Transcript: Roy Feig | Honeycomb Insurance | How Do You Build an Efficient Insurance Product?

Eyal David: [00:01] Hi friends, it's Eyal. You've reached Product Builder, where you can listen to the most interesting product conversations. My journey began in 2012, when I learned what product is at my own startup. Since then I've worked in the industry, and since 2017 my company has been advising a wide range of companies. My drive is to help all of us get better as product people. And before we start, if you enjoy listening to Product Builder, please rate us five stars on Spotify. It gives me great feedback that the podcast is giving you value, and it would make me very happy. So yalla, let's start.

Roy Feig: [00:35] Hi Roy, how's it going? Good, hello. How fun, did you just come from the area of the new office? Right, we're actually moving to new offices next week. I'm excited, everyone at the company is excited, yes.

Eyal David: [00:46] You'll tell us where in a moment. So with us is Roy, VP Product at Honeycomb Insurance. Right. You're moving from Herzliya to? Moving here, to Tel Aviv, Midtown. Gorgeous, so it was really easy to get here. Right. Tell us a bit about Honeycomb so we get a sense.

What Honeycomb is: an insurtech that is also the carrier

Roy Feig: [01:03] Sure. So Honeycomb is a startup in insurtech. Meaning, a technology company, we build technology, and we're also an insurance company. Meaning, unlike some of the tech companies in the insurtech world, we mainly build technology for ourselves; we are the insurance company. We insure buildings in the United States in commercial lines, meaning for businesses. Think of our typical customer: it's an investor or a group of investors who own properties, and all the properties are rentals. They don't live there; that's very important, right? That's the difference between personal lines and commercial lines. These are usually much more expensive types of insurance, and the process of... I'll say underwriting, and let's say underwriting is the ability to assess the risk or assess the property, whether we want to insure it or not, is much more complex and difficult. And we identified that this is a segment we can enter and really change how things work in it, and show, unlike the rest of the industry in this segment, that we're very efficient, and that we can create a process where our efficiency generates profits, when the industry doesn't count on those profits. We're really changing how the business model works.

Eyal David: [02:27] It's probably all about efficiency; let's break that down in a second. You talked about buyer groups or developers, that's also a very Israeli thing, right? There are a lot of Israeli investments there like that, but then you're actually targeting a different segment. What does go-to-market even look like for something like this, an insurance company that...

Roy Feig: [02:43] Yes. So we, let's start for a moment with the product hat, one second, and I'll say how I see the... who we target and who we build products for. So I said, one thing is that we build products for ourselves. Meaning, we build a product for the underwriting team; the underwriting team is the body that assesses the risk, whether to insure a property or not and on what terms. And we target, our external customers are really insurance agents, they're the main channel. It's the main channel in this segment in general, and for us too it's the leading channel. But in the end, the customer, the one who pays us the premium, is the insured, or the group, or the company that owns the properties. And we also have a channel, what's called direct, where they can bypass the agents' channel and simply go to honeycombinsurance.com, enter their property details, and get a quote immediately.

Eyal David: [03:39] So let's stay on that for a moment, and we'll get back to the questions I have there. So what are the property details? Is it literally the building, its address,

Roy Feig: [03:47] you know what it looks like. Yes, so the process is entering an address, that's the most basic thing, and based on that address we know how to pull in a lot of data, but in addition there's a fairly short questionnaire, we made it very simple, with things like the area, when the building was built, questions like that, what coverage amount they need. The questions are pretty basic; usually people in the field know how to answer them, especially if they're represented by an insurance agent, they run through it very fast. But we don't rely only on those answers, we do our own risk assessment work, and that's the heart of the work. That's where most of our investment goes; most of the company's IP sits in those areas: how we build a system that knows how to take that data, add a lot more information we bring in, and in the end analyze all that data together and reach a decision whether we want to insure this property or not, where of course we'd want to insure properties where the chance we'll need to pay a claim on them is low,

Eyal David: [04:55] So that's the heart of the company. So take us back a bit: how long has the company existed, can you tell us about it, did the founders come from the field, what does it actually look like?

The founders and the market failure: why California lacks insurance

Roy Feig: [05:04] The founders touched the field a bit, but you can't say they really came from it. There are two founders, Itay and Nimrod; Itay is the CEO, Nimrod the CTO, and they simply identified this problem in the market. Each came from a somewhat different angle, they knew each other from before, and they decided there's another way; it's sort of their vision. There's another way to deal with the market's inefficiency. You also need to understand, it's a market where in some states it's much more pronounced, in the US, and in some less, but for example, if we take California, it's a market, a state where it's very hard for that typical customer I mentioned to find insurance. And the reason is that insurance companies took losses there; there are giant companies there, and over the years, or in recent years, they took a lot of losses because of wildfires and all kinds of natural phenomena we don't know here in Israel, but that definitely exist in the US, states like Texas with crazy-sized hail. So there are really big losses for insurance companies there, and they simply do categorical underwriting or categorical risk management: they just exit a state or exit an area, or across the board they say buildings over a certain age won't be insured. And then you're left with people or companies like that who want insurance and can't get it, and that's where we enter this market, because in our view, even in the founders' view, it can't be that there are no good buildings to insure in California, right? It doesn't make sense. So let's build the mechanism that lets us identify those buildings we'd want to insure, that we thought were very high quality and should be with us. And that's how it started, and of course people from the field joined the company, people who really know how to do underwriting work. So there's a chief underwriter, and we work closely to get into his head and the underwriting team's head, and see how we can take all that knowledge and the manual, Sisyphean work process that he and his group do, and by the way the whole industry does, and turn it into something easy for them if they need to do it manually, or even automatic, so they won't need to look at any request at all, and it'll just happen,

Eyal David: [07:29] you'll get a quote, because we think it's a good property. Just close the loop for me: how long has the company existed? You know you've been there about three years? How long? I'm almost at three years.

Roy Feig: [07:37] We sold the first policy in June '21. Got it, so it's somewhere around there.

Eyal David: [07:44] And actually tell us for a second about the industry. What does a traditional process look like today?

A relationships business: what traditional underwriting looks like

Roy Feig: [07:51] So as I said, in our segment it's mainly through agents. Pretty much as soon as I joined, they told me, listen, this is an industry that's a relationship business, right? Meaning, it's a lot of conversations, probably a lot of negotiation of sorts, and some relationship between those agents and those underwriters, the ones who decide. That's what the market looks like, and insurance companies try to assess the risk by investing, let's say, half an hour per request at minimum, going into all kinds of data sources, wanting to see photos of the property and looking at the roof. That's what the industry looks like. We're basically coming from a point where we're challenging that relationship a bit and saying, here, there's an easy interface, it gives an answer immediately, and it can also give a positive answer in places where others won't,

Eyal David: [08:48] And that's exactly where we come in. And your focus is on underwriting? Are there other areas you focused on, that you identified?

Efficiency, loss ratio and a product team organized by persona

Roy Feig: [08:55] Beyond, of course, a fast user experience and all those things. So that's it: the underwriting world, the efficiency and the automation that are supposed to help us with two main metrics. One, of course, to help grow sales, because everything is done fast and efficiently, but also to make sure we keep a low loss ratio. Loss ratio is how much we have to pay out divided by how much we bring in in premiums. So that's the focus, but beyond that, as I said at the start, we're an insurance company. Meaning, we have users, there's an interface, and the product we sell in the end is a PDF that says insurance on it, so you need to make sure documents work well and properly. So we have a chain of customers, whether it's different channels and in the end the insured themselves. And that's how we started, and we expanded to a few more insurance types that complement, or really help us, say, improve the funnel, so we can offer insurance to those we previously couldn't accept, and we built them an insurance product better suited to the type of property or the risk that comes with them. So describe for us what the structure of the product group looks like. So the product group is made up of three product managers, a UX person and a data analyst. And how is it split according to the different areas you mentioned? What a good question. First, it's work to think about how to split them, because in a startup, especially when you arrive pretty early and you're the first product function, I did everything. So I know everything, and not only me; engineering was small and underwriting was small too, so when everything is small you oversee everything and don't need to think much about who does what. And really, afterwards the challenge is how you grow or build a team where each person truly has their own area and knows it and understands its problems and opportunities. So the way I split it, and how we work today, is that each product manager has a persona and a metric they need to improve. So for example, we have the persona of the agent, the insurance agent, who has an interface in the system, and we want to make sure there's good retention there and they come back to us, keep requesting insurance quotes, and actually buy the insurance through us. So that's one PM. The second PM works on underwriters and underwriter efficiency, meaning on the underwriters, underwriting, the risk assessment world. And the third focuses on slightly more innovative areas like partnerships and other channels we're starting to work on, and also on the insured themselves in the end: how we make sure we don't forget who ultimately pays us,

Eyal David: [11:55] because really, right now most deals are closed through the agents. So let's stick for a second with the first persona and the KPI you mentioned. You basically said that everything, as you heard about it at the start, is a relationships business, right, and what you're saying is there's a relationship between the insurance agent and the underwriter, and so we need to be better than that so they come back. What does it look like today? Like, do those who come in return, are there real problems there, what does it look like?

Agent retention and low-touch onboarding

Roy Feig: [12:27] So those who come in do return, but we have a goal to improve it. The idea is to see how we create the most positive experience right from the start, when sometimes not everyone gets a positive experience, because we don't approve everything, certainly not in the first products we launched, because it was important to us to make sure we only accept the best properties. We were at the start of the road and really wanted to avoid claims early on. Today we already have insurance products that can allow more, but it can be very frustrating: an agent who comes in, and the first property they try to get a quote on is declined, and the second and the third too, and then, okay, even though our process is very easy, I'm not going to waste my time on the fourth property. So we need to make sure we manage to help them reach successes. That's one of the challenges: how we make sure onboarding happens low-touch, without a support or revenue person being with them, helping them use the system or teaching them and explaining what's called our appetite, which properties we'd want to insure. And that's where the product person's work comes in: how the system itself represents the relations now, how you don't need all the people around to support,

Eyal David: [13:49] and it's just clear and easy and you really walk away with a positive experience. Totally, that's where I'm taking it, and I'm curious. If I were an insurance agent, wow, I have great relations with the underwriter. I think you'd win me over either by me really putting in a deal and it going through, or by you giving me a bigger cut, right? So I'm wondering where that sits product-wise, because if they come with a deal that, wow, neither the underwriter nor we will pass, then how do we pass it and how do you give a good experience here? That's interesting, I'd be happy for you to answer that. And also, do you actively look at areas where you know there are opportunities, and then you reach out to those insurance agents and bring them in: hey, come look, here I have... what does that actually look like?

Roy Feig: [14:28] Yes, so how we reach out: we don't do underwriting or risk assessment for the whole US in advance. We really wait for requests to come to us, but we, I'll mention that today we write, or actually sell, insurance in 18 states in the US, and of course we're expanding, but it's territorial work. Everywhere you need salespeople and underwriting people who are experts in that place, and in the end that also translates into our models' understanding, which assess risk per location and how to look at it. But we bring on the big agencies in each area. That's how we get to properties; we don't suggest to them, this property is good for you. Totally, no, I think what's his initial incentive, that he came, or that, uh, they put in the first deal and it didn't go through. Okay, what can we do now? So first, explain to them what should happen in the second deal. Meaning, let's learn from the first deal what didn't work, that's one thing. Second, those agents also have a relationship with their insureds, or whoever they're trying to get the policy for. And that relationship is important too and we can help with it. So even if we can't give you insurance, how can we create a situation where you as an agent can explain it very well, and it's easy for you to explain, and you can recommend to that property owner what they need to do, or maybe where they should go if not to us.

Eyal David: [16:00] Meaning, you can find those places where it's still worth it for them to go through our system, and they'll always get value. And another question that comes up for me: as an insurance agent, will I usually upload one property, or do I have some Excel with lots of properties that I throw all in and...

Roy Feig: [16:16] It really depends on the agent, but sometimes it's one property. Sometimes they test us with one property, and they actually have a lot more business to bring, and they want to see how we, or how we do at renewal, meaning after a year, that we still exist and everything's fine. So there are those, there are those who immediately bring some huge spreadsheet, and there are those who simply have a client with lots of properties, really clients with 50 or 100 properties, who need insurance on all those properties.

Eyal David: [16:44] So there's also a process where, instead of one by one, we say okay, bring some bulk sheet and we'll give a quote for everything together. So what's the KPI that product manager is actually measured on? Closing deals, new customers, what does it look like?

Roy Feig: [17:00] No, so the KPI is producer retention. The product manager doesn't close deals, we're not measured on that, but rather on managing to move the metric of how many of those appointed agencies come back to us and generate policies every month. When generating, buying a policy, that already means they were there before; there's some pipeline, they've already submitted, some were declined. Meaning they need to submit enough quality submissions, some of which can be approved and even convert and mature into a policy. That's the metric. So because your cycle is really quick, within a month you can actually find a lot of those, like, you can already bring value and actually see them come back. Interesting.

Eyal David: [17:43] Okay, so that's probably very different from the traditional way, I assume. Who are your other competitors in this industry?

The competition and who the insurance agent is

Roy Feig: [17:51] Most of the competition is the incumbents, the big traditional insurance companies, which, as I said earlier, can simply exit whole areas or not renew many of the policies they already write or hold in some area. There aren't many companies like us in the segment. There are a few more that are a bit more technological, but they're in slightly different segments, whether that's smaller properties or riskier areas, dangerous in terms of storms and hurricanes and such, which we don't enter. So we don't really have direct competition from a tech company, but competition exists, because in the end we're coming to do something a bit different, and we need to change a certain mindset or way of working of those agents, and we're competing against players who've been there a long time.

Eyal David: [18:50] And with very deep pockets. So give us, just so we can get it straight in our heads, what does the average user look like, that insurance agent?

Roy Feig: [18:59] They're people whose goal is to grow the policies they hold as much as possible, to serve as many users as possible, meaning their time is very important, and they would very, very, very much like ease of use, ease of doing business, and not to deal with problems or requests or passing along all kinds of messages between the insurance company and the client, like he didn't pay or he forgot to do something, or... They need to justify their place. Exactly, but they want to do it with very little effort. Meaning, the more we help them streamline their work process with the insureds, the happier they'll be, because their goal is to bring new business and not to go back and resell the same business they already have. So you're kind of, as they say, their secret weapon. Totally. The nicest thing, or very interesting cases that happen, is all kinds of agents who come to us because they lost a deal to another agent who came with Honeycomb. And that's cool; it's already happening in several states where our presence is bigger, and you hear stories like that, and it's fun to hear. Of course, those are probably also the best customers. Totally. Tell me, what does your day-to-day actually look like when you get to the office? So I make a point of arriving early, early being around eight. I clean the coffee machine, make coffee for myself and for whoever manages to arrive at eight. And then I have some quiet time with myself where I look at numbers, see what happened yesterday, go over things that happened in the US while I was asleep. And then around ten we have a product team daily. We share, talk, and officially start the day. And then it's time for a lot of conversations, whether with my team or with engineering or with anyone who's in Israel, broadly, but it's mainly those functions, engineering, data science. And the afternoon hours are conversations with the market, whether with our underwriters, with the sales team,

Eyal David: [21:14] with customers. So the focus shifts there. Gorgeous. I just want to understand: first of all, what metrics do you look at in the morning? Can you share?

Morning metrics and measuring feature impact

Roy Feig: [21:24] Yes. First, to see what happened in terms of sales. How many policies were closed, and then also dive a bit into which segments. Meaning, whether it's a new product, or a product where I'm looking for the product impact on that metric. And then I go to another stage of the funnel: how many submissions there were, what happened with them. And I try to understand and see whether things we're working on, I can see they're still there, or things we shipped recently, they exist, they have an effect. If we shipped some feature that's supposed to improve efficiency, is there really more automation? Those are the things I look for. And how easy is it really to identify a correlation between product and business? With efficiency, probably yes. Exactly, it really depends on what you're doing. For example, we now have an effort to improve the underwriters' work. So one way to improve it is to make a lot of changes in their flow. How they see all the data in an organized way, how we make things accessible to them in repetitive places, we'll give them a dropdown instead of having to make up a reason every time, right? So it's just templates and they can pick something. Those are things that are a bit harder to measure, compared to, let's give a tool that simply decides for them, and then it's very clear what happened, right? Suddenly, from 40 requests a day that someone has to deal with, you see they no longer need to deal with them, and the requests simply close automatically. So some things are much easier to see. And some things are very clear-cut, if for example we launched a new insurance product. You color it a different color in the dashboard, so it's very easy to see its impact. It didn't exist before and suddenly it's born, so that's easy. So it really depends on what, on which feature, but most things, for us, are aimed at efficiency, risk management, and ultimately growing sales and our attractiveness in the eyes of agents and insureds.

Eyal David: [23:36] So take us for a moment into the underwriters' work, and the example of, hey, let's put something automatic against it. What did that look like from the moment you joined, or from the moment you started working on it, and where is it today?

From zero automation to 60% with models and AI

Roy Feig: [23:50] When I arrived, there was zero automation. Oh wow, okay. Let's put it this way: there were models, we brought in a lot of data, but there wasn't really automation. There were a few, let's say, zip codes where we simply didn't write because we'd excluded them in advance, but every submission, someone had to look at it. And at first it was a process of, okay, let's understand what these people look at, and give them a mechanism that, using rules we'll think up together, let's give you some rules, just configure something, and here, this you don't have to look at, this you do have to look at. But the interesting point was when we managed to identify their risk-assessment patterns and build them as a model that can predict what they would say or decide, and simply do it in their place. And then, on top of the automation layer, we added an AI layer that decided on its own. So today we're at around 60 percent of requests approved or declined automatically; nobody looks at them. And now we're in a project of, let's grow that from 60 percent to 80 percent and make sure most of the business is decided automatically.

Eyal David: [25:16] Because basically, today it's clear why you're doing it, but today lots of insurance agents, lots of customers who want to give more business are knocking on the door, and it's stuck on scalability, basically?

Roy Feig: [25:28] It's stuck, or about to get stuck. Meaning, we're a growing company, as you said at the start: we started with 20 people, today 130. So obviously that comes with the company's growth, and it constantly grows sales and really our need to handle more and more insurance requests whose risk we need to assess. So one option is, okay, let's hire more and more of those people who know how to decide. And the second option is, okay, let's hire super developers and engineers and data science, and I see that we can actually build a startup here whose job is to provide a technological solution, and our cost on every new policy will be nothing, or, I don't know if it's really nothing, but it won't be bringing in another person who has to assess the risk. How? Not definitively, but I'm saying, will this jump from 60 to 80 put

Eyal David: [26:25] you in some safe zone where now you can grow and go after anyone you want, or is it still not there? We don't make it conditional on that; we're still growing.

Sampling models and measuring accuracy

Roy Feig: [26:34] Just curious, why that number specifically? Because, first, we needed to set a goal, so we set it, but in general, also from looking at those areas that today we can't decide on, we managed to see how big they are and where we can get with the next jump, when the goal isn't necessarily to reach 100%. I think it's always good that some of the business passes through our hands. Also, every model we release, we sample it. Meaning part of it still reaches that same work pipeline of the underwriters, who look and assess, and then we feed that data back to the data science group and say, okay, this model works well, this one now needs to be revisited, and let's make sure we get back to the numbers we'd expect in terms of our accuracy. Nice, you saved me the benchmark question.

Eyal David: [27:30] Another question I asked earlier: so what's actually the metric that indicates correctness? What does it look like?

Roy Feig: [27:38] Because it's not enough to just stitch together lots and lots, right? So maybe I can give an example of a few models and what correctness means, but the bottom line is we need to decide at the property level, at the submission level, whether we want to insure, and if so on what terms, or not to insure. So that's the underwriter's decision on a property, so what would we have decided? That's one check. But along the way there are all kinds of models we run that help the decision. For example, what's the condition of the roof? What's the condition of maintenance based on a front photo of the building? All kinds of other, more specific models that check features. Inside the structure. In the end each of them also has some level of quality or accuracy that we need to make sure is there,

Eyal David: [28:31] because obviously it affects the final model that decides in aggregate. And going forward, the focus is obviously to stay in the US, because you're probably also limited by the type of input you can receive and handle. I'm saying input because, hey, you said I have photos of the property, I have all kinds of insights about the building, so other places probably not necessarily. Beyond that, in insurance it's probably, wow, there's a lot of room to break through inside the US, but the brand also matters inside the US. So we're currently in 18 states, no thoughts of going outside before you stitch up more inside the US, I assume, right? Yes, in the coming years we're staying in the US. Right now there's no plan to go out; it's to expand within the US and to add more products in the US.

Roy Feig: [29:15] But this thing and this technology can work in other places. I, you could reach places near the US, you could expand to be equal, but I think it can absolutely reach other places, but right now it's not in our plans.

Eyal David: [29:31] And today, out of the 18 states, there are surely a few at the top. Which states are actually at the top? Yes, there's that thing, there's the top 3, there's Illinois, where we started.

Roy Feig: [29:41] Why? For a very bureaucratic reason: it was easiest, or fastest, to get approval to sell our insurance in Illinois. That's why most new insurance companies start there; it's simply easy to get approval there. But California is a very strong state for us, also because of the need and the market failure there. And our product is very attractive in Texas too. But there are more; the other states, some are bigger and some we're trying to grow. Not all of them grow at the same pace. It really is a geographic product: every state in the US has the big players in it and its own dynamics, even at the agent level, how much commission each one gets. So there's really a need for geographic understanding, and in each such place we need to see how we break through. But the core is the same. The core is how I know how to assess the risk, and let's make sure I can support the other things in the product, whether that's with the agent, or making sure I have some coverages that may be missing in a particular place. Yes, there are places where maybe flood coverage is more interesting, compared to other places where, I don't know, earthquake coverage is more interesting, so it depends.

Eyal David: [31:01] Like, at any given moment you could also grow too big and then one of those giants buys you; that's also an option, because they're all very strong players.

Roy Feig: [31:10] That could be, but right now we're on our own path and trying to grow, and not to grow only in one state or in the top three I mentioned, but to make sure we grow in all the states, and there are a lot of opportunities there. And another question: tell me, how do 130 employees and three product people work? We're always busy. How does it work? It works. We try to match some ratio between the number of product people and engineering people, and hence also the split of each product person to some world or some persona they need to understand. And right now it's a ratio where we feel the pressure but manage it, and I think sometimes the pressure is also good, because it makes you make sure you're working on the right things. After all, in the end we also need to prioritize our time and make sure our focus is in the right place, and it helps make sure all the investments that we ultimately dictate, what engineering works on,

Eyal David: [32:14] also go to the right places. Nice, so you teed it up. Tell us, what do you think are the most important product skills for a product manager?

The most important product manager skills

Roy Feig: [32:21] I think something I look for is, simply, people who get it. That's a general statement, but in the end product people need to make sure the important things get done and actually happen. If I break it down into slightly more specific things besides getting it, it's the ability to really listen, to listen to many people, to understand what they really want to tell us, whether it's salespeople or customers or engineering or our fellow product people. Everyone has some story or something they're trying to tell us, so be very attentive, that's on one hand. And on the other hand, make sure we connect it to the opportunity or the problem, which really connects it to data. I think the combination of the ability to listen and the understanding that this thing is significant is the critical thing in product management, because in the end it's understanding that I'm working on the right things. And that comes before which solution I bring, because a solution can be something iterative, or I can solve it another way, but that's better than coming with an amazing solution, for example, to a problem that's too small or isn't interesting. So I think that's the area that's a bit more significant in the product management role. Gorgeous. What does the process of driving toward a solution look like for you? I assume... I can see you're a process person, tell me what it looks like. Again, it looks like simply listening to many sources. Now, we have people who already know the industry better, and it's nice to listen to them, and they help us understand problems or tell us what would have happened without Honeycomb. So that's one way. A second way is to really see in our numbers what works, what doesn't, what we expected to happen and isn't happening. So we look at the data a lot and really try to connect it to the insights of those who understand and those in the field. And then we decide: these are our agendas, this is what we're coming to solve, while very often we don't know what a solution will look like.

Eyal David: [34:34] And that's how it happens. And how do you examine the data? Meaning, say I worked on some feature; content probably matters most, I'm setting aside for a second the UX we talked about, but content really matters and I want users to see it to get feedback on it. So what does a cycle like that look like for you? You say content, what do you mean? I'll explain. I now have the underwriters; they need to see as many approved deals as possible, but of course ones that make sense, and I don't want to waste their time. So how fast, that's an example for underwriters; it could be an example for users that I'm now showing some content or other.

Roy Feig: [35:10] What's the actual process, how fast do you expose it in order to optimize, to identify a mistake? It's something where we try to expose the features we build, or the models, or whatever we can offer, as fast as possible, of course in order to be able to iterate, to improve; we really try to make that happen. We have all kinds of ways to do it when it's less impactful, or could cause damage if we think it's not mature enough yet. For example, we can release models where we just show the result and don't yet make a decision based on them, or models that will only work on some of the cases, where cases we think are more problematic, we'll decide at first not to handle with those models. But we try to get to market very fast, and with the external users too, our customers like the agents, we simply release things and see what happens. They learn like design partners; do you have a few people like that you can... There are a few we talk to, but usually if we release, we release almost every feature to everyone. There are some features where we definitely do some gradual rollout and want to see it's received okay, that we understood the problem well,

Eyal David: [36:32] but most of the work simply happens and goes out. Right, and you also have the matter of territories, so doing GA is probably much more convenient.

The challenges: distance from the market, onboarding and growth

Roy Feig: [36:41] Where do you usually start? Is there one state that's the one? It really depends on what. Some states need something specific, and sometimes you do something for that state, so obviously they're in focus, but most of the things we do are across the board, and whoever's interested will use it and whoever isn't won't. But there are data points, for example, that are very, very geographic: there's a data point specific to Los Angeles and a data point specific to Chicago. Meaning there are those cases, and if there's a real need there, we add something very specific to the state. Wow, okay, so what are your challenges? Yours, the product department's? Um... so the challenges are, first, all of us, me and all the product managers, come from other worlds and not from insurance. I think one challenge is understanding the world we're in; in the end we're building solutions for that world, and even the fact that we're in Israel, far from the market, is a very big challenge in my view. It took me time to understand and manage to speak the language of insurance, and I see the PMs in the group too, some are still trying to get there, and they'll succeed, they'll get there, but I remember how I was when I started, and it's a hard process, and it's hard when you're far away. And I see other functions, not PMs, that we have in the US, but other functions that join the company, not from product and not from insurance, and they already speak the language very fast, they get insurance much faster than us. And I think the distance is a barrier, so that's one challenge: how we get closer. And how do you deal with it, like, what does onboarding look like for you? What did you think? So onboarding means, first, exposure to a lot of content and recordings and calls, and of course onboarding with every group we have, even when it's far away, making sure people get the content, but you don't absorb it immediately. It's a process that takes time, so you need to travel or fly to the US and conquer, there's no way around it, you have to, and you need to be available at the hours to talk with customers and to talk with our counterparts in the market, in the US, and it just takes time. Got it. Other challenges? We have growth challenges, we're constantly growing. I described a bit earlier, you know, the work of one product person, then two, and suddenly three; now we're already four product managers, including me. And how we divide responsibility among ourselves, and also we don't only work with ourselves, we have counterparts in engineering, so how do we create this growth together? How do we make sure there's a good connection between the product people and engineering, and how each product person knows how to represent their persona in the market well, and make sure there are the right resources to do it. So this thing is constantly changing; when I say this thing I mean our work process, how we do planning, right? When I first arrived, all of engineering was together, sitting with me, we were super-planning. But it was small, it was contained. And slowly this changes, and it keeps changing to support the size, to make sure we manage to give PMs and teams the independence to do what they need, that we don't waste people's time sitting in very large meetings for nothing. So those are challenges, and it's constantly changing. So it's much easier to do it with people who understand that change is part of the job.

Eyal David: [40:39] So you're saying, I'm reading between the lines, that to make the work more efficient you actually do a lot of planning meetings.

Roy Feig: [40:47] And that's probably now, that's another challenge, it's a matter of your time, really, of the product people's in the end. Yes, but we reduce them by having each person be in the planning that's relevant to them, for example. It just keeps changing. We also look at ourselves as a body that's constantly growing, and what fit us yesterday probably doesn't fit today. So we, and I, are attentive to what's happening, attentive to my PMs and attentive to engineering, and try to make sure that together we have good processes that can ship products fast in the end and give solutions to those who need them, whether that's helping sales or helping the underwriting team do their work fast.

Eyal David: [41:28] So basically what you're immersed in now is probably planning for next year. Do you have time to do all this?

Roy Feig: [41:36] You find time. As I said, you find time for the important things. Not everything happens, but the important things happen.

Eyal David: [41:43] I understand. Okay, I won't take us into '25 planning; you really shared a lot and it was very interesting. Let's hear a bit about you so we know who we're talking to. So Roy, do you want to run through your career, which actually started in medicine, right?

Career: from medical devices to Kenshoo to insurance

Roy Feig: [41:58] In medical devices, yes. The first place I worked was a startup in medical devices, AngioSlide, a company that doesn't exist today, a startup for balloons for catheterizations. I was there; I was looking for work after I finished my engineering degree, biomedical engineering, so I was really looking for work in the field, and somewhat strangely, for someone finishing biomedical engineering it's hard to find work in medical devices. Because usually, or at least at that time, they strongly preferred either software engineers or mechanical or electrical engineers, depending on the area. And they didn't know what to do with those biomedical engineers, but in the end I found something. And I started there as an R&D engineer, I was there almost 7 years, and did several different and interesting roles there, and it really was a fun experience, because every so often it changed in terms of responsibility and what we did. We also built a production line here in Israel, and then sold it. And in general it's a very satisfying field in the sense that you're building a medical product. It's even very exciting to stand next to that doctor doing a procedure and opening a blockage in someone's heart, and you're standing next to him, and you say, wow, that's my product in there? That's a lot of fun. But I'd had enough there after that period, and I also have to say it's a world that moves slowly; it's a world where a release is, say, the founder comes with the idea and the release happens, and that's the company, and it's a lot of regulation. So after that I moved to a company, Kenshoo, today called Skai, where I had my first product management role, although I think in the previous company too many of the elements touched on it, or those same needs of understanding a problem and how to solve it, because it was still a small startup where we did everything. So I think the mindset was there, but I didn't really know what product management was, and I said, yalla, let's try and see what it means. It was very strange at first, and I didn't really know what I was doing, I think. I remember the first feature, I also remember the first day, when I really didn't understand what was going on there. I arrived at some, there was a release right then, and I said, what, you're doing a release? Yes, we do a release every two weeks. Not suddenly, like, it was, it was very fast. So it was a kind of shock at first, but inside the company, inside Kenshoo, which builds a SaaS product for managing online campaigns, meaning all the activity on Google or Facebook or Amazon, it was a product that lets big ad agencies and big brands manage all their online activity. So that's where I learned what it actually means to be a product manager, and during my journey inside Kenshoo I got the chance, or it worked out well for me, to build a few products or areas that were new, which exposed me to more functions in the organization, and that was already less a product manager on a team and more, go build support now for... for example, a new channel called Pinterest, which didn't exist, and that was nice, because you're exposed not only to your customer in an existing product, but you're going after a new area. And after that there was more work with Amazon, which was new and interesting at the time. I remember we were a small team working with Amazon, which had just come out with their Sponsored Ads; for them it was some beta we joined, and it simply exploded. It was a crazy growth engine, and it was a wave we caught with really good timing. Those were experiences that were a lot of fun and taught me a lot, and exposed me to all kinds of functions in the organization that I think, in the role, inside a team, are a bit harder to be exposed to. So that was my time at Kenshoo, which was also almost seven years, and then I changed fields again and came to the insurance world. I see a pattern here. Yes, yes. Tell me, and what was it actually like,

Eyal David: [46:29] what's it like now to work, you worked with corporate America, you said, what's it like now with insurance agents, with this world, underwriters?

Roy Feig: [46:39] It's a very different world. It's a world that's much less tech-savvy; it uses or expects this technology to serve it much less, in the same way as, let's say, marketing or online advertising, where it's obvious that's what's needed. So there's a challenge here that needs bridging, some gap even in language or in understanding how we build a product, whether you can go out with a product that isn't fully mature, whether you can get the understanding of how tech products are built and even be part of a company like this. We have people who come from the insurance world and use our product, and they're part of our process as product managers to understand the problem and think about solutions. So how do we create that language and their understanding of what this release is really going to look like, what this first version that goes out is?

Eyal David: [47:41] That's a challenge that's also fun in a way. So, toward the end, you pointed to another challenge, as I understand it. The user experience, a multitude of personas, some of them not techie. At the same time, they're probably used to other products, so you do need to keep them engaged. That's what I meant when I asked you earlier who the user or persona is, because I really wanted to understand it better, but you answered it, and great, very interesting, I think. Toward the end, let me ask you: first of all, it was very good in my opinion, it was interesting, thanks. Where do you see yourself in 10 years? We said you have cycles, a sabbatical year, 7 years. Where do you see yourself in 10 years? What do you wish for yourself?

Roy Feig: [48:24] It's hard for me to wish myself anything in this period. Honestly, I wish for us that in 10 years we'll have an amazing country here and a good place to live, and that the hostages come home tomorrow and not in 10 years. Gorgeous. And really, one last question to wrap up.

Eyal David: [48:41] A product manager who now wants to enter a new field, and you really did that every 7 years. How do you recommend, when you start a new job or interview in a new industry, how do you recommend actually getting into it?

How to break into a new industry

Roy Feig: [48:56] It needs to come with a real desire to learn and understand, that's one. And second, be curious and really ask why, many times, because sometimes product management skills really help, and help you think a bit differently. We too have examples of things we didn't build, or built very differently, only because of questions like, wait, but why? The fact that the industry does it this way doesn't necessarily fit us, whether for a technical reason or because maybe it really isn't the right business direction. There were a few things I'm very glad we didn't build, and there are things we built very differently, to make sure we built them in a modular way so we can use them in other places. Meaning, that place of, I really don't know, I need these people who are the experts in the field, and not thinking I'm inventing a new insurance world here, for example. So hear and listen, and also ask why, why is it like this, and can this idea work. And many times I hear no, but in the end you find the way. So be people who get it, curious and focused, basically.

Eyal David: [50:06] Right. Gorgeous. Roy, thank you very much for coming, I had fun. Me too. Yalla, bye bye. Bye. Hey friends, thanks for listening. If you found this podcast valuable, you can subscribe and follow us, of course, for more episodes on Spotify, Apple Podcasts or any other app. And of course if you didn't find us on some app, I'd be glad if you wrote to us. We'd really love five stars on every platform and for you to follow us, so more listeners can discover us and find the podcast. You can also find previous episodes on any app or on the YouTube channel; we have links in the description. So until next time, yalla, be effective, and bye bye.