Making Risk Flow | The Future of Insurance
00:28:53 · Jamie Wilson, Juan de Castro
Like, do I want them to see what the marginal impact of this policy is to my portfolio? Do I wanna see them what other quotes that have come through recently look like this one and what the pricing kind of decision was at that point? And kind of say there's all this opportunity, but you just can't get there with spreadsheets is the key point. It is the kind of, that's always been the limiting factor. And what love about my job to a certain extent is I get to go and kind of work with pricing actuaries across the market. Not all of them have been restricted by like kind of just Excel in the past, but a lot of them have. And I get to sit down with them and talk about, okay, now you're not limited by Excel. What's your opportunity? Like what could you do? Like kind of what can the future look like for your pricing function and for your underwriters? And rather than pushing out pricing tools that are almost a burden to underwriters, kind of saying, how can you build out pricing tools that really enable your underwriting team? So yeah, I think kind of that's what the future can look like. [00:29:52] S0: Definitely. And then Simon, you're sponsoring this transformation within Markel too in providing insights. How are you thinking about it? How are you thinking about moving from providing a number, a technical rate, into providing insights to underwriters?
00:30:09 · Simon Cooper-Williams
[00:30:09] S1: I mean, I guess our underlying pricing strategy that we've done recently is relevant to this. And Jamie's comments there, leads it into this, I think quite nicely. But one of our key points of our strategy is that pricing has to be fun, which I think is quite a nice point. And the subtleties behind that are like, it's gotta be easy, it's gotta be integrated. We re don't keying, like single pane of glass feel has to be there. We want people to enjoy it and not be, the system shouldn't be a burden for when people click on it and go, I've gotta do my pricing now. Like if you're trying to change behavior and make pricing be like a valued and critical part of the process, it's gotta be enjoyable to use. And that's really what we're trying to do. We want more data in there. Examples we're doing now, now we've moved to HX, we're putting a ton of external data in there. We don't know how it affects the price. We're not having it load or discount the price, but we're putting it in and it's shown as a rating factor. It just doesn't do anything. And we couldn't use to do that and people can see it and go, okay, I might change my decision now. It might not affect my technical price yet, but at least it's there and people are looking at it. And, you know, Pete, I think the whole point of what we're doing is just trying to make that pricing process more insightful, but then also easier for everyone at the same time.
00:31:26 · Juan de Castro, Simon Cooper-Williams
[00:31:26] S0: So are you thinking about it capturing more data through hyperexponential, more external data for each risk. Initially, probably not changing your rating algorithm, as you said, based on that. But as you capture it, then you can run analysis later on and say, Okay, did any of these factors correlate, or any of these data points correlate with the loss experience? And then I guess the additional beauty of not using spreadsheets is you can seamlessly start embedding those data points into your writing into your writing calculations, right? Without having to distribute in a new version of a spreadsheet. Is that how you're thinking about it? [00:32:06] S1: Exactly, yeah. Yeah. We've got functionality like we can batch rate, we can run lots of our older policies through a new model. We can do that in a process, in a controlled process. We don't have to start a VBA code from scratch and sort of help us to do that. We can just ultimately press a button and it'll happen, which is pretty transformational for us. [00:32:25] S0: Definitely. And you touched there on something which piqued my curiosity. You talked about pricing strategy and making pricing fun. What is a pricing strategy? Is it like a vision for pricing? Is it... Do you think about it?
00:32:40 · Simon Cooper-Williams
[00:32:40] S1: Yeah, mean, one of our points on the strategy is for it to be fun. I mean, some of the other stuff, we develop pricing to quote. Historically, a lot of people will save up pricing models and they'll do it post bind and they'll use it as a benchmarking tool to use afterwards for a rate change calculation or a rate adequacy calculation. We look to put external data integration wherever we can. We've signed up to a lot of new external data providers over the last year under this pricing function. And I think a big one for us is moving to incremental calibration. I think that's a really big point. If I look at what happened before or can happen traditionally, is a model is looked at and then it's left, and then it's left for a long period of time. And then when it's picked up again, it's usually, Right, let's just start again. Let's rip it up. If you think about... I mean, that's a nightmare for data analysis, because you have to then map all the old columns to the new one. It's probably a new underwriter by then. And you basically start from scratch every time and you forget about the old data if it changes too much, which it often did. But a big one for us is getting to a business as usual calibration. And for me, that's incremental development. I wanna put SLAs around the pricing function. So I wanna say if an underwriter asks for a rating factor to be added, we should have that in within a month, two months. We haven't decided on that yet, but we wanna be able to add things quickly. So you don't have this backlog that is so large value you get to it that you might as well start from scratch and you could add things as you go and you can see the impact and you can do scenarios and you can really say, should we add it? Yes, no, let's experiment. And we can just be a lot more agile with our development like you would in sort of any sort of software development. And that's the vision for us is to make that pricing function working like that. Again, it all comes back to that rate adequacy and pricing being a critical part of the underwriting process. And if it's gonna be like that, that's how it should be.
00:34:37 · Juan de Castro
[00:34:37] S0: I like it. It's something that often, like you called it incremental calibration, right? Yeah. Jamie, for those of your clients who've been using HX longer, is that a reality?
00:34:49 · Jamie Wilson
[00:34:49] S2: It's a great question. It's, I think always, I think for some it's starting to become a reality. I think it's something that I always try and encourage our customers towards, because I guess there's two things there. One, you need a piece of technology that allows you to make changes very easily and seamlessly. And kind of say, a lot of our customers are very used to the mindset of what Simon was talking about. They've moved from a place where pricing tools get changed once every five years. And actually it's not just your pricing entries that need to change culturally for that to work. Your underwriters need to be able to accept that as well. Underwriters have got very used to kind of pricing models and pricing tools being very static. And if you come in and you make kind of like changes suddenly they're like, well, hold on a second. This isn't the pricing tool I'm used to. And so kind of there's a lot of cultural work that needs to go into that, but you start to see it happening. And it's kind of, I guess two things. One, starting to embed processes so that you can improve your pricing model over time. But then I think kind of having a kind of a calibration strategy as well. So like when you first release a pricing tool, thinking about what is the data I wanna be collecting so that in a year's time, I can do some kind of incremental data analysis. Rather than waiting five years and realizing you haven't been capturing the data you needed. And actually that last five years of data capture has been completely pointless. Instead kind of like from day one, think what am I going to do? What kind of incremental like calibration do I want to perform? What data am I gonna need? What level of granularity am I gonna need to add? And starting to put a plan together. Because I think to that point of getting underwriter buy in as well, if historically they had a spreadsheet rater that didn't connect to anything and they didn't trust and it wasn't accurate. And they can now actually be in a world where you've got this kind of like all singing, all dancing pricing model, which connects to things, which makes their life easier. That generates quote documents. There's all of these things for them.