David Grainger is as reinsurance statistician with MS Amlin, an insurance company that is a leader in the property, marine and reinsurance markets. He has worked in the IT and data space for several years and has been at MS Amlin for the past 12.

David joins us on this episode of Fibonacci, the Red Olive data podcast to talk about the challenges around data in insurance in what is a very conservative industry.

We also discuss where the opportunities are for businesses like MS Amlin that want to use analytics and insights to future-proof their technology investments and bring reporting and business value to the entire organisation.

Here are the topics we discuss with their timecodes:

• The challenges that the technology teams have to keep up with the pace of change in the business (2m 53s)
• The journey from disparate systems to a single, enterprise-wide platform (6m 6s)
• Difficulties in delivering a single platform when there are many individual programmes going on (7m 59s)
• Dangers of projects that are delivered in silos, and a ‘business as usual’ approach (8m 49s)
• Becoming a data-led business when you have multiple, business intelligence tools (12m)
• How the hub and spoke model works for the organisation (14m 26s)
• Tying the Wherescape, Infosphere and reporting platforms together so regulators can see evidence that there is good governance of personal data (14m 57s)
• Improving fraud catching capabilities (17m 11s)
• Becoming leaner to be fitter for the future (18m 22s)
• What to do when programmes don’t deliver their aspirations (20m 27s)
• Moving to the cloud, future-proofed by the Wherescape tool, and working with Red Olive (21m 33s)
• The effect of Brexit on the business, and moving data when the data is in multiple systems (26m 51s)
• The skills a data professional will need entering the industry (30m)

Hello and welcome to Fibonacci the Red Olive Data Podcast all about data

and analytics, where we hear from leading specialists and get their take

on the industry. I'm your host, Nicky Rudd. Today, I'm joined by Dave

Grainger, reinsurance statistician at MS Amlin, a leader in the property

and casualty Marine and reinsurance markets. Dave has worked in the IT and

data space for several years and has been at MS Amlin for the

past 12. We talk about the challenges around data

and insurance in what is seen as a very traditional industry,

but also where the opportunities lie for businesses like MS Amlin, who want

to use analytics and insights to Future proof their technology investments

and bring reporting and business value to the entire organization. Let's

find out more. I just wondered if you could tell me a little

bit about how you got into the world of data.

I stumbled into the world of data back in

probably 1990 1991. I started a job in insurance, in the insurance industry

when I was 17 as a accounts account clerk. And as is the

way in insurance businesses, I worked for a very small company that got

bought out by a large motor syndicate back in 1990.

I'd started to be the computer person in the accounts department so I

was playing with the PC that had been installed, obviously all had green

screens in those days. And my boss put me forward to the IT bar, it

wasn't called IT in those days. It was called data processing.

I'm that old, right? So I got put forward to just speak to

the head of that department and my boss, Jerry Burcham at the time

that I interviewed with him, he was a massive cricketer as I was.

So my interview was pretty much about how fast I bowl and what

number did I bat, etcetera. So I got my job through sporting connection

really and really enjoyed working with him. He really sort of shaped my

early part of my career with a lot of young men.

We worked in a computer process department, probably had about 40 young

men. I mean, just to give you an idea of how old we

were. One of my good friends at the time Simon, he was 25

and we named him old chap. So just gives you... Quirky. An idea of

basically 18 to 22 year old boys try to learn how to

develop computer systems. And I had an actual

affliction to reporting and just started building reports for the finance

team and then started to build out stats, databases, etcetera, from there.

And so that was my sort of first introduction

into data from what it can do for businesses.

You've been at MS Amlin for a while and done this huge reporting

and re reporting sort of project. Do you want to tell us a

little bit about that? So I've had a long career at MS Amlin. I've

been there for over a decade so when it comes to data,

I know where most of the dead bodies are. So yeah, there's lots

of challenges with systems so they start off

with a great intention, whether that be an insurance application to store

data or for platforms to build data. And unfortunately,

I think one of the biggest challenges with data is that the technology

teams struggle to keep up with the demand of change that the businesses

want to do. So a new class of business wants to be set

up out by the business to do that properly on the platform,

the IT teams will put a proposal together, spend

weeks prepping that. And obviously a lot of the time we've experienced over

many years is that budgets just aren't there to do

platforms as well as we should. And therefore they go further the quick

win to get the business classes up and running. So what that means

is that they hijack the system. They'll put some business logic around the

system that doesn't exist in the system. So they'll prefix a policy with

a specific code to therefore, they will know what those policies are but

the system doesn't the downstream platforms don't etcetera, etcetera. So

part of the landscape that we've been learning through and evolving is understanding

those things and trying to transition that into the reporting layer layers,

so that everyone gets that single version of the truth. Amlin has as an

organization that's has gone through many big changes over the last 10 years.

It's gone from independent legal entities that are master of their own destinies

reporting into a group PLC business to the group PLC business reshaping

itself, to have more control over those legal entities and bringing them

together to get best of breed across their

insurance entities. And then subsequently at the end of that process,

we are bought by a parent company MSI and they, just because there's no

longer need for the PLC business that got disbanded a couple years ago.

So now we're back to the legal entities being in charge of their own

destiny, again, in that sense. So we are going from a sort of Constantina

of a data world in that we, we bring data together for group

wide reporting. Now we're splitting it apart for legal entity reporting.

I'm sure at some point that they'll come back together again,

right? So we have all of these challenges. And what that means is

that we have lots of despair underwriting systems and operational systems

because people make their own choices about what they wanna do for their

business. And they'll share information, especially when we look at our

re insuring programs where we wanna re insure our inward risks.

So we still need to share that data.

So therefore try to build a reported report data architecture platform

over that landscape is very complex, time consuming

and ultimately very expensive. So they are the things that.

We've learned over the years. I think most data professionals know these

lessons. It's just actually having the time and the team sizes around you

to do the best practices. I would say we're 50% of the way

of our journey to get to that single

enterprise wide platform that can be trusted and used by all, all users

of that data. Be it that individual legal entity or

partnerships between the legal entities to share data. So we've made some

good progress over the last two, three years, especially with, with the

Red Olive business, helping us on some key projects. But yeah,

we've still got quite some way to go. Obviously it's massive and it

is changing, but where do you typically start with the project? And do

you have your own way of thinking about it? Like you say,

you've learned over the years and sort of honed down from mistakes made.

And I always think that it's, that's that difficult thing of actually having

the end goal. what do you actually want it? And finding out from

all the different stakeholders within the business, how they would use the

kind of information that the data shows, but do you have a process

that you've now adapted so that it works, or, you know, what's your

approach? I think we're, we are evolving our processes. So again,

because of the disbarment of the PLC business in the last two years,

I was part of a, what was the business MI function,

which was, you know, was a PLC function. So it was a

group wide reporting function. So with the disbarment of that PLC business,

we've now evolved into essentially having data functions within each legal

entity. And so we are all learning. So parts of that team got

federated out into the separate legal entity teams. We're all still evolving

out ways of working. But I think the

challenge you have with data programs is, you set out your vision around

what does the end goal look like? And I think with individual programs,

that's very difficult to do. I think the lessons that we've

learned, I think from the recent programs we've run and we are, we're

starting some more at the moment is that individual programs contribute

to your data architecture landscape and you may have a, an ultimate vision.

So yeah, from my perspective, I'm responsible for the business intelligence

tooling and the reporting platforms that support those across the organization.

So I only want one business intelligence tool to only want one reporting

platform because therefore that's cheaper, easier to maintain. It could

be focused on supporting those platforms much easier. At the moment we've

got multiple business intelligence tools, we've got multiple platforms,

projects, and programs deal with those individual things. So they're gonna

contribute to our ultimate vision of having single platforms that can be

used easily, etcetera. So I think starting up

projects and all programs, we're starting to really drive out data piece

as early as we can into pits bits, etcetera. So we get programs

and projects talking to the right people. I think that's the key thing.

So typically you having big projects and programs is that we need support,

we need support from third parties. So we'll, we'll bring in a partner like

Red Olive that will have the right skillset to help us build our

platform, but that not necessarily have the understanding of the data landscape

Amlin has. So that's a very difficult thing to pick up first and

foremost. So if are not careful, what tends to happen is that projects

and programs tend to work in silos, 'cause they're delivering their specific

part of the jigsaw puzzle. And without knowing the Amlin data infrastructure

and how things work necessarily, you can look at data structures in our

database and see exactly how things work, but actually knowing the nuances

of how that data's actually being used is a real challenge.

So from my perspective, the great thing that I've got from working with

Red Olive, especially their lead consultants, Mark Fulgoni, David Searro

is, our engagement model with the stakeholders across our organization.

What does that look like? Because we are a business as usual team.

Too often, we just get you know how to do stuff.

Can you just do this for us? And then we'll go off and

do things very quickly and that doesn't ultimately build a proper relationship

out with our, with our stakeholders. So we're, we're now really evolving

towards detailed engagement requests. So they understand the full implications

to what they're asking, because just add that thing, that table,

that field to the platform will be fine. Isn't necessarily the right thing

to do. So, yeah, we've got lots of learning to do,

I think, as ever, we never have enough people to do that.

So that's another, it's another challenge for us is

how we build up our internal teams again, to support the demand and

be able to really articulate how that benefits the business. Obviously within

the insurance base, it's seen as very much a sort of, I would say,

traditional business in the way that actually, you know, sort of data flows

and yeah. How an organization is sort of set if you like,

how have you managed to sort of pull everyone together into thinking differently

in this kind of new way of working? Is it challenging them upfront

about get what they want the data to tell you? Is it a

communication to sort of, a bit of education? We're very much at the start

of our data maturity level. So the insurance industry is typically slow

moving toward that change. So we have a lot going on,

but it takes time to do that. The RFS 17 accountancy regulations that

are coming in that's been a big, slow moving beast, thrust to move

towards. Our legal entities, have aspirations of being data driven companies.

They're very aware of the disruptors in the marketplace and they really

want to evolve. There's some really good initiatives going across the organization

around robotics and data science. Our data science team are doing some good

stuff with specific parts of the business in our London digital hub.

So there's definitely the right steps forward. I think the challenge that

you have is that you have to have the platform in place so

that you can support the business so they can take that journey.

So too much time is being spent to actually just deliver the core

foundations I was talking about around, we've got multiple business intelligence

tools that some people still have to rely upon because not all the

data's available in other reporting platforms and other business intelligence

tool. So we need to get that structure in place and foundations in

place for them so that we can accelerate that data led culture.

Right. Which is what everyone is crying out for in our organization.

How do you do that in such a sort of large organization,

so that whole digitization and affecting the industry, as well as kind of,

other regulations and compliance, sort of guidelines and things like that.

Do you break it down into steps or have you got a kind of, we've

got a 3 year plan or? I think hope that's a really good

challenge because obviously we've got the legal entities having their own

business change initiatives. And I think that's where

the, criticality is for us right now as an organization is that we

need to ensure that we've got the right people

plugged into all of those projects, because some of those we completely

driven and led by the business and it's making sure that we've got

the right people with touch points on there, so that they can steer that

in the right. So we make sure that we get the best benefit

of those initiatives into our data architecture platform.

I think it's quite easy for the business teams to just drive forward.

They see these things, they want these things,

we've all got handheld devices that have got

so much information at our fingertips. So why can't we have our insurance

data at our fingertips, right? So that's not a too difficult service. It's

actually trying, the cell is actually doing it in a constructive manner

so that everyone understands where that single source of the truth is.

So as far as how we are doing at MS Amlin, we've got,

as I said, local data functions in each entity. So we've now got

a hub and spoke model around the data organization. So

we've got these data functions that are responsible for an interesting point.

You said around the regulatory demands, that is becoming more and more stringent

and will continue to do so. Just seeing how the banking sector's gone

over the last decade, I'm sure insurance sector will follow there them as

well. So it's really critical. I think for us to be able to show

that we've got control over our data. So there's been some really strong

project initiatives over the last 18 months around data lineage, data capture,

data management, and they have been at the forefront of

how we tie everything together. So how we are doing that as an

organization? I think we've had some really great support from Red Olive

to show us how to do that in a better way.

So as you'll know, the west cape CAP platform that Red Olive are very

skilled in. We've used that to help us support our, data and image. That's

given us a good starting point out of the box. We've got,

really great data capturing, info fit all and IBM tool, which has got

our business glossary and data glossary in there. And what we've been able

to do is using the skills of Red Olive is be able to

tie that platform up with our reporting world. And we're really starting

to evolve our reporting platforms from a regulatory perspective to actually

showcase that we do have good governance over our data

so that when we do, returns to the regulators will be in a

position. So when they come and say, "Right, show us your data controls"?

We'll be able to do it in the future as a push of a

button. And it'll just be part of our processes. There's obviously like

the current what's going on, but there's also this sort of,

like you say, the future of when you're gonna be asked stuff. And

also you've got the kind of, sort of big trends in technology that

are sort of pushing with the emergence of machine learning and,

sort of applying data to solve problems. What do you see as some

of the big areas of opportunity to apply machine learning in the insurance

sector? Is that something that you are currently looking at as you mentioned

earlier about some data science team doing some interesting projects. Is

there something you could share on that? I certainly think that's definitely

on our radar. I'm not so sure specifically about machine learning at the

moment. I know there's been some really good used cases in the claims

space, certainly in the, in the insurance sector.

We are obviously always looking at evolving our processes.

One thing that I know that we are starting to look at is around

our fraud catching capabilities. So I think we're looking at seeing how

potentially advanced analytics can help us identify fraudulent claims quicker.

I think that's probably gonna be the best space in the immediate term

for us to, look at that. But yeah, I think the data science team

have been doing some good work around looking at our losses and our

risks that we right. So they've been looking at how we can protect

ourselves better with better information. So we now have a modeling tool

on our Marine business that checks where the vessels go, that data's all

available the maritime data. So we've plugged that into our risk models

and they've done some really great work with the Marine underwriters to

capture more premium is probably the best way to describe that.

Where we, where we're finding out that vessels are visiting excluded territories

so they don't go into, but they do. Right. So that's been some

really insightful stuff that the data science team have done.

I think from our machine learning perspective, I think once we've got our

platform more organized and consolidated, I think that then allows us to

better make use of that type of technology.

Where are the biggest challenges and opportunities with the data and extracting

it, because I'm guessing that picture's building all the time with the amount

of data sets that are coming in and the more information you're collecting.

Yes. I think that's a really interesting point because obviously the businesses

are going through a lot of change at the moment, either through,

performance re evaluation. We've had some tough years as a, as an

insurer and reinsurer, certainly in the last five years, it's been a very

difficult marketplace for us. So there's been a lot of consolidation going

on. So we are I suppose getting ourselves leaner is probably the best

way to be fit for the future. And I think we've got these

big programs like RFS 17 that are preparing us for the future.

So we'll be looking at underwriting our books of business in a different

way. Once that's settled down, going into I think next year is the

first year that we'll start reporting on RFS 17 basis

that will then allow us to look at opportunities. And I think the

key thing is the growth that the entities will be looking at.

So to do that, I know we've got

as an example we've got some online platforms coming online this year and

plugging those into our backend architecture is gonna be a key piece,

certainly from our financial and reporting perspective. Once you've got

that in place, then our analytics can then start to go to work there.

So there are lots of challenges. I think that, is ensuring that our platforms

can keep up to pace with what the business wants. The data function

and the IT teams will always be behind the business decisions because some

of 'em will be hugely confidential. IT teams won't even know

that there may be a small merger on the card or an acquisition

that's happening. But I think the platforms we've got in place at the

moment do allow us to onboard data as quickly as we can.

So there's lots of interesting scope, I think that's coming down the Pipeline

next year. Would you say that, just being able to be a bit

more agile with those platforms is kind of opening up opportunities or maybe

managing frustrations where in the past perhaps the IT department, would've

specced out a system that even if it wasn't fit for purpose after

five years, for example, you still had to get the ROI from that

investment? Do you think that sort of changed in the last few years?

I think that's a really good question. I think obviously there's always

challenge where things don't deliver the whole scope. You consistently see

big aspirations with programs and this is why I was saying to my earlier

point that programs shouldn't be drivers of aspirations. The aspirations

should be there and programs and projects should contribute to that roadmap.

And I think too many times programs try to do those things by

themselves. And I think the key thing is having the tooling available to

be more agile around data ingestion, et cetera, but ensuring that it's done

in a strategic way. So one of the reasons why we'd like the

WhereScape platform, the tooling when we selected it a couple years ago

is the fact that it's essentially technology agnostic in the... We've got

aspirations to move to the Cloud, we are a SQL server shop at

the moment. And I think there's definite aspirations for us to get to

the Cloud using the WhereScape tooling. It means that my team only have

to learn one tool and we can then deploy.

It's not gonna be seamless and it won't get everything right in one go,

but certainly the bulk of the platform will be created by the tooling. And

then we just be left with the tricky 20% just to get right.

So I think that future proofs our data platforms that we can move

with the times. And if someone decides that the Cloud is the best

way to go for everything that we do, then we don't need to

go and reinvent the wheel in the Cloud. We can get there certainly much

quicker. And I think the Cloud is certainly something that enables us to

be more agile in the future. We can stand up proof of concepts

much quicker going forward. Obviously you mentioned about Red Olive and

working with them and WhereScape being kind of a key tool for you. You've

obviously been working with them for a couple of years, I think.

Was it useful having external consultants coming in and challenging the

way that you were thinking? I'm not so sure they challenged the way

we are thinking. So my experience of working with Red Olive was that actually

we were quite aligned in what we are thinking. It was actually the

execution. The platforms that we've built so far with the WhereScape technology

used to support our financial reporting. That is

a guy heavily normalized database at the heart of our

data capture as you'd expect, it needs to be very strong 'cause it's

dealing with our financial transactions, it supports our general ledger,

it supports actuary reserving process. So that platform itself is great

at what it does. And obviously it being highly normalized means that it's

not great for reporting and analytics. So we needed to build out reporting

marks that would support easier reporting. When we engaged with Red Olive,

initially it wasn't a case of showing us new things other than actually

look at this tool that we can deliver traditional reporting marks we were

used to working with via this tool and we did a proof of concept

that was about a week long. The biggest challenge of that week was

actually getting people logged in into our platform because we were in lockdown

during COVID. So that was a real challenge. Just being able to get

someone to log into our network, to connect to some of our data,

but I had a very light touch on the POC, but

Mark was able to demonstrate how quickly the tool could help us transition

highly normalized data structures into more typical star schema, data dimensional

modeling structures. Very quickly, they built in documentation of the WhereScape

platform and the data used in tracking made it very, very appealing as

a tool to move forward. The tool itself, when I was looking at

it, I could probably write the code and my team could write the

code it generates just as quickly but the key thing was really around

the documentation. The data linears tracking, ease of use of the tool was

quite a key thing as well. So it was definitely a put your

ego aside of, well we could write this code anyway. But that tool can't

write code as well as I can but... And it's not magic.

It doesn't do everything you'd wanna do it out of the box,

there's always a nuance with tooling, but for me, I was really sold

on the ease of use of the tool and how quickly Mark was

able to demonstrate that. Now, obviously Mark's very experienced WhereScape

consultant. So not everyone's a Mark Fulgoni. But for me, it was a

key selling point of that platform and how we can make use of

it going forward. How quickly could you see the value of implementing the

system? Or is it kind of a very quick...

That you were running before we working it that fast?

It was literally at the end of the POC terms a week.

I was like, I'm sold on the benefits the tool bring.

The project it was about to embark on was a challenging one anyway.

Again, to come back to projects and stuff being run in isolation with

other things that are going on is always a challenging space.

So the platform that we built together took longer to do than was

first thought because I knew it was a very complex project,

but other people hadn't really grasped the fact that it was a challenging

space we were working in. But I think the tooling is great.

It's making sure you understand the pitfalls of where you're starting what

are your data sources, et cetera, how complex are they?

Does it have all the data in the platform that you're sourcing to

build the analytics that you need to. Now One of our challenges was,

as I said before, we've got data in many different reporting platforms and

data sources and business intelligence tools, so not everything is in one

place, and that's the journey that we're still on.

On a completely different note, London's Financial Services Center was a

big point of discussion in the run up to Brexit, I kind of mentioned it

with all the political shenanigans going on at the moment. As someone working

for a London insurer, do you think Brexit has made much difference to

your organization? And if so, what? It certainly cost us a lot of

money as a first point. So as part of the Brexit deal,

we actually had to transition all of the risks that we've written in

Europe to Lloyds Brussels. Lloyds of London set up a Lloyd's Brussels office.

Now, I don't know all of the insurance side of that,

but by law, we had to transition all of our European list from

Lloyds of London to Lloyds Brussels. So yeah, we had a dedicated part

seven projects as part seven is actually the Lloyds contract, I believe,

that did the transition. That's been a really challenging project. Again,

we come back to our data is not in one place,

so therefore, how do we get all of that data? So

very briefly, most of our European business that we broke was through a

property and casualty business in the UK that used

four systems for its underwriting. So getting all of that data out of

those systems in one place in a format that Lloyds platforms could consume,

it has been a very challenging space, and it continues to be today.

So I think from a actual insurance perspective, from a business level, it's

probably not been too traumatic, but from a technology perspective, it's

a real challenging thing to do. And obviously, it was all new, right? It

was new for Lloyd's to understand what was needed, so we had very

evolving timelines and requirements. So requirements would change in the

Lloyds marketplaces. Obviously, it's a great place to do business, but

all insurers and all brokers do things slightly different. Again, there's

not one single Lloyd's market platform that everyone uses, insurers and

brokers have their own. So trying to find that common ground around how

insurers write their business has been a challenge for the Lloyd's platforms

to receive. But I think the feedback I've had from certainly the project

team that was working on it, the collaboration between the Lloyd's syndicates

was really strong on this initiative because no one's ever done it before

and they had tight deadlines. So it's January 21 was the drop down

deadline date that this was happening, and we only had

effectively a year to do it. So it's been a real challenging

space, but so far, it's been pretty well done. So apart from costing

a lot of money, I'm not sure what the benefits of Brexit were for

the insurance industry. I was gonna say it's still not mentioned within

the office, the V word. No. Okay, well, we started off with how

you got into the world of data. I just wanted to ask if

somebody is thinking about coming in, finishing uni, or they've decided

that actually data is what they want to be doing. What advice would

you give a younger data professional starting out? What skills do you think

they should be kind of honing, and what kind of things do you

think they should be looking to do? It's a really good question.

We've had some graduates that come from another third party that have onboarded

into Amlin in quite recently. So I've spent the last couple of years,

they've been supporting us on the reporting project as well. And I've been

talking to them quite a bit actually around

what do you like doing? What are the things that you wanna do?

And I think that it's a really interesting space, data.

The first thing is, do you love it? Because if you don't love

it, then it's a difficult, boring space to be in. But I love

data. I've just grown up with maths in my head as a small

boy, so I played, as I mentioned earlier about my first sort of data

or IT job interview. I love cricket and football, so sports.

There's tons of data in sports, especially cricket. Coming out of uni, one

of the things to go for, so I think there's

the cool fun stuff to look at So you'll hear machine learning,

you'll hear python, you'll hear... But actually just understanding data

structures and understanding the business domains that you're working in,

that goes a long way. So the technology side of data is one

thing, but understanding the business domain you're in, I think is the key

differentiator between being just a person who builds and does reporting

as an example or someone who can actually make a difference and take

a lead in the organization you're working is really understanding the domain

knowledge so that you can take your skill sets you've got from your university

courses and stuff, and you'll be able to run with it. That's my

main advice, get to know that domain knowledge, the business teams understand

themselves. I think that's a really valuable piece of knowledge because

technology changes and evolves the whole time, but actually if you can use

that as a tool to actually see the bigger picture. Yeah,

no, I think that's the key thing. You can learn a new technology

skill. Even I, as an old man have learned some new stuff in

the last few years. WhereScape Red, if you've got lots of technology skills,

you're able to just fit technology and understand it to the domains you're

in. So you really need the domain knowledge to help you do that,

in my humble opinion. Some interesting points there about setting some objectives

for your data projects at the outset and using tools to help,

not hinder. Join us for the next episode of Fibonacci, the Red Olive Data

Podcast, where we'll be joined by another data expert sharing their thoughts

on the latest trends in AI and big data, along with some great

hints and tips. Make sure you subscribe to Fibonacci, the Red Olive Data

Podcast from wherever you get your podcast to make sure you don't miss

it. That's all for today. Thanks for listening. I've been your host,