Making Risk Flow | The Future of Insurance
00:20:45 · Jamie Wilson
[00:20:45] S2: of the pricing actuarial team, like we had to kind of push out rate change analysis and reporting from all of the data that was captured in the pricing tools at the time. And the way that that worked was each actuary wrote some very hacky code to pull out data for their pricing tool. And they ran that maybe once every few months. They realized the code was wrong. They would try and correct it. They would download that into an Excel spreadsheet and then they'd send that Excel spreadsheet around to kind of all the underwriters in the company via email. And so kind of what it meant was that we had this like horrifically manual process that took, I don't know, probably three weeks of actuarial, the entire pricing actuarial team once a quarter to produce rate change reporting. And then after about five years of doing this, we suddenly were like actually kind of this is kind of at the beginning of the development of the global predictive analytics team I built out. We kind of suddenly realized we can actually take some of the kind of standard things that you would expect from a data engineering structure or like a proper BI kind of work stream. And we kind of effectively brought in data engineering skills, took that incredibly manual process and said, well, let's just build a pricing data asset. Like automate all of the piping of this data. Kind of was running every single day, kind of automate the development of like a web based application report. And so suddenly we kind of moved from this place where actuaries were doing hacky data work and pushing out reports in Excel to suddenly every single person in the entire company had live access to all of the pricing data they could possibly want. So they could go and do self serve data collection. But then we also had like these dashboards, like rate change dashboards that were suddenly live on a daily basis as well. So kind of you had underwriters looking on one screen, they would have their pricing tool. On the other screen, they would have this dashboard, which would show what rate change have you achieved in the year to date, or what price adequacy are you running out? Or like, what's your portfolio look like? And it wasn't... You kind of moved away from this, a whole team of actuaries doing this manual...
00:22:44 · Jamie Wilson
Massive manual effort to push out a bunch of Excel reports to suddenly kind of this thing being live all the time. And I think that's what we're starting to see more and more of, is kind of actuaries and kind of... And to be honest, where actuaries don't do this, kind of non actuaries, basically realizing something has to change and doing it instead. But I think kind of effectively actuaries realizing there is a much better way, like kind of more advanced technology exists. There's more advanced reporting. There's more advanced kind of data kind of tooling and coding at the heart of a lot of that as well. Like kind of actually getting upskilled in coding in order to kind of like push out these more advanced solutions.
00:23:22 · Simon Cooper-Williams, Juan de Castro
[00:23:22] S1: In terms of like engagement with that system as well from that front end underwriting, there must have been such a big amount of improvement. I mean, like a similar example with us on like how we would develop. You'd spend a lot of time developing a spreadsheet with an underwriter, and they have to give a lot of their time for that. So it's on top of their day job is iterating with you, developing it. And like, we've observed people have been filling out these spreadsheet raters for years. They put a lot of value in, but then the effort they put in is equal to the value that they get out, and that's where you lose them. So they do this and then dashboards. They don't get data. They don't get analysis that uses the data they put in because it's not accessible. And that's just one of the biggest sort of limitations around that legacy spreadsheet system of a pricing system, which a lot doesn't work. [00:24:13] S0: I mean, there so many issues with that, right? I mean, I've seen the first thing is how do you make in the previous role, right? How do you make sure everybody's using the latest version of the spreadsheet? Even something as that, right? Because if you're distributing the spreadsheets via email, that's probably not...
00:24:13 · Juan de Castro, Jamie Wilson
[00:24:13] S0: I mean, there so many issues with that, right? I mean, I've seen the first thing is how do you make in the previous role, right? How do you make sure everybody's using the latest version of the spreadsheet? Even something as that, right? Because if you're distributing the spreadsheets via email, that's probably not... [00:24:31] S2: Yeah. No, like, how do you make sure that you've got everyone on your email distribution list? How do you kind of, if someone corrects the spreadsheet, like you say, how do you make sure everyone's got the most recent version of the truth? And it just becomes this nightmare of trying to manage a tangle of other things, as opposed to having like a true production system or solution in place where everyone knows where they're working from. [00:24:54] S0: Exactly. And then the other challenge we've seen is then, as you said, Simon, you've got all these underwriters filling all these spreadsheets, the inputs to those spreadsheets get lost. That's not data that's stored anywhere. It's stored in a file system, but it's not discoverable, right?
00:24:54 · Juan de Castro, Simon Cooper-Williams
[00:24:54] S0: Exactly. And then the other challenge we've seen is then, as you said, Simon, you've got all these underwriters filling all these spreadsheets, the inputs to those spreadsheets get lost. That's not data that's stored anywhere. It's stored in a file system, but it's not discoverable, right? [00:25:12] S1: Yeah, completely. Yeah, I'm very happy to go hard on spreadsheets. Are completely trying to move off them, so happy to dis them as we go. But yeah, mean, I think another theme with them is that the user interface and just that UI functionality of spreadsheets, we just lost people on that. I think you'd see whoever happened to develop it would have a different style. And if someone wanted to pick it up, they would completely make it their own again. It just, from a usability perspective, it just didn't feel like it was part of the company's infrastructure. It wasn't part of that underwriting workflow at all. You just completely lose people given that. They're also incredibly hard to pick apart as well. Mean, we've all been there where someone joins, someone's developed a previous model, you have to start and try and understand that. And if you want to add something as simple as adding a rating factor or even integrating external data. That's what a lot of people talk about, putting an API functionality in there and automating that data to come in. I mean, these things were just incredibly painful and pretty much impossible doing it in those sort of systems.
00:26:21 · Juan de Castro
[00:26:21] S0: Okay. So perhaps this is a good point to ask you, Jamie. Most of the questions so far have been about your history as an actuary in different insurers. A lot of what you're doing now within Hyperexponential obviously is helping insurers move away from spreadsheets and having a much more nimble way of pricing. So let's start with what does it look like? So like what does it mean? We've already complained a lot about the history of spreadsheets. What does it look like with a more modern rating engine?