communication metrics. It's the same thing with the claimant or claimants, or if you're subrogating, like the quicker you can get at everything, the better the result typically is. But you see the big M metrics are growth for underwriting, and loss ratio for claims, and certainly we did at the companies I was at, I'm sure most do, and your NPS scores, right? So are you improving the quality and the service you're providing to your producers and your claimants? And that's NPS or transactional NPS. And I think that's a really important metric to try to track as well.

Juan de Castro: Especially because it's the NPS at the moment of truth. That is where you fulfil the promise you made when you underwrote that business in the first instance.

Bill Bloom: Right. Which actually gets back to what I said up front, which is why this industry is so important. Because you know what? Stuff happens. And we made a commitment. It's a contractual commitment, we committed to be there when stuff happens, whether it's to an individual or a company or a ship or whatever. Our commitment is to try to make them as whole as possible, as quickly as possible. And that's why we're in business.

Juan de Castro: Definitely. One last question. I don't want to wrap up the episode without asking the question that we have to talk about Gen AI. Let's start with Gen AI. Obviously, many people refer to 2023 being the year of the testing, the POCs, et cetera, and 2024, as the year of production. I'm specifically interested because as you work with insurers, InsurTechs, et cetera, how do you see Gen AI being deployed, the interest from the boards, et cetera? Give me your overview.

Bill Bloom: So, I mean, starting backwards, the interest from boards and advisors could not be any bigger. And I think that boards, the boards of directors or advisory boards have basically got it, what the technology can do, not just text, but text, video, pictures, whatnot. The use cases are interesting, but it's back to making an incremental improvement to what already exists. So I don't know. I ran operational things, but I am looking forward to seeing where smart people like you take the industry.

Juan de Castro: Actually, interestingly enough, in contexts like the one you described around claims, we apply in production Gen AI in both underwriting and claims processes. It is in those contexts where you don't know what you are going to receive from a broker or from a police station where the Gen AI use cases are strongest. Because I think the industry cracked how to digitize an accord template years ago, right? It's now more about digitizing an input where it can come in a paper document from the police station in the middle of Montana, right? How do you digitize that? And I think those are use cases where only Gen AI can deliver value in those contexts.

Bill Bloom: And I agree. And honestly, I know this is not a commercial, but that's why I'm betting on you guys because you're much more thoughtful, creative and forward-looking than I think I was over the course of my career. And I'm hopeful that you're able to help solve this because it's important.

Juan de Castro: I appreciate that. But as you said, these podcasts are not about Cytora, but I appreciate, obviously, your support. Bill, I've thoroughly enjoyed this chat. It's always a pleasure catching up. It does feel very natural, like a fireside type of chat. So thank you so much for joining me.

Bill Bloom: And Juan, thanks for inviting me. I'm glad we were able to schedule it, but I have to go play golf right now. See you soon.

Juan de Castro: See you soon. Thank you, Bill.