Indico Data receives top position in Everest Group's Intelligent Document Processing (IDP) Insurance PEAK Matrix® 2024
Read More
  Everest Group IDP
             PEAK Matrix® 2022  
Indico Named as Major Contender and Star Performer in Everest Group's PEAK Matrix® for Intelligent Document Processing (IDP)
Access the Report

Unstructured Unlocked episode 15 – Why is insurance intake so challenging?

Listen to Christopher M. Wells, Ph. D., Indico VP of Research and Development, and Michelle Gouveia, VP at Sandbox Insurtech Ventures, in episode 15 of Unstructured Unlocked as they discuss insurance intake, why it’s such a challenge for insurers, and how to make it intelligent.

Listen to the full podcast here: Unstructured Unlocked episode 15 – Why is insurance intake so challenging?


Michelle Gouveia: Welcome to Unstructured Unlocked,

Christopher Wells: A podcast where listeners discover how enterprise leaders are confidently automating document intake and accelerating their workflows to increase capacity and drive top line revenue.

MG: I’m co-host Michelle Govea

CW: And I’m co-hosts Chris Wells. Welcome to the podcast.

MG: Welcome to another episode of Unstructured Unlocked. I’m Michelle Govea, and I’m joined with co-host Chris Wells for today’s topic on intelligent intake. So Chris, this one is, this one’s kind of my fault. I feel like we’ve been <laugh> we’ve been talking about data and the value of data within the insurance industry and how it influences or impacts underwriting workflows and claims workflows. And I think it probably warrants its own deep dive specifically Yeah. Submission intake on the underwriting side cuz it’s such, it’s such a key workflow and key component of, of the insurance industry, right? That’s, and a lot of times that’s how consumers first interface with insurance carriers is, I need insurance. Here’s, here’s all of my information. And so on the flip side thinking about how insurance carriers take that information in, what they use it for the process at which that that happens, I, I think is really interesting and it is an area that we spend a lot of time looking at. And I think that you can, can chime in on some really interesting takes on the intelligent automation and RPA side

CW: If you’re, I’ll try, I’ll try. You know, I know just pretty much everything about intake and insurance, but maybe you could tell our audience Absolutely. What, what intake is and what, and you know, the, I’m sure there are layers of complexity here that you can help us unpack.

MG: Yeah. So I think the, the first thing I would mention is it looks different. It, there is not, there’s not one standard submission process, right? And it varies by product line. So, you know, your experience, I’ll talk about it from the in insurance care side, but even as a consumer, your experience, if you’re personally searching for insurance, you’re looking to buy insurance is gonna differ from a small business trying to buy insurance. Okay. As well as large corporations looking to buy insurance. And, and those processes are all different on the personal line side, right? You’ve seen and, and all the, the big, the big names, you know, advertise this all the time. It’s so easy to get a quote, right, because you just go online. Yeah. Right. Most people are buying their personal, their home and auto insurance online. So that, that’s one form of intake where there’s already a digital front end that the carriers have. And all, all of that information theoretically is feeding directly into a database of some sort where, where those fields are already mapped. You’ve got, and can

CW: I, can I ask one question there? My, my mental model of insurance is you have the end user, like the policy holder, and then you have like a broker between, and then you have the carriers mm-hmm. <Affirmative>. So on the personal line side, are, are brokerages just gone now that we have these web portals?

MG: Not, not gone, but definitely disrupted, I would say on Okay, on the personal line side. Now, now that model, very much so, and, and this is, this is my opinion. You’ll, you’ll have different schools of thought on this depending on, on who you talk to there, there’s some, there’s some that believe that the small commercial insurance industry, so small business insurance is going the way of personal lines where you’re going to have that, that I go online and try to buy my business insurance. And then you have another group myself included, that thinks that for, for small commercial or commercial insurance it’s more complex and the needs of a business will shift as the business grows and develops. And so you will need that, that, that special touch that, that knowledge base from an agent or a broker that says I, I understand your, your insurance needs and there’s a lot out there, and you may be underinsured or you might be you know, missing, missing a few key policies.

So, so that actually gets to my second point. So thank you for teeing that up. You’re welcome. From a submission standpoint insurance carriers are also getting their submissions from different sources. And so if you think of a car insurance carrier has partnerships with four or five or six brokers or, you know, numerous agencies, they’re getting that information from those brokers and agents in different ways. If they’ve got an agent portal that those agencies have to sign into, and then manually input information if it could come in via email, it could come in depending on, on how they’re interacting on a monthly border row where it says, th these are all the policies that, that I’ve bound for y for your company this month. Like here they are not put ’em into your system.

CW: Sorry, I have to ask, what’s a border row?

MG: Think of it as just like one long excel list with all of the information about the policies that are written. So the name, address, policy number you know, all, all, all of those, all of those data points insurance coverage, et cetera. And so that is how do you ingest Excel? Sorry. And, and one could argue, well, in Excel it’s probably structured, right? You’ve got a column, you’ve got all the inputs, but if you’ve got, right, you’re, if you’re okay, if you’re working with 20 brokers, they might have all the same data points, but have it formatted differently or input it differently. And so then on the backend, any, any type of like rules based or template automation Yeah. Breaks down, right? Yeah. Like it, tons of errors. And then you’ve got, you know, I’ll say middle market commercial insurance, which is usually your big corporation.

You’re, you’re trying to get one huge policy with a ton of different coverages in one, right? Okay. and that’s, that’s very email driven things Lockton, P D F, and at email attachments going, going directly to these underwriters in their systems. So so intake looks different for each carrier based on it’s what channel and what line of business those things are coming in through. And there’s even from just that kind of three minute overview probably a lot of pain points that you can isolate <laugh> on what’s happening at the insurance carrier level. Yeah.

CW: Yeah. My, my experience talking to insurers about these challenges have highlight, highlighted a handful of pain points which were, you know, they were in there. I mean, one is everything, especially on the, you know, large commercial lines, like you pointed out, everything’s super bespoke, right? Mm-Hmm. <affirmative>, and so mm-hmm. <Affirmative>, you know, if you wanted to automate this process, right? That means you have to read the email and emails. I’ve told people this email is like the most unstructured of all of the unstructured data, because you can say whatever you want. You don’t have to stick to any kind of template. You can attach anything you want, including more emails to the email. And those things may have attachments. It’s this huge nested crazy data structure. But anyway, you have to read the email. You have to ask, figure out what someone’s asking, you know, do they identify the coverages that they want?

Do they reference documents? Are those documents referenced then attached? And you have to go find those things and you have to pull out, like, it might be in a accord form, right? So you have to pull out all of the infinite detail that’s in, like in Accord 1 25 mm-hmm. <Affirmative> plus policies, plus, you know, say they’ve attached something like a loss runs history. And you were talking about, you were talking about bordero being big tables. We have a saying in indico R and D that there’s no such thing as a table, because, you know, people do, they’ll dump whatever data they want and they’ll put table one on top of it. It’s like, I’m, I’m sorry, the, you have like six levels of nesting here. This, this is not fair. You can’t call this a table. So that, that’s one of the pain points. It’s just the overwhelming amount of data and the need to be able to reference one data point, like the email against the attachment that came in and tie all of that together in a bundle.

MG: And there’s the, there’s the identification, right? Of, of the data that’s sitting in that, that unstructured, even, I’ll even call it semi-structured yeah. Input, right? And then there’s ingesting it in a way that’s you try to automate it in a sense so that it’s not just manual entry, which takes a ton of time. Human error can, can happen. So, and then how, how do you make it, I, I talked about this on the, the first time we chatted about making that information usable, right? So you’ve got the data, yeah. You, you’ve got it usable in the sense that, that you’ve pulled it out, but now how do you use it to make business decisions? How do you take all this information and decide that you are going to underwrite this policy that you do wanna bind this policy? Like how do you, how do you elevate that the data that’s come in hits up against some, some processing exception that requires a deeper look or a different workflow. And just, just how, how do you, how do you make that process as efficient and accurate as possible? And that’s not even now, I’m just on a rant.

CW: No, stay on the soapbox as long as you want.

MG: Yeah. Yeah. That’s not even getting into the, how do you take the, the data that’s coming on the, during this process and validate that it’s even correct. Right? So what other third party exactly information are you pulling in? What system does that happen in? Right? Where are those checks and balances? What, what are the, the, I’ll call it the pings that like alert someone that it needs a second look, and then how do you continue that workflow? Yeah.

CW: And that, that gets to one of the other pain points I was gonna mention, which is, and we use humans to solve all of these right now for the most part, which is like mm-hmm. <Affirmative>, you have to get the data so you have a bot, right? And it finds, here’s the word, and that’s the word you care about, but then you have to rationalize it. Like, you’re probably gonna have to look up the credit worthiness of whoever you might be underwriting, right? And they might, they may not use the corporate entity name that’s in your database that then connects to like Dun and Bradstreet or whatever it is to figure that stuff out. Mm-Hmm. <affirmative>. So it’s not just like, you don’t just need a computer to read the information for you. You have to, you have to actually reason about it and connect the dots to your point. Well,

MG: And that’s a, and you just touched that. This one is not, not specific to, to intake per se, but on that, that data accuracy side, one of the, the biggest pain points on the commercial side is like classifying a business, right? And depending on how that, that business is classified, will, will identify how big of a risk it is, right? So if you’re like I don’t know you submit your N A I C S code that you’re a restaurant yeah. And there’s, you know, there’s, there’s like four digit, five digit, six digit, and they all get very, very detailed. But so you’re a restaurant and you don’t identify that there’s like, I don’t know, liquor things or like a bar, right? But that’s probably a different risk than if you’re just like a sandwich shop restaurant, right? And so if that information is not made available to the underwriter they’re probably pricing that risk inaccurately. Yeah.

CW: Interesting. Yeah. And they made the best decision they could, given the data that they had, right? So that, and that, that’s another pay point, right? Is just how, how do I know when I’ve got everything, right? How do I know that the submission is complete? Yeah. And so I Exactly,

MG: Yeah,

CW: Yeah, high level, the, the pain point is really just takes a lot of human gray matter to do this really well. And so you end up with knock on effects, which I, and again, not an not an expert in this, this is just from talking to folks where you, you can’t quote all the business that’s submitted to you cause you just, you just don’t have time and you don’t know if you’re quoting the right business because it’s just like, you know, there’s this constant cue that’s building up. That’s what I’ve seen.

MG: Yeah. Yeah. And you know, this gets to the, the whole, the whole idea or the, the, the happy place, right? The happy state is you get, you get all this data in, in a way that it’s, it’s complete. So you’re not missing any information. You validated the accuracy of it in some kind of checks and balance that that happens in your system. You hit it up against, so from an underwriting perspective, you’ve got all this information, you hit it up against some type of, of rules-based underwriting. Yeah. you know, box that says, yes, yess is checks these boxes. This is a standard risk automatically under like quote, bind that, right? Like, like low touch, like high frequency, low touch work that is just like super easy to get through is is where you wanna get to. And then if there’s something that hits up and is an exception to those underwriting rules or needs a different look, or maybe there’s a request for higher policy limits or, or something nuanced, then that’s where going back to just the, the human capital, right, is that’s where people can spend their time.

It’s, it’s more more valuable to spend their time on the stuff that, that needs a another look as opposed to just like, I don’t wanna, I don’t wanna use the term pushing paper, but essentially just like, yeah, yeah, yeah. Checking the box paper that like, you, you’ve reviewed it and that it’s, that it’s gone through. So that, that to me is like the end state of, you know, you hear a lot about like automation goals and, and streamlining and making people more effective and more efficient. And I think that’s, that’s, that’s an area where intake can, can really help. Interesting.

CW: Yeah. Yeah. The, the other, the other nice side effect of having that intake maybe if that, if you’re, you just described the dream state, I’ll describe Nirvana, which is you capture all of that data on the decision making, right? And we talked about this last time, and you can cycle it back through, and now you have this long queue of things. Well, these five, these 10 in the row, I’m not gonna quote, I just, we just don’t quote that. It doesn’t turn out well. So skip those all together. And now not only have you streamlined the process individually, you’ve streamlined the whole queue and you’re only quoting and as fast as possible, you’re quoting and binding the stuff that you really want. That all has to happen in the context of an existing tool chain. And I, I don’t know a whole lot about what the types of software, the types of tools that folks are using. So maybe you could tell us a little bit about that.

MG: I can try <laugh>. Okay. Yeah, because, because like all my other answers today, <laugh>, it’s, it’s nuanced and it, and it differs. Yeah. You know, you’ve got, you’ve got the core, the core systems, right? So that do policy, administration, claims handling, billing. So presumably you’ve got a, a number of carriers that have, have upgraded their core systems where some of, some of this is out of the box capability, or it’s been customized where it’s just easy to, to connect in and say like, this field should plug into this field in my, in my system. Right? And then it’s, it’s in there automatically. You know, the, there’s a lot of there, there’s a, a huge push many years ago, and that probably still exists where a lot of carriers have built their own agency or broker portal. Yeah. So, right.

So if I’m an agent sitting at the Michelle Gove agency, like I have to log in to a carrier portal to, to quote business for them, right? Got it. And so in some cases, I’m doing that for multiple carriers or sometimes there’s, there’s kind of a catchall, but like how, so how is that front end? Or how’s that portal integrate to your policy administration system? It’s probably a different connection with the same goal in mind, right? The same information you’re trying to capture. If it’s personal lines, your website that should be linked behind the scenes. Excel is probably still something that’s getting used. Of course, it’s

CW: Course it is. Yeah.

MG: And then, you know, there’s, there’s all these other like, like software solutions and things to help with reporting. And, and to some extent there may even be, you know, one, once something comes in the door, some type of a report that gets run to assess like what’s, what’s in the queue.

CW: Okay.

MG: And then, and then organized by maybe skillset by underwriter type you know, potentially like there, there’s a lot of homegrown like workflow, like work management systems too. I know that in, in my time at, at carriers, there was just a lot of homegrown like work management, like okay. Or like assignment front ends that were built

CW: Interesting stuff that you would do now with like Asana or Jira or some kind of tool like that.

MG: Yeah. Yeah.

CW: So you’ve named like five or six different systems that all of this data has to pass through in, in some fashion that’s, that is complex.

MG: Yeah. And, and you know, you’ve got some carriers that are, are only doing like, you know, only personal lines, right? And so you would, presumably there’s, there’s a pretty, it, it’s, it looks the same internally, but for these multiline carriers that are writing personal lines, commercial lines, you know, life insurance, that this is all, it’s different data into different systems that’s probably got different rules on, on how you store that data, how you can use that data for, for pricing and underwriting. And even how long you can keep that data. So there’s, there’s a lot, there’s a lot there. There’s a lot there. But, but it all starts with getting it in the door in a way that hopefully is not super manual that’s accurate. And that can be, can be pushed through in a, in a, in a, in a usable, usable format.

And I think what a big opportunity is less so in the, in the lines that are more standard, right? More like, like, yeah. Or commoditized, even like, these really niche like specialty insurance lines is where, like your, is is really where an automated solution can be really helpful because you’re collecting a lot more detailed or granular information, right? So if you like interesting. Like if you’re like, like pilot insurance, right? Like if you like pilot, like a personal plane, like that’s a whole different that’s probably a very lengthy application. Yeah. with a lot of, you know critical, critical details and how absolutely. That, but even, even on the personal line side, right? Like, it can get complex probably if you’re, if if you’re submitting a request for a primary home, a secondary home Yeah. You know, maybe you own a, a boat in addition to like your, your vehicles. Like there’s,

CW: There’s some kind of umbrella you have to collect for all of that.

MG: And Yeah. But I’m talking about from the system perspective, right? Like how do you, how do you then say, turn on to say they’ve answered this question this way, so now we need to open up this whole other subset of questions on the submission, and we need to make sure that it gets into the system even if it’s not standard. Yeah. I guess.

CW: All right. Well this makes sense.

MG: I’m quickly getting outside my, my expertise in, in the area, but

CW: Really interesting information. It makes me think that if you’re out there working in the data org at a insurance company, you should send a thank you email to your CDO for worrying about all of these things at night. You said a couple things in there which peaked my curiosity. One of them was that it might be that automation would help most in the specialty cases. And I wanted to double click

MG: On that. My hypothesis. Yeah.

CW: Yeah. That, that’s interesting cuz normally you know, normally folks think of, well, let’s automate like the bread and butter, the simple stuff cuz it’s so high volume mm-hmm. <Affirmative>, but you’re, you’re pointing out something else, which, which these can be so complicated that it’s worth having an automated double check on them. I think, I think that’s what you’re telling me.

MG: Well, that, that, but it’s because it’s so complex and it’s usually very manual what, what incremental benefits can come from automating the intake or doing the check so that that, that the real time is spent analyzing the data that’s in that submission Yeah. As opposed to just getting the data in the door for that submission. If that, if that makes sense. Right. because there, there’s some of those specialty lines just are so complex that they are always gonna need that. They’re not gonna be one, one of those risks that you do just pass through the system and say like, it’s stable stakes. So you may not, the, the automation isn’t gonna happen end to end on those, but to automate a piece of it to then spend the time where, where the true value of, of that underwriting function sits within that line, I think is Yeah. Is an area that, that is, is ripe for, for the benefits of intelligent automation.

CW: Yeah. That, that raises a good point, which is that I think a lot of folks think about automation as I have this end-to-end process. It’s not automated until the whole thing is just one smooth tube that it goes through mm-hmm. <Affirmative> mm-hmm. <Affirmative> where whereas in fact, you’ve pointed out that these insurance systems across carriers lines, the whole ecosystem, they’re really a lot of different systems that talk to each other with sort of human hands moving stuff from one to the next. And you can find benefits in automating intake along the way. It’s not all or nothing like you can carve it up into relevant pieces and then streamline those and then go to the next lily pad and build something, right?

MG: Yeah. And, and that all that depends on what are the, what are the goals that, that the carrier or whoever the entity, right? It could, could be an MGA that’s trying to do this. It could be a broker that’s trying to automate some of their workflows internally to Yeah. It’s, I don’t wanna say that it’s just insurance carriers that, that are, are doing this. Yeah. Okay. You know, it, what, what are they trying, what, what is the goal that they’re trying to meet? You know, what, what is, what are the metrics that that they have today? What are they trying to get to? What’s the r o i that like intelligent automation or intake can, can help them with? And, and that’ll vary, right? And some of it is that pie in the sky. I want, I want this completely automated. I, and some of it is I just, I really need to get the data in the door in a way that doesn’t take, you know, five different double checks and manual entry into the system and then Yeah. You know all of these, these, these approvals and things like that that have to happen.

CW: Yeah. And I, I’ve seen a range. You, you raised the specter of r o I mm-hmm. <Affirmative>, and I’ve seen quite the range of approaches and behaviors out in the wild across industries. But, you know, insurance is very much the same with regard to this. And I would say there’s like a spectrum, and on one end of the spectrum is, well, the automation has to be highly accurate. And you know, if you can’t show me a hundred percent accuracy, then we’re not gonna do it. And I’ve said this many times on the podcast, that’s dumb naive at best, <laugh> in the sense that yeah, your process has to be a hundred percent accurate. That’s like, of course table stakes. You can’t make mistakes, we get it. But accuracy doesn’t have units on it. Like you can’t directly convert it to dollars.

Like you could something like time, right? So more sophisticated organizations are looking at metrics that actually influence the business outcomes, right? So, you know, if you’re in the, you know, if you’re in the small business or the personal lines, like what happened to our quote to bind time when, when we, when we introduced this automation, and of course you’re checking the accuracy, but now it’s the accuracy of humans plus whatever automation and rules, and it’s RPA and it’s ai, it’s all of these things. But sophisticated organizations that are talking that kind of language and they’ve done the homework about like, this is what moves the needle and these are what we think are the gears behind that needle that we need to replace with new gears. Those are the organizations that I want to talk to because they’re actually going to do something.

MG: Well, you can speak directly to a business outcome, right? What’s Yeah, exactly. You were able to, yeah. So which I think I said the first time we chatted too. That’s what exactly, that’s what, when you’re looking to sell to insurance carriers, what is, what is the business outcome that you’re selling, right? The technology is important, like I said and, and hearkening back to, to when we, when we first started chatting about this, like you, you have, you have to have something. The technology’s important, but yeah, like a rules based, you can sell a rules-based template to somebody. Oh yeah. But it’s gonna break the instant that that something falls out outside of that where the intelligent intake side and you’re by far the expert on this, not me. It is, I, I can identify that this information is here, it’s not in the right place, but because I’m smart enough to know that I, I’m not breaking your process or I’m not, you know, putting a name where an address belongs or something like that. So to get to your accuracy standpoint, right? Like that, that’s the inaccuracy that, that eventually Yeah. People not believe in the capabilities of, of automating some of these things.

CW: Yeah. And that, that, that’s a good point I didn’t even mention, which is like, there are different levels of inaccuracy, right? Like if you can, if you can still go through the whole process, then maybe that data element that you pulled out, it might be valuable somewhere else. It might just be a part of the legacy part of the process. You know, you should really think about your process before you automate it.

MG: <Laugh> very good point. <Laugh>. Yeah, good point. Auto automating a broken pro like process is still gonna re result in, you know, broken outcome.

CW: Yeah. They’ll just, it’ll just break faster. <Laugh>. Yeah. <Laugh>. And then more repeatedly <laugh>, you know, last time we talked about this, you know, we were giving the advice to the InsureTech entrepreneur out there, but this is also advice to like the InsureTech salesperson, if they’re not speaking the right language you’re gonna go through some pilot with this prospect, you’re gonna do do some P O c and you’re both gonna waste a lot of time and end up heartbroken. So I, I think of this and look, not a salesperson, I’ve never sold anyone anything as far as I know. But if, if you’re not on the same page, then this is, this is not a pursuit for you. This should be part of your qualification process upfront.

MG: Exactly. And, and that goes back to what I like you, you could sell the technology, but if you’re not actually sure what, what, what the goal is then there’s no there, there’s no incentive to, to say, well, I’m gonna take a shot on this tech versus this tech. If, if it, if you don’t really understand the, the desired outcome. Yeah.

And, and but most importantly is, as the buyer, you have to know the desired outcome, right? Is it to automate a piece of it? Is it to automate the whole thing? What, what is that range of, to, to your point of accuracy, right? Or, or it might even not have to, like, I, I’m gonna counter what you said, like a hundred percent is important. A hundred percent accuracy should be what you aim for. But if current accuracy is 50%, oh yeah. You can automate to 75, that’s an incremental, you know, progression to towards that 100. As the system gets smarter, then hopefully you get to that a hundred. But can already, can already improve it.

CW: No, absolutely. Now, now you’re, now you’re implying that people actually know what the accuracy of the processes is. And I, I I don’t see that a lot, honestly. Just a shame.

MG: <Laugh>. Interesting,

CW: Interesting. Yeah. I just aired some dirty laundry, but I didn’t use any names, so I think it’s okay. <Laugh>. speaking of accuracy, and you know, you, you actually mentioned it earlier, and I wanted to circle back to this cuz I was curious exception handling in the intake process. Yeah. Like what do, what do, what are the exceptions and like what does that do to the process? Like how does that tree branch any thoughts you had? I, I’m just curious.

MG: Yeah, so like from, from my experience working at within the insurance industry, yeah. And it was very, it was specialty lines. It was, it was very, very specific. Like some, some of those exceptions could be and yeah, some of the exceptions could be like, you want higher policy limits than, than what is standard, right? And so how does that change the underwriting? How does that change the pricing? Cuz cuz you have like your sta standard underwriting box, right? Where things fit nicely. And then a lot of times if there’s maybe so an exception could be maybe they don’t underwrite that particular industry code that your business falls into, right? Okay. So do, do you kick it out? Is there do you just, do you just decline that? Is there like a nuance to that where they might underwrite it but they might price you higher because it’s a little bit riskier or the policy is written a little bit differently?

Like what are so, so that’s one example, right? And I’m thinking commercial lines, specialty lines, yeah. Specifically on that. Like, like the, the big example, right? Is like cannabis, right? Like there’s a lot, like there’s a lot of, like if, if you’re a cannabis like, like sell that, like a lot of insurance carriers won’t, won’t underwrite that, right? Yeah. So that’s an industry code that’s an exception to, to that. But you may not know that. Like you’d have to get all the right data and the right N N A C S code in to understand that that’s what that, that business does. So that, that’s one example. So how does that kick out? And then like you could even from a personal line standpoint, just yeah. You know, exceptions would, so what’s interesting too is I wonder, this is not an area I’m an expert in, but I wonder if like, so you get, you get in the submission intake, right? And you identify that there’s an air, like it hits up against some third party dataset and you’ve identified that there’s an an inaccuracy on that submission, right? Like how does Oh, that kick out, right?

CW: Like, so yeah. Back to like the credit worthiness, right? Like yeah, yeah,

MG: Yeah, yeah, yeah. So, so you, you know, you said that, and I’m not, I’m not, I’m not putting out there that anyone is insurance fraud, none of that. Like the, a actual just errors like mistakes. Like you forgot to add the plus sign on, like, I have a second car, right? So it’s like, is there, like what’s the process when you’ve done validation and it says like they submitted their, their application saying that they had a single car and now they’ve got two others to ensure what’s the process? They’re like, you know, on, on a digital front end, like it says, are you sure? Like, cuz usually it’s, it’s hitting us against some database and it tells you what cars you have, but to the extent that you’re doing it manually or, or not in that system. Huh there’s some, some of that too. I’m sure

CW: That’s interesting it all. It also reminds me that one of the, one of the benefits you get from digitizing these workflows is, you know, you’ve sort of, you’ve sort of built a map of the process and some of it’s done by humans and some of it’s done by bots and some of it you reach out to an external database, but exceptions are places where it like leaves the normal order. And if you’ve built the system or bought the right system, then you know how often that’s ha happening. And I, I have to imagine out that out there, that there are exceptions that aren’t terribly exceptional. It’s just, you know, we haven’t built a process specifically for them.

MG: That that’s a great point actually. And, and what, what are the exceptions that we don’t know about yet, right? Especially if you’ve got a new product that you’re rolling out or you’re rolling it out in a new geography or something where there’s, there’s some nuance to you know, what, what questions you can ask on the application. Cuz going back to that highly regulated industry that’s done, you know, district by, by district or state by state, but 56 jurisdictions that, that differs. And so you, so internally actually those systems need to be programmed with the, the regulatory landscape in mind. So, so yeah, what is the process if you’re, you know, if you’ve issued, if you’re writing cyber in California versus cyber in New York versus cyber in, you know, any of the other states there, there could, there could be a different process that you internally have to, to account for.

CW: It’s like filling me with crippling anxiety, like the levels of complexity you’ve talked about, right? Like you have your jurisdiction, mostly states, but not exactly states, right? Mm-Hmm. <affirmative>. And then you have what line of business are you in? And then there are levels of nuance there in terms of what’s possible what data has to be captured. Like this is a lot, it’s a lot.

MG: It’s, it’s, it’s a lot. Again, back to it’s a ton of data being captured and some of it is, is captured and kind of, i I would imagine sits in a system somewhere untapped. And so what, what are the insights that you can pull out of that if you were able to access it? And so part of that is intelligent intake. You, you’ve got the workflow there, but you’ve also gotta have the system in place to, to be able to take in all that data and then Absolutely. And then push it, push it down the line, and that that comes into whole other, you know, other systems and things that you have to, to account for, to, to manage your reporting and everything else. So,

CW: So giant, massive problem space many systems, much complexity a lot to chew on. So we talked about it’s wise to carve this up and not focus on building like the end to end necessarily from the beginning. What are in, in, you know, on this you travel in different circles than I do. So what are you seeing in terms of what parts of the problem you know, various different parts of this ecosystem trying to solve when it comes to intake?

MG: So, so there’s the, there’s actually, we didn’t even touch upon it cause I talked a lot about the, the underwriting submission process, but there’s, yeah, there’s, how do you ingest even internal documents in, into systems, right? How do you, how do you keep a record of those? So you know, you’re in the procurement process when you’re, when you’re looking to, to partner even with, with third parties, what, where do those documents go? How do you ingest that information? How do you track those invoices? How do you, which invoices, you know, is they’re all differ, but they’ve got pretty much the same information in them. I understand that. But, but you’ve got that internally, right? You’ve got the billing system. So how do you, how do you generate those documents? How do you pull in all the right, pull in all the right information from your existing systems to maybe automate the, the creation of those documents and then to then push out to automatically distribute them, whether that’s digitally, electronically, or by snail mail.

Yeah. <laugh>, it still happens. Facts. yeah. Fellow fax. Yeah. yep. And my time insurance carries, I did use the fax machine quite often and that wasn’t that long ago. <Laugh>. That wasn’t that long ago. But, but yeah, I think you know, there’s, it, it depends on back to this digital transformation, where are you in that journey? What, what systems that are you’re using are ready to, and, and have the capability to, to connect with via API to Yes. Or just, you know, front end a a lot of these, these data inputs to, to use them. And so you, you can, you can do this at any stage within, I’ll call it the policy administration life cycle, but even within, just like the day-to-day, what makes the carrier move you know, from from department to department or unit to unit there’s, there’s a ton of opportunity, I think to, to automate anything that’s coming in and that’s information that, that should be captured.

Yeah. and we didn’t even talk about it on the claim side, right? Like you, how do, oh yeah, you get data, data in for the claim. How do you make sure it’s the right data? How do you follow, follow up? And, and that it’s can get into like, you know, intake of, of images and things like that from like, you know, a claims event and Oh, yeah, yeah, yeah. There’s a, there’s a ton there as well. But, but that’s not the topic i I force you to talking about today. Say that for another one. <Laugh>.

CW: I, well, I wanna foreshadow it a little bit with a naive question that I’ve had for a while now, which is I understand why personal lines would want to, you know, accelerate claims with automation because like better customer service and you want to keep people from going to the other company, but like, I don’t know, is why would a, a large, like large commercial lines insurer want to accelerate claims that that’s like how the money leaves. So I’m, is it, is it customer, is it actually just the same customer service story? Is there something else that I’m missing in terms of why someone would want to do that?

MG: That is a very complex question. Yeah.

CW: Okay. Well, we can, we can table it for now. We have, we have more episodes coming.

MG: I think that that claims related discussion would be that, like, that’s, that’s not a two minute

CW: Answer. Yeah. Stay tuned for the next episode, <laugh>. All right. Well then, you know, we’re coming up on the hour. What didn’t we talk about in this ecosystem? And, and, you know, your, your experiences on the, in the InsureTech investment side, at least the last, you know, little part of your career, so mm-hmm. <Affirmative> what haven’t we covered there? Because that, I mean, that’s a, that’s a big and massively growing area, and I wanna make sure we give it it’s justice before we wrap up for today.

MG: Specific to like intake or a writing

CW: Or, yeah, yeah, yeah, yeah.

MG: I, I think I, I might have mentioned it, and it, it goes hand in hand with what we talked about today, but what, from, from an underwriting standpoint, what we’ve seen a lot of is I’ll use that automation word again, but just supplementing the submission with third party data. Mm. That one helps you underwrite faster because you don’t have to ask all those questions upfront, right? Like you, you’re, you’re basically validating as opposed to, to asking. And so we’re seeing a lot of opportunities for for third party data vendors that have combined the data in, in unique and proprietary ways that can give you insights to what’s happening with your book of business even validate that, that what you’re underwriting is accurate. So some of those n I C s code solutions and as well I think new and distinct data sources as well.

So how do you, okay. Underwrite someone when you wanna use something that that’s not their credit score, right? So what, what are the alternative data sets that you can use to, you know, behavioral analytics is an area that that’s growing as well. So how do you, how can you take this information to understand how someone what kind of risk somebody is just by the way that some, some, some com some components that, like how are, how are they on social media or oh wow. Like, you know e and there’s tons of data op opportunities, even like with telematics, how does this person drive, right? Okay. Like, what, what predictably, what, what can this tell me? All of that goes into the consideration for, for underwriting. But again, you have to have a basis of information to, to hit up those, those third party right data sets up against, to, to compare and contrast and then ultimately make, make the right underwriting decision or, or figure out what variables you use to price a risk, and then by default, pass those on to the states and get them approved by the regulators to be able to, to use that in your, in your pricing.

CW: So you, the components of the pricing model have to be approved by the state regulators. I did not know this.

MG: Yeah. And so so the regulators are, are, I’m, I’m oversimplifying this, but okay, but are there to protect the, the consumer, right? Like they don’t want the, you know, insurance carriers can’t just use whatever they want to underwrite you because they, there could be bias in those models. There could be bias against, you know, certain demographics if, if you do that. Yeah. and so it, it, it is an art form. This is also something we could probably dive into of just what are the variables that you use in your pricing model? How do you weigh those variables, then are they approved by the regulators? And then there’s a whole art to how you submit that, right? Like, you don’t wanna give away your secret sauce Yeah. So that everyone can, can, can figure out how you price risk. But there is some, some, it can’t just be a black box. Like there has to be some level of transparency there.

CW: Some interesting information theoretical topics there, which I think I think marketing will definitely approve the claims episode. I’m not sure they’ll go for that, but I’m interested, I wanna learn more. Well this has been extremely informative and you’ve been listening to another episode of Unstructured Unlocked, my co-host for today. And as long as she’ll keep coming back as Michelle Goya and stay tuned for another episode coming to you where we’ll talk about all things claims. Take care.

MG: Bye everyone.

CW: Thanks for joining us for this episode of Unstructured Unlocked. You can find all of our episodes wherever you listen to podcast today, Spotify, apple, everywhere. Be sure to follow at indico data on Twitter and YouTube. Have a good day. Automated.

Check out the full Unstructured Unlocked podcast on your favorite platform, including:

Subscribe to our LinkedIn newsletter.

Unstructured Unlocked podcast

April 10, 2024 | E44

Unstructured Unlocked episode 44 with Tom Wilde, Indico Data CEO, and Robin Merttens, Executive Chairman of InsTech

podcast episode artwork
March 27, 2024 | E43

Unstructured Unlocked episode 43 with Sunil Rao, Chief Executive Officer at Tribble

podcast episode artwork
March 13, 2024 | E42

Unstructured Unlocked episode 42 with Arthur Borden, VP of Digital Business Systems & Architecture for Everest and Alex Taylor, Global Head of Emerging Technology for QBE Ventures

podcast episode artwork

Get started with Indico

1-1 Demo



Gain insights from experts in automation, data, machine learning, and digital transformation.

Unstructured Unlocked

Enterprise leaders discuss how to unlock value from unstructured data.

YouTube Channel

Check out our YouTube channel to see clips from our podcast and more.