JBKnowledge podcast network.
The InsureTech 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 into the technology we see changing the industry. We are taking you on a journey through insurance tech, so enjoy the ride, and geek out!
Rob: Hey everyone. This is Rob Galbraith. I am flying solo this week of the InsureTech Geek podcast. Unfortunately, no James riding along. So, this is the time the copilot gets to take over and earn his wings a little bit. So I’m real excited to be with you guys this week, and glad to be joined by Colby Tunick of ReFocus AI. Colby, it’s a pleasure having you on.
Colby: Thanks for having me on the call, Rob. Great to meet you and great to be here.
Rob: So we’re real excited to get to meet Colby, learn a little bit about him and, learn about ReFocus AI and what they’re focused on. And we’re going to get a little geeky, talking about the difference between AI and machine learning and go a little techy and then a little salesy this week, so I’m really looking forward to it. Before we get started, with our conversation, don’t forget that you can subscribe to the InsureTech Geek podcast by texting GeekOut to 66866 and make sure you never miss an episode. You’ll get our newsletter and you’ll be in the know to get this episode and all future episodes. So I encourage you guys to text GeekOut to 66866. Back with Colby. So Colby, we always like to start every podcast by getting to know a little bit about our guest and having them give a bit of a background, you’ll be shocked to know that many of our guests did not originally start out to be in the insurance industry, and so it’s always interesting to hear, the winding path that you took. I don’t think you’re any exception to that rule. So, just tell us a little bit about yourself.
Colby: Absolutely, Rob. I hear that a lot when I talk to people in the insurance industry, one of the questions I always enjoy asking is how did you get involved in insurance? And most of the time it was, well, it just happened. It was so unexpected. It was, a lateral move. It was an opportunity that a friend presented, something like that. But, something that I’ve enjoyed about insurance is that people are very value and mission-driven and insurance is really one of the ways that we can help people with more sustainable lives. My journey didn’t start with insurance. When I graduated from college, my first job out of college was working as a defense consultant grant writer and international researcher, and I was running some of my own projects. One of them that I always like to mention as I was helping run a coffee farm in Hawaii, not at all related to the defense part of the job, but it was just a task that came up. We used to joke that friends don’t let friends drink Starbucks. I encourage everybody who’s listening to go find a local coffee roaster and give them your business. I did that job for a couple of years, it was a great learning experience about how international companies go about seeking support from governments, in particular, I was writing grants for both the U S department of defense and UK ministry of defense, in that role, and just learned a lot about how large companies operate, and that transitioned me into a role working at a space technology startup, which was extremely interesting. I got to put on my Elon Musk cap for a little bit, and learned a lot about what you could consider cutting edge technology in an industry that’s actually very antiquated. When we think about space, we think about new frontier. It’s exciting, everything’s shiny, but most of the techniques and technologies they use are 50, 60, 70 years old. On average, to build a satellite, it takes five years. That’s something a lot of people don’t realize and interestingly, there’s a lot of parallels between insurance and satellite design or space tech design because of the tool set and methodologies that are still in use there today. So after leaving that job, I took a position at the California earthquake authority, which is the second largest single line, natural catastrophe insurance company in the world. It’s an insurance carrier specifically focusing on residential earthquake insurance in the state of California. It’s comparable to the federally run insurance companies that book Japan and New Zealand have for earthquake insurance. And again, of course, those companies are on the Pacific rim and sustain regular earthquakes, just like we do in the western United States, and working there at that insurance carrier. I realized that there was an opportunity to use our data, to help drive not only our business, but our sales decisions. And it’s interesting because insurance is one of the few industries that can claim that they have too much data. Most other industries suffer from what you would consider a data shortage, they’re relatively new. There’s not a depth of history there to pull from, but insurance produces so much data. In many cases, it can actually be crippling, so at that point I was actually working on a different technology for social media marketing and he wasn’t my co-founder at the time, but my co-founder now our chief operations officer, Elisha Cheng, started talking to marketing people about my idea without telling me, and he went out on my behalf and did this research for me, which was very much appreciated when I found out he was doing it. But, early results from that were not promising. I was a little downtrodden. I had been spending a couple of months understanding this market and the industry, so I called an advisor and I said, Hey, I’m not seeing a lot of luck here. It looks like there’s just not the need I thought. And he said, “you know, you work in insurance. I don’t actually know anything about insurance, but is there an, is there an application of machine learning for insurance?” And that got me thinking again on the data question that we had internally been discussing and trying to resolve, and that’s where we are today, a little after a year. It’s been a long year and a quarter, because we started the company and then we have COVID-19, and that has been quite a journey, but here we are today and I couldn’t be happier with the progress we’ve made.
Rob: That’s an amazing story, Colby, and I know you’re coming to us from the San Diego area. So, in Hawaii, were you actually in Hawaii or were you managing it from California?
Colby: So I was primarily operating out of my offices in San Diego, but I was able to actually go to the coffee farm in Hawaii and understand the coffee process intimately.
Rob: Was this on the big Island?
Colby: It was, this was in Kona, near Captain Cook, if you’re familiar, highly recommend if you’re ever in Hawaii going and sampling some of the local coffee, it’s absolutely outstanding.
Rob: Yeah, you’re talking about two of my passions already, coffee and Hawaii. So I’ve been privileged enough to actually go to Hawaii three times, twice on business, because I got to go to two conferences in Hawaii. One was on Oahu in Honolulu. So I got to see Pearl Harbor do a little bit of tour, but then I did take a flight out to the big Island for a day and just fell in love out there. Flew into Hilo, but they kind of made the circle round and got over to the Kona coast and walked through volcanoes national park a little bit, and then actually, had the opportunity to go back and, I speak at a conference on wildfire in Hawaii. And, when I flew out there, actually, not soon after landing had a 6.9 magnitude earthquake, which was a fun to go through the middle of, and everyone was like, Oh my gosh, it’s. So thankfully no major damage, but it did precipitate some lava flow that ended up going on for several months. I got to go back to volcanoes national park and like a day or two later it shut down for months and definitely some homes and cars melted. Very interesting time. And then I took the family last year for vacation. And so we stayed out on Kauai for the week. So that was a really cold, but you’re right, the coffee’s amazing, some beautiful coffee farms out there. You mentioned space tech and then you kind of mentioned moving over to the California earthquake insurance. Were you doing something similar at the space tech company when you moved to CEA, what was the connection between those two jobs?
Colby: So what I was doing for the space tech company was business development. My background has been sales. At the Calendar and Earthquake Authority I was doing professional development, which was a little bit different, but it was a unique opportunity. And I know when you’re thinking professional development, what does that have to do with data science? What does that have to do with machine learning? But when you’re in professional development, you need to learn every single position at the company in order to support every single person’s professional development, and often what the job entails is identifying gaps that you can tell with training. A lot of the gaps had to do with data and machine learning and business intelligence and things of that nature, which is actually how I was able to understand the problem set a little bit more intimately than somebody who was perhaps in a different business function that wasn’t speaking to everybody at the organization and didn’t have visibility into the scope and even the scale of the problem. Accenture, I believe in 2016, did a survey of insurance companies. And of course Accenture is that global titan, the international consulting company. And they found that insurance companies only use 10% to 15% of all the data they have available. From everything. And that really just speaks to the prevalence of data and even too much data within the insurance industry. So tapping into that data is key to creating efficiencies, lowering customer acquisition crossed, boosting retention, and just overall creating a better customer experience for our insurance.
Rob: Yeah. I’m glad that you seized on that, and I do find that folks such as yourself, that are innovators and entrepreneurs in the industry tend to come from a non traditional background or a non-traditional career path, you know, not just starting at the front lines and working your way up through a claims department or underwriting department, and so kind of coming in from the side parallels my career path, spending a whole lot more time in the insurance industry, but I know I kind of started in as a sideways position. I’ve always felt like I wasn’t up the ladder in one particular vertical, and so I think that’s always given me a unique perspective on the industry. So I love the comments that you mentioned about professional development, getting to know everybody, and then that being your first exposure to the insurance industry, and you’re absolutely right about data and is kind of interesting, right? Data is the lifeblood of course, of the insurance industry, but so many other industries are data poor and we’re data rich, but we don’t know what to make sense of that at all. You told us a little bit about the genesis of ReFocus, so let’s talk about ReFocus, what was this idea? When did you decide to go out and venture on your own and what’s your value prop?
Colby: ReFocus has been a great collaboration with me and my two co-founders Elisha Cheng, our chief operations officer and Alexander Pearson–Goulart who who’s our chief technology officer, and both also come from very unique backgrounds. Elisha out of college, started an import export business for telecommunications out of Southern China, and then sold that business, came back to the US and then was a campaign manager for a social media influencer and grew his cross-platform following from 10,000 to 100,000 in less than three months, and then we got him, and his specialization has always been really finding and honing in on a specific target demographic and figuring out who the market really is and what they want. I met Alexander at San Diego State University during our undergrad, we actually ended up working at the space technology startup together, and just have remained close friends ever since. Currently he works as a software architect for a biotechnology company in the oil and gas industry and has really specialized in the complex solutions to the cloud in a sustainable and scalable way. And it’s also interesting because a lot of what we think about insurance technology is what we would consider on premises or on prem, meaning you have a server running in a closet somewhere. And one of the companies we’re speaking with, actually just the other day had their entire system go down because their on-prem servers got flooded by some of the weather events happening in the Southern United States. So the fact that we have someone of Alexander’s caliber to help us make sure that our solution is 100% available and is always functioning at its best, and it’s optimized for what we’re trying to serve for our customers is great. We started September of 2019, right about there. It was kind of a soft go and it began just by speaking to 50 people in insurance, and we did that in the first couple of months, we reached out to everybody we could get our hands on and just ask them questions. And through that process, we had the ability to speak with people from the vice-president level, all the way up to the C-suite at very large, even some international companies who told us that the first problem was trying to use their data to drive their business decisions. And then we began to go down the value chain and talk to actual insurance practitioners, and as you mentioned these are your frontline salespeople, your agents, your brokers, the people day in and day out trying to serve their customers to the best of their ability, and we realized that there was a gap there in terms of was the technology available to someone in that position, in an insurance versus any other industry? And that’s the gap we felt ReFocus is a machine learning platform or an AI platform for sales professionals in the insurance industry. We’ve built a better mouse trap. We are able to also tell you where to place that mouse trap to increase your revenue, to reduce your customer acquisition costs. And improve your customer experience. We often speak with people that have been selling insurance or in the insurance industry have owned their own brokerage or agency for 20 to 40 years. One of the things we explained to them is there are many ways to sell something. They’re all equally valid and in most situations equally applicable. But if there is a tool that would help you sell more and leave less money on the table? Why wouldn’t you use it? and that’s what a platform like ReFocus provides. It’s a next generation sales enablement product that allows insurance professionals to use the data in their agency management system or customer relationship management software to drive their sales decisions and gain key context into what their customers are likely to do, and when they’re likely to do it, and that’s really exciting because it’s been explained to us by one of our customers. It was like they had a crystal ball and they picked up the phone and called the customer and the customer said, wow, I was just about to call you about that. How did you know? And that was a really validating experience for the customer, because that was the best experience that customer could have ever asked for. It was validating for the agency because they were able to pretty much grow their book without putting in that much more work, and it was validating for us because we realized that we were creating meaningful change and helping people save time and live their lives in a more meaningful way.
Rob: Yeah, I love that focus, and I think it’s really, really fascinating. So I’ve got a couple of follow-up questions. The first is we know that there’s mom and pop agencies that are out there. Sometimes I say it’s like the last family business in America. and I’ve met some that are three, four, five, even six generations old, which is a pretty fascinating. And then, there’s some really big ones out there, and larger agencies, there’s a lot of acquisitions in this space, right? As people are looking to retire and sell their agency and there’s consolidation in this space, and then there’s Marsh and Aon and Willis towers, Watson, some of the really, really big players as well. So who is your solution targeted at? And then you mentioned these agencies, they already have agency management systems, other customer relationship management systems. How do you get the data? Are you guys cloud-based and so how are you getting the data in, what is the output look like? Maybe just get a little bit behind the mechanics of, you know, what would the process flow look like from the producer point of view?
Colby: So when we talk about our target customers, primarily we work within the commercial insurance space, but the product is equally applicable to personal lines, life and health. One of the great things that we provide our customers is the ability to cross sell from personal lines to commercial lines or vice versa or cross sell from personal to health or personal delight and back and forth, and identify those opportunities. Our target customers at the moment are what we would consider middle market or kind of the top 100 insurance companies in the U S. And these are companies that are large enough to want a data science team, but they’re not large enough to afford that $10 million annual expenditure that would come along with it. There’s no reason why a mom and pop, a fourth generation business couldn’t use a platform from ours. We provide the ability for them to scale. What we do is we remove the need for every million dollars in premium. You write to have to hire another insurance professional because we’re able to help you identify prospects among your leads that you should be reaching out to, and then we’re also able to help you upsell, cross sell and retain customers that you already have as part of your business. And just as an example, at least on the retail side, we also work with wholesalers and carriers. And I’m happy to talk about that a little bit as well. Rob, if you’re my insured, I can tell you if you’re going to purchase policy type X this week. And that may be different from what you’d be likely to purchase next week or the week after. And that’s also different from, are you likely to have a claim this day, this week, this month, this year, are you likely to be sold more coverage? Has there hasn’t been a life event that has changed your risk and now you need more coverage to cover that. And at least on the retail side, the people that we work with on the smaller end understand the value that’s in their data, and they’re looking for efficiency. A lot of what we see on the smaller side and in smaller insurance businesses are they have a process in place that has worked for them. They’ve grown their book of business to a sustainable level where they’re living off. Premium and referrals at that point. And they’re not really looking to scale larger. And for those people, I applaud them because that’s ultimately where we all want to be a stable income that you just have almost on autopilot in effect. But for everybody else, who’s looking to scale sustainably. Who’s looking to use their data to increase their revenue up to 10 times, our product is something that can help them do that. How it works to get into the technical weeds, and I’ll put a little disclaimer on this we’re a machine learning platform, which means we need data in order to function, we can tell you on the lead list you buy, who you should be reaching out to based on what they’re likely to purchase, how many lines of insurance they’re likely to purchase through you and items like that. But if all you have is a first name, a last name an address and a phone number, there’s really not much we can do there. And I think that’s one of the great limiters is people have to start looking at their data as a resource and not necessarily the systems that it’s stored in is justifiable in cabinet, because there’s so much more you can do when you can harness your data and it really powers your business forward. So we integrate with any agency management system or CRM that has an application program interface or an API. And we are able then to seamlessly send out a back and forth between our platform in the cloud and their platform. If it’s a solution that’s where we would consider on-prem, not in the cloud, we can work with that as well. The second an option, a lot of companies have a centralized database that they use as a policy repository, and we can hook into that extremely easily. And then the last option for companies that perhaps don’t really have a CRM or an agency management system, is we allow people to upload CSPs or other flat file transfers to the system. So those are the three ways we get information in and out. And the ultimate idea of this is to provide context for every customer interaction. When someone calls in with a policy question, we know that that is an opportunity to provide an excellent customer experience. And if you see there’s an opportunity for them to buy policy X, why wouldn’t you offer them that policy? What we’re trying to get away from are the days where you run the report in your agency management system and it says they just don’t have these 10 policies. And that’s really not useful because no one likes to say, Hey, you don’t have policy ABC and D which one do you want? And the answer is they probably want B, but because you overwhelm them with options, they’re not going to purchase any. But what if you could just approach them with option B. And that’s, what’s really exciting and a step change for how we’re allowing customers to use their data when they make sales and just improve that customer experience.
Rob: That’s really fascinating Colby, my minds buzzing with a lot of different possibilities based on what you described. So I want to geek out a little bit. Our listeners know that we like to go here a little bit on the InsureTech Geek Podcast, it’s in the name. One of the things you guys talk about is democratizing AI, for agents and brokers, and then you mentioned machine learning. And so, maybe you could explain for the audience what is the difference between AI and machine learning and get into the nuances a little bit, you know, how does it work? And then when you talk about democratizing AI, what do you mean by that?
Colby: Machine learning is a part of artificial intelligence. We think of artificial intelligence as the ultimate umbrella that encompasses many areas of computerized statistics or computerized inference. So artificial intelligence at the ultimate level is the idea that a computer can see a situation that it’s never experienced before, apply logic from a separate but similar situation and react accordingly. And when we often think about AI, we think about self-driving cars, we think about automation, but all of those things are not actually. Artificial intelligence, because again, there’s a requirement for artificial intelligence that computer can automatically relate between two similar, but different events and apply logic from one to the other. If you were to go to a zoo and see a Lion in a zoo, and then you were to go and see a Bobcat, you would be able to say as a human, I learned in the zoo that this sort of big cat is dangerous, I should back up and not approach it. A computer would walk up to that Bobcat and say, Hey, let me pet you, and then it would die because it’s not able to apply that sort of complex thought that we can or dolphins can, or elephants can, or chimpanzees can to unique situations. So ultimately where we’re at now is machine learning and machine learning is looking at data at scale. So we’re no longer just looking at two points of data. We’re looking at hundreds of points of data across thousands of different instances in time. And this allows us to understand within that limited scope, what is going to happen next? A couple of great examples of this. If any of you use Gmail, Gmail now has this great text recommender system where when you type something in, it tries to predict out what the rest of the sentence is going to say. And it does that with a machine learning model called GPT3. And this is a machine learning model that was trained on, I believe it was a trillion lines of data. And the thing is, it’s great for predicting what you’re going to say next most of the time, but a machine learning model is just an algorithm that a data scientist has written, so you can’t apply it to space travel and say, when are we going to see aliens next GPT3, because that’s not how it works. And so machine learning is the ability for a computer to take in a specific set of data. And there is some ability to generalize, but that’s a bit of a deeper topic, but it takes in a specific set of data, and then it’s able to infer, right? Again, these are all predictions. There’s a high likelihood that predictions will happen. Our platform has between a 90% and a 97% accuracy rate. So if we say, Rob is going to purchase policy A this week, and we think he has an 85% chance of doing that, what we’re saying is he has an 85% chance of doing that plus or minus 3%, if it was on 97% accuracy rate for the model, which is pretty good. I know I would want to call you as a prospect, if that was the information I have. And that’s really where machine learning shines is looking across a specific data set. and then being able to take a new piece of data and say, based on existing data, what’s likely to happen next. This is how robo traders work on the stock market. They look at 150 years of stock data, and then they look at all the data that’s coming in and they’re saying, how do we think this is going to play out the next one, two, three, six months. And that’s really where machine learning is limited because it needs a lot of data about a specific topic. Interestingly, this is something a lot of people don’t realize, the way you train a machine learning model is the same way you train the child trying to catch a ball. When you think about trying to teach a child to catch a ball, it takes you a couple of times throwing the ball at them, for them to realize that what they should be doing is catching it and throwing it back. And the same as how you train a machine learning model, you train it, and then you give it an instance and you have a predict on it. If it predicts correctly, it reinforces that previous prediction, otherwise it learns from its mistake and then eventually using the same ball metaphor, it learns to catch the ball and throw it back. And that’s how we get these very highly trained and accurate models for specific situations in machine learning. Looking at that from an insurance practitioner’s perspective, one of the questions we get is I only trust 80% of my data. What do I do? One of the great things about machine learning is we’re no longer looking at two data points for five people, we’re looking at 200 data points, anything that’s on an accord form, anything that’s on a carrier form anything that’s on a supplemental form for thousands of different policies that these people have purchased. So if there are an accuracies in your data, of course, that’s going to affect the outcome, but you can still achieve a fairly high accuracy rate that’s perfectly viable in a sales situation using the data you already have without the need to go out and purchase third party data or enrich your data. And that’s also a mind shift change we’re trying to get people to think about is the data you have in 95% of the situations is already good enough for you to understand what this prospect is going to purchase or how likely your current customer is to leave.
Rob: I think you make a great point. In fact, I was at a conference earlier this week, talking about third-party data, frictionless underwriting, some of these concepts, right? When you want to get straight through processing , instant quote type capability, and the homeowner doesn’t know the square foot of their home. They certainly don’t know to back out the garage. But to include the basement, if it’s finished off or things like that. So you go get third-party data and train the model. So that’s really fascinating that you can leverage so much of the data that they already have, even if it’s imperfect data. And I loved your analogy on the child catching the ball. I guess what I’ve read is that it might take a child a couple of times, but it would take the machine like 6,000 times. You’ve got to train it way more. So in some ways like machines are not the pattern recognizers that humans are, you have to use brute force, right. Computing, power and whatnot. If you do give it enough data, then it can come with these patterns and can definitely see things that humans can’t. So in some ways, your Bobcat analogy is a great one where humans can tell, Hey, there’s a distinction and even a small child can quickly pick up on some of these distinctions, and yet, there are so many powerful applications like self-driving cars. So it’s always fascinating when we talk about artificial intelligence and most of the time we’re talking about it in a very narrow sense, to your point, right? In terms of practical application versus, artificial general intelligence, which is kind of that how. Or that cyborg or Terminator. And there’s a lot of fear out there, and the sense I get in books I read and experts I’ve talked to is that that is still a really long way away. Kind of curious what your thoughts are.
Colby: So artificial intelligence is the ability for a computer to recognize its own existence and think of itself as an individual. I don’t have a crystal ball really. I can’t tell you if that’ll happen or when it will happen at the rate that technology advances, it will probably happen at some point where we have a sentience computer. There was a couple of interesting computer science experiments in the early two thousands, where they had two computers communicating with each other. And after only a couple of hours, they created their own language. Humans could no longer understand what they were communicating and they had to turn the experiment off. So the idea that if there is a sentient computer, we can even understand what it’s communicating, that in itself is also a big question. There is no requirement that a computer that is sentient wouldn’t create a more efficient language of communication. That we would, as humans be able to communicate with something, that has access to so much more computing, power and resources than we in a human in a single lifetime could ever hope to accomplish. And when we think about AI, there’s a lot of exciting potentials for the ability to automate risky jobs, dangerous jobs, manual jobs, the ability to make transportation safer. Self-driving cars. If everybody has a self-driving car and all the cars can communicate with each other, you’re going to get places much faster with no accidents, planes aren’t going to be falling out of the sky. Internet will be faster. Your doctor will know before you know that you’re sick, things of that nature, but we are a long ways off from AI and we don’t even really know what it looks like. And again, because there’s also no requirement that if a computer is sentient, it will communicate with us in a way we understand, and it’s an exciting thing to think about, because it also kind of expands upon the idea are we alone in this universe? Are there aliens out there? Wherever you fall on that spectrum, if computers are sentient, doesn’t that also kind of fulfill the life outside of homosapien and I think that’s an interesting idea concept to think about as well.
Rob: Great thought experiment. It would be fascinating to see what the future holds. So I’m going to reel it back in a little bit, and, kind of get back to this use case that ReFocus is really laser focused on right now. So, in some of your materials you’re talking about improving sales, you’re talking about customer retention and, you talked about this concept of recapturing half a day. So your brokers can focus on what they do best. Can you expand on that? What do you mean when you say, recapture half a day?
Colby: Machine learning is the ability to take statistics and automated and then automatically infer what’s going to happen next. When we talk about what this actually looks like, we’re looking again at all of the data in the AMS or the CRM across all the different customers that have ever existed, both future present and historical and all of the different policies that they have written. And this is a wealth of data. And then what we’re doing is we’re able to mine that for insights and allow the insurance professional to reach out to people that are likely to purchase a specific policy in a specific timeframe, and retain customers better. A lot of insurance professionals reach out to their insured once a year on renewals, there’s the three month outreach, the one month outreach and then, Hey, the policies due. In this market, a lot of carriers are choosing not to necessarily just rewrite the policy automatically. The agent has to do a lot of work to try to get another carrier to take on the policy and, prices across the board seem to be going up, especially in commercial insurance, we’ve had a bunch of natural catastrophes and carriers have had losses the last couple of years, and taking that approach to it if you knew six months before renewal that your customer had a concern or was likely to leave, wouldn’t it be better to reach out to them then at that moment, rather than wait for renewals, when you also have to give them the bad news that their rate’s going up. So that’s what we kind of talk about as a practitioner, improving sales and improving customer retention. The best way to illustrate this is we’re capturing half a day use cases with one of our wholesale partners. Our wholesale partners are fielding more policies than perhaps ever before because specialty complex risk is not being written as easily by your frontline carriers. So brokers are sending it to wholesalers and then the wholesalers are trying to place it with carriers that deal on complex risk, but this process is entirely manual. So what does this look like? One of our customers on a daily basis receives 400 to 500 inbound policies. Each one of those policies has to be reviewed by a human underwriter before they’re willing to try it and place it with the carrier for it. That’s just internal process. And then once they’ve tried to place it with the carrier, they have hundreds of carriers they work with, and that means they’re sending out 20, 30, 40 different accord forms or carrier specific forms. And all of that just takes time. What does ReFocus do in that situation? Our product allows a wholesaler key piece of information. Right now, they’re only able to process 10 to 15 policies a day with human underwriters. So it takes two to three to four weeks sometimes for them to respond to a broker agent. I’m hearing from the people I’m speaking with that anytime they send something to all sellers right now, it’s two weeks to a month before they hear anything back. And that’s just because humans are in the loop and what we can do is help. Our wholesale partners automatically qualify their inbound policies, and what we’re able to do is say anything over 60% certainty, you automatically try to place with the carrier. Everything under that amount gets reviewed by a human underwriter. The agents and brokers are getting out much faster response because they underwriters are looking at less. And then the next step of that is trying to automatically match the policy to the carrier based on market conditions automatically. So rather than them having to send out 20, 30, 40 different forms, we’re able to say, based on market conditions, these two carriers are the people you should only bother sending this policy to. They’re the two that are going to write it. Ignore the other 38, because they’re not going to even look at this, but then tomorrow when you get a similar policy, we can say, send it to these two different carriers because the market is constantly evolving and we can help wholesale professionals trying to improve their efficiency, placing policies as well. All of this comes back to building a scalable insurance business where you’re serving to make sales, where you’re serving your customers to answer their needs to service the policies, of course, but now you’re spending your time making sales to the people that are actually likely to purchase something, reaching out to people that are at risk of turning and automatically qualifying prospects so that you know who you should bother to work with. When you buy a list of a thousand people, it doesn’t do you any good if you have to pick up the phone and call up a thousand people to get five companies. But if you can just find out the three companies that are most likely to buy, suddenly you’ve spent four fewer hours or eight fewer hours of your day, and now you can be more effective in what you’re doing. So that’s why we say recapture half a day, it’s spending your day differently so that you’re more effective. And you as a human are more scalable.
Rob: I think that’s so compelling Colby because there are a lot of inefficiencies in distribution. I think there was a wave of InsureTech at the very beginning, let’s say mid 2010s that said I was just going to go direct. We’re going to cut out these agents and brokers. And those that have realized that, Hey, unless you’re willing to spend a billion dollars in advertising like a Geico, or progressive, actually those agents and brokers are your best distribution channel, plus they know the customer best. The market is that way for a good reason, particularly when you’re talking about mid-market and larger risks and complex risk, et cetera, going beyond simple personal lines or small commercial, which is easier to go direct or sell over digital channels. I think there’s been a lot more focus in this space that I’ve seen even direct carriers like startups, then changed their tune later on and say, Oh, we love independent agents. You’re right. It’s, Hey, I’ve given you a submission, but I’m waiting to hear back and it’s like, I don’t need you to go through the full underwriting. I just want to know like, Hey, are you planning to compete for this or not? There’s this kind of delicate dance, right? They kind of go in between agents and brokers. Wholesalers if, like you said, it’s a complex risk. So there’s lots of email going back and forth. There’s frankly, a lot of inefficiencies because of the humans, and being able to enable a solution that’s smoothing down that process and making it more refined and more effective is really a win-win. It’s a win for the the producer side. It’s a win for the carrier and certainly, it’s a win for the customer, which is great. So as we wrap up, you’ve talked a little bit about that types of systems that you integrate with, and I just want to get your thoughts on this, you mentioned on-prem and we talked about in the space technology firm that you were at and how there can be some legacy systems, and out of date technologies. And there’s this whole idea of previously siloed systems that are now becoming platforms and ecosystems, you mentioned integration via API. Your thoughts on this topic of insurance platforms and ecosystems. And then how does ReFocus see itself in that context?
Colby: If the banking industry can ultimately become connected, I know insurance can as well. There’s lots of great technology for banking. Now that was. Just mentally impossible to think about 10 years ago, the ability that you as a consumer can have one app that you can connect to your 10 different banking accounts, and that it automatically pulls in transaction information and balance information. That’s the sort of connectivity that we see that is going to have to happen to insurance, and the reason it’s going to have to happen is because I think insurance agents and brokers are realizing the power in their own data, and are kind of getting tired from the fact that carriers are putting a lot of the processing burden on them. When the carriers created submission portals, a lot of people thought this as an opportunity to reduce the amount of work. Now you’re not necessarily sending an email with all of the accord forms attached and you’re not communicating over an email poorly, you can log in, but then every carrier had a portal and really what they were doing was reducing their level of work and making the agencies and brokers work harder. I think that gave insurance professionals a bad taste in their mouth from thinking about connected systems, because that was their first experience. And that is ultimately unfortunate because what connected systems will allow you to do are have a centralized repository, where you can see everything that’s going on in your agency and not have to worry about having 10 different pieces of software to do 10 different things. And that’s ultimately what the connected system is about because now you have one hub that can connect to everything that’s going on without having to use the platform that this carrier has approved or works with this marketing tool. And that’s where APIs will really help reduce customer acquisition costs and servicing costs and servicing time. Something interesting I found out, I didn’t realize how costly it is for someone to acquire a customer and insurance on the personal lines side. It can be anywhere from $400 to $1,200 to bring in a new policy holder. On the commercial side, it’s upwards of $2,000. And if there is a way to your data to enable and streamline your sales process, you reduce that customer acquisition costs. Now, what we’re talking about is, more money in the agency, less money going out the door in marketing and creating a really hyper scalable business for the next generation of the family member that’s going to own that company.
And at least the ReFocus, we never see independent insurance agents and brokers going away. We don’t see captive agents going away. One of the companies we’re speaking with right now is a large top five captive and one of the issues that they have expressed is when someone retires or chooses to leave at their captive, the carrier needs to decide what to do with the book of business. And this is either an MNA opportunity, or this is an opportunity for the carrier to try to keep and retain those policies and divvy them up to other agents in that area. And our platform can help them with both of those. We can evaluate the value of a book. When you buy a business, any other business but insurance, you don’t just purchase it on that year’s revenue, you purchase it on what it’s expected to generate over the next three years. I personally think it’s ridiculous that there’s a double standard in insurance that when someone purchases a book, they’re only purchasing it for that year’s premium. It’s like one and a half percent of the premium is the purchase price are going to get. And why is that? For every other industry, when you buy a business, you’re purchasing it for future growth and our platform can help people considering to sell their book evaluate it for future growth and ultimately get a better price. Similarly, we can help carriers who are looking at trying to retain those customers, if that agent decides no longer to be a captive or work exclusively with that company, better place the policies with people that are likely, or at the highest potential of retaining that customer. And these are all areas where software can create efficiency and just better the lives of people using it, at the highest level.
Rob: So many use cases that really emerged during our conversation Colby, so I’m really excited about what you guys are doing, at ReFocus AI. It’s been really great to get to know you a little bit. Obviously you’re a relatively new startup, half of the time has been during the pandemic, but it’s great to see your success to date, and I definitely wish you the best of luck going forward. Where can people reach you and find out more if they want to learn more and potentially explore a partnership with ReFocus.
Colby: They can drop me a line on LinkedIn. They can go to our website, refocusai.com and drop us a message there. We’re always happy to pick up the phone and have a call. We like to meet our partners and understand who we’re working with, and I think that’s something that a lot of our customers express as well, because insurance is still very much a face-to-face business, and we see a lot of value in that. So if they are interested again, LinkedIn, they can send us an email, always happy to talk and pick up the phone and see how we can help them.
Rob: It’d be great to sit down over a cup of local coffee or on a Kona coast and learn more, so great to have you on Colby. So, we’re going to pivot now to the news segment of the InsureTech Geek podcast and I’ve flagged three items this week. The first is, I got my save the date email already for InsureTech Connect 2021. It will be held October 4th-6th and they are planning to have it in person. It will not be at the MGM this year in Las Vegas. It will actually be nearby at the Mandalay Bay hotel. They said there’s a lot more room there so it will be a better experience overall. So I’m curious, Colby, have you been to an ITC before? Did you go to this year’s version that was virtual? Just curious, if you did attend, what your thoughts were.
Colby: I had the opportunity to attend ITC this year virtually. I was excited by the quality of the platform and the speakers that they had, I think they did a fantastic job, at least from a technology standpoint, recreating a lot of the reason we all go to conferences, which the number one reason is networking, in a virtual platform, but I’m very excited to attend next year’s ITC in person, and just have an opportunity to meet everyone we’ve been speaking with and shake their hand. That’ll be really rewarding. Vaccine approved, of course.
Rob: If not, I’m asking an elbow bump maybe, but it’s a little weird to think at this moment as we’re going through our third wave here in the United States of COVID cases, setting new records every day, for the nation, it’s hard to think about in-person gatherings, but definitely fingers crossed. Good news this week from Pfizer on a potential vaccine. I know Moderna and others are having a lot of trials as well. So we’ll definitely keep our fingers crossed for that. I think we’re all facing Zoom fatigue. I know I’ve been part of some virtual events this week, and I’ve got two more next week and it feels like a lot of these actually are cramming into the end of the year, because they were in person in the spring and then they’ve kind of had to pivot and repackage. And so, it’s always great to learn and to connect, but definitely the best way to do that is still in person. It’s good that we have virtual as an option, but I definitely miss in-person events.
Another news item is that Brown and Brown, which is a really big agency, acquired several other agencies. And so it’s definitely one of the larger players in the commercial agent or broker space, they’ve actually acquired a digital insurance agency that many of you may have heard of called CoverHound and its CyberPolicy Unit. Congratulations to Keith Moore. We know each other on Twitter mostly, I guess a little bit LinkedIn. There was an article this week in the insurance journal and you can check that out. Yet another InsureTech exit. We’ve been seeing a lot of those in the second half of this year, so good news, we’ll continue to monitor those for you guys as the series goes on a week after week on the InsureTech Geek podcast, and then finally. I have to share this with you. I had the privilege of being on a brand new series, a YouTube series that my friend Ed Halsey from the UK, it’s his brain child, only he could have come up with this at Evermore Digital. It’s called Grinding My Gears and it’s, it’s not quite a game show, but he had three guests on and they’re supposed to talk about what is grinding their gears about insurance. What do they hate about insurance? What would they throw into the depths of Hades, as he would say, and he’s got lots of flame imagery. I happen to be on a group with Nick Lamparelli, Billy Van Jura, and Seth Zaremba, and I got to be the judge of the very first pilot episode. We were the Yankee version, and insurance advertising was the thing that we decided to get rid of, that there’s way too many ads, and we would love it if that money was better spent, returned of all the policy holders, maybe invested in technology, such as yours Colby. And then there’s episode two featuring some of my favorite UK influencers out there, so you can watch both episode one and episode two. We’ll include those YouTube links in the show notes. Tell us who did it better, the Americans or the Brits? Definitely check it out, it is quite fun. We don’t have enough fun in insurance, quite entertaining. So congrats on the launch of your new series, a special thanks to Colby Tunick for joining us today on the InsureTech Geek podcast.
I’m Rob Galbraith, thank you all for listening and we’ll see you next time.