Daily Mail & General Trust plc ( DMGT Investor … › ~ › media › Files › D › DMGT › Investor...1 Daily Mail & General Trust plc (‘DMGT’) Investor Briefing 2 July 2019

1

Daily Mail & General Trust plc (‘DMGT’)

Investor Briefing

2 July 2019

Transcript

Disclaimer

This transcript is derived from a recording of the event. Every possible effort has been

made to transcribe accurately. However, neither DMGT nor BRR Media Limited shall be

liable for any inaccuracies, errors, or omissions.

2

KEY

PZ: Paul Zwillenberg

KW: Karen White

V: Video

MK: Moe Khosravy

CB: Cihan Biyikoglu

MB: Mohsen Rahnama

RM-W: Robert Muir-Wood

MW: Matthew Walker

AM: Alex Mees

KT: Katherine Tait

PW: Patrick Wellington

AW: Adam Webster

AMa: Annick Maas

CC: Chris Collett

ND: Nick Dempsey

PZ: Good afternoon everyone. It's my great pleasure to welcome you here

this afternoon to DMGT's Investor Briefing where we're focusing on RMS.

It's fitting that we're holding this event here at number 10 Trinity Square.

This building was originally completed in the 1920s when London was

still at the centre of all trade. It was built to house the Port of London

Authority. Every day more than 1,200 people would come through its

doors to pay their port rates. It was then rebuilt after being bombed and

destroyed in the blitz, and to honour its history, the Rotunda Room

downstairs, which many of you walked in through, was modelled on a

nautical compass with the pillars decorated to look like ships’ ropes.

After the war, this room was the very first meeting of the General

Assembly of the United Nations in 1946 when the good and the great met

to discuss how countries could ensure peace, prosperity, and trade to

build a brighter future. Later, this building became the European

headquarters of the insurance broker Willis Faber Limited. It's fitting,

because a place steeped in the values of trade on one hand, and

managing risk on the other, is the ideal backdrop for a business like

RMS.

But before I talk about RMS, let me take a moment to put it in the DMGT

context. As you know, when I became CEO, I set out three strategic

priorities: performance improvement, portfolio focus, and financial

3

flexibility, and I'm pleased to say that as a result of making great progress

on all three, we've moved into the second phase of our transformation.

We are now focused on growth, more specifically good, profitable

growth, and improving cash flow. Continuing to realise the full potential

of each business in our portfolio, all of which are market leaders in their

sectors, all anchored in must have proprietary content, data, algorithms,

stories, and all providing the products and services that their customers

rely upon each and every day. RMS plays an important role in DMGT's

portfolio as a significant driver of future growth.

Before I get to RMS though a word about what the Group looks like now.

We've done a lot of heavy lifting, and we've made some very difficult

decisions. As a result, we are now much leaner, fitter and future focused.

Our increased portfolio focus means that we are now in six sectors, not

the 10 when I arrived. We have 11 operating companies, not the 40, and

our net debt to EBITDA ratio is now 0.3, down from 1.8 when I arrived,

and that is after returning nearly 40% of our market cap to shareholders.

We've been busy, and that means with this focus that I get to spend more

of my time on the DMGT of tomorrow ... and I probably should be going

through the slides as I do this.

Turning up in the right places is half the battle. So we've deliberately

shaped the portfolio, focusing on the markets with lots of potential, with

long runways that play to our strengths where we have all number one

positions. Part and parcel of the next phase of our transformation is

looking at where the future value lies for our shareholders, where it will

come from. We are very, very good at building and nurturing strong

brands grounded in proprietary content, be that stories, be that science,

be that software. We are very good at taking okay businesses

underpinned by great brands, and fixing them to unlock real value, and

we are very good at incubating new businesses, taking advantage of

technological change to create value for consumers, customers and

shareholders alike.

So how do we translate this into how we think about our businesses?

Well, I look at our businesses in three buckets, in terms of their roles,

expected contribution, and how we allocate capital. The first bucket at

the top are our ‘Predictable performers’. They're amongst our largest

businesses, mature and defined by their predictability. They might be

growing just ahead of inflation, or they might be declining, but they're

strong brands in great demand and essential for the customers they

serve. Importantly, they are our economic bedrock, providing the cash

flow that underpins our dividend and investment elsewhere in the

4

portfolio. We will, of course, continue to invest in these businesses in a

disciplined way organically and inorganically to maintain their

leadership positions and develop new avenues of growth when the right

opportunity comes along.

At the bottom, you see the ‘Businesses for the future’, early bets at the

beginning of their journey, often in areas where technological change

brings opportunity. These businesses may succeed, they may not, but

they're small enough for us to be able to absorb the risk, and if we get it

right, the prize is big. We have form in this area, Zoopla, MailOnline,

Hobsons’ Naviance just to name three. Finally, there's the central bucket,

are ‘Growing and delivering’ businesses, and this is where RMS is. These

are all businesses that are well positioned in attractive markets with long

runways that play to our strengths, and this is where you will see the

most growth over the coming years in both the top and bottom lines.

Of course these businesses, as with all DMGT companies, we invest for

the long term with a bias towards growth over margin and we will deliver

that growth organically and through bolt-ons. Over time, you can expect

to see more underlying growth and Cash OI, both of which can be better

from these businesses. This is our absolute focus for the next phase of

our transformation. These businesses, dmg events, Genscape, Hobsons,

MailOnline and RMS represent the economic engine of the future of

DMGT.

As all of you in the room know all of our investment decisions are driven

by a long term perspective. We invest for long-term shareholder value.

We now have and are committed to maintaining the financial flexibility

that we need to invest in both organic growth and bolt-on acquisitions

and we will continue to prioritise those opportunities with the greatest

value creation potential with a laser focus on return on investment

across the portfolio.

Let me now turn specifically to RMS, a significant driver in the future of

value creation for DMGT. As you all know, it has not been plain sailing

for RMS since 2014 with enormous frustration for our customers, and for

you, our shareholders. The impairments we had to make as a result of

RMS(one) have been a disappointment to everyone, which is why I'm so

pleased that you are all here today so that you can see for yourselves the

remarkable progress we've made in the last 12 plus months.

Now, we're by no means there yet, but we now have an exceptional

leadership team who combined the very best science with the very best

5

in enterprise, in SaaS, in software, in AI and ML, which is why I knew it

was the right thing to do, and I saw that recently at RMS's customer

conference Exceedance, where the excitement among its customers,

among the people who use RMS each and every day was palpable. Now,

I'm going to leave it to the team to talk about the business and the

changes that they have made, and their ambitious and exciting plans for

the future, because no one else can tell it as well as they can, but before I

hand it over to them a word on people.

A crucial part of my role as CEO is putting the right people in place in

each of our businesses to realise the full potential of each of our

businesses. People are our foundation. From early on in my time at

DMGT, I realised we needed a different team at RMS. As I got to know the

business better, I could see all these exciting opportunities, and how the

business was holding itself back.

So, I brought on board Karen White, whom I've known for five years

before her appointment. Karen and I spent so much time together

between California, New York and London, and I knew that she could

build the team that would take RMS to great places. She had the

background and experience in software, in data and information. She

had a track record, not only of breakout growth, but of solving really,

really hard problems, and RMS was a really, really hard problem.

Not only that, but she's built an amazing team around her. A team that is

the envy of Silicon Valley. Together they have systematically

strengthened the core. They've tackled the hurdles that got in RMS's way,

and they started unlocking exciting new adjacent markets. Speaking

personally, what Karen and the team have achieved over the past 12 plus

months has cemented my confidence not only in them but in what this

business can achieve, which is a lot.

We live in an uncertain and volatile world, which with an increasing

need to quantify risk quickly and accurately. This represents a huge and

exciting opportunity for us, which Karen and her team are best placed to

talk about. So, I'm going to hand over to them. Karen is going to walk us

through the business, the market, her strategy and some of the

accomplishments we've made in such a short period of time. We'll then

take a short break and her team will come up and go into some more

detail about the software, show you some demos of our new products,

newest products, which are out in the market today, and our models and

our model science. Then, we'll take another short break, come back for

6

some Q&A, and wrap it up with cocktails outside. So, thank you very

much indeed, and over to you Karen.

KW: Thank you. Thank you, good afternoon. I appreciate you being here in

such force. I was just told 10 minutes ago, this is the first time DMGT has

had a single one of its portfolio companies come front and centre, and I

appreciate all the interest here. I have to say as I was getting ready to

leave California on Friday as the market closed, I got ... I'm signed up for

the news flashes of DMGT and RMS of course, and I get a news flash

from Yahoo Business, and perhaps some of you saw it and did the same

double take that I did, but these were the numbers before us.

Suddenly DMGT's stock at closing had climbed from $780 per share to

$1.4 billion per share, and all I could think of was the look on Warren

Buffett, and Jeff Bezos, and Bill Gates’s faces as they realise that each and

every one of the shareholders in this room had now overtaken them as

the richest men alive. Well done you. So it was, it was kind of funny. I

appreciate the great confidence and enthusiasm, but I think it's safe to

say you've really overshot it here, and I think your expectations coming

into the room are a bit outsized, if this is what they are.

No, quite seriously. Thank you for being here today. I'm first going to

start by taking a step back and looking at RMS historically as a business

before we step into our future. I've been here a little over a year now.

RMS is celebrating its 30th anniversary this year, so we had a lot of fun

with that at our customer conference a few months ago. RMS helped to

pioneer the industry that it leads today starting with a natural

catastrophe and then of an earthquake in California 30 years ago, and

building upon that base to create its leadership in the market. It leads

that market today, and when I came on board, you might be curious,

given my background in software, which is rooted even underneath of

that in data, why I ended up joining this company in natural catastrophe

modelling.

I've made my career in innovation and I've been pretty lucky at picking

some great companies to join, but what I look for in anything I want to

spend my time with, is that point in time where an industry is ripe for

transformation. Whether it is ready to accelerate, or whether it's coming

kicking and screaming, like some of the more stodgy industries that are

regulated that we've seen undergoing transformation, but

transformations are only aspirational unless the technology that can be

leveraged into that transformation has matured to a place that it

becomes very interesting.

7

You can use those technology advances in ways that you couldn't have

even thought of five years ago, and 10 years ago, and that's getting the

timing right. Over my years as an investor, I've gotten the timing wrong

when I have been a little early, or when I have been a little late. Being a

little late is better, by the way. Being a little early turns out to be very

expensive, knowing your ... Everyone is learning on your nickel.

But I think the timing here is just right, and when I came into RMS, as I

think about my career, when I joined Oracle, the company had lost $60

million the year before I joined. It was not an obvious choice to come on

board then, but the industry was changing. At the time, there were 50

web servers in the world. 50 web servers in the world, and we were going

from client server computing to internet computing, or end tier

computing, and all of that wasn't obvious to us then, but we saw a change

in the wind and a transformation coming.

I met Larry Ellison actually on a plane, and we argued about technology

for about five hours. By the way, I was right. We were talking about

network technology, and what was going to happen next. I got very

intrigued by this vision of this transformation, and I really believe that

the technology was there that was going to fuel it, and in my time there,

not only that transformation of the entire computer industry occurred,

but a second transformation in our business occurred, and that is going

from being a database platform and taking that power that we had as a

database platform and transforming ourselves into a strategic partner for

our customers where we were in their mission critical business with

applications.

So, we rode two transformations then when those two vectors came

together. My last company was Addepar, a financial technology

company, or fintech, and people laughed at us because we were talking

about redoing a 30 year old industry about investment management with

well entrenched players and you can't solve that problem. Well, we did

solve that problem, and we now have $1.3 trillion on the platform

everyone said wasn't worth building.

I love to be at a time of great industry transformation where technology

can prove conventional wisdom wrong. That's what I feel like here at

RMS. I've sat on 12 boards of directors in my career, and more and more

so the meetings became about risk, risk, risk and risk. I look at this

business and at our core, at our core we are in the business of risk, and

we have focused that understanding of risk on a very interesting area,

natural catastrophes, and we've built models. I think that is a marvellous

8

beachhead from which to approach a broader market, and I'll talk about

that broader market here this afternoon with our team.

A business is only worth joining if you have some moat, some

differentiation, and there was a philosophy that Ellison believed in at

Oracle that I have adopted throughout my career. It is simply this, is that

true differentiation occurs when you have something of great value to

the market, that only you can say out loud without lying, and if your

competitors say they are lying, and that is what I found at RMS. There are

really three things, three moats that we have that I really like as the hand

of cards we're playing as we approach the emerging risks market.

One of those is simply the people. We are in the business of intellectual

property, and all that matters is the team of innovators and inventors

who think out of the box and who know how to work together. We have

nearly 300 scientists at RMS working together who know how to do

things that the rest of the world simply doesn't know how to do in the

way that we know how to do them. If I had to compete with RMS, and

replicate the team or try to get them, I wouldn't like that hand of cards. I

like this one. So the people in the team are quite differentiated,

particularly in the area of the sciences that we built out over the years.

The second is our model portfolio, which has been amassed over a

period of 30 years, and underneath of those models is data that we have

yet to unlock into the market appropriately, and we will now, that no one

else in the world has. So, anyone else who claims these in the way that

we have them is lying. So again, I like this hand that we're playing.

V: 1989 the world wide web was invented, the Berlin wall fell, and Chicago

was dominating everyone's boombox. 1989 was also an important year

for the insurance industry. In March, Exxon Valdez ran aground in

Alaska. In September, Hurricane Hugo devastated the Caribbean and

south eastern United States. In October, the Loma Prieta earthquake hit

the San Francisco area, and RMS was founded. Over the last 30 years, we

have seen so many catastrophes and walked through the journey with

you.

A large number of industrial facilities were damaged. The terrorist

attacks on September 11th, 2001 resulted in massive damages.

Hurricane Sandy battered the east coast from the Caribbean to Canada.

2017, 2018 California wildfire seasons were exceptionally destructive.

9

How can we prepare together for the next 30 years of risk?

KW: It's always good to understand your heritage before you start to look at

going forward, and I thought it'd be useful to ground us in the kinds of

natural catastrophes that we've been modelling for the last 30 years, and

with each and every one of those events we've learned more. Today's

agenda, I'm going to be up here for a little while, another 50 minutes or

so, so settle in, to talk about our current business, our product lines,

where we've been, where we're going, some new market opportunities,

and to chat a little bit about Risk Intelligence.

You've heard about artificial intelligence, AI, business intelligence, BI.

Well, we're introducing Risk Intelligence, RI. We're going to take a break

and then some of my colleagues are going to come up, Cihan and Moe

will really talk about the data and the risk analytics, and our new

products that we've evolved over the last year, and talk about bringing

models to the cloud. Then my colleagues, Mohsen and Robert, will get up

and talk about our core model business, and what we see in our crystal

ball for the future of that aspect of our work. Then I'll come back and

wrap it up with our opportunities for growth, and the investments that

we're making, and then we'll open it up for Q&A after a short break.

Our heritage, we are the market leader, and I think you've seen versions

of this slide over the years in terms of how we lead in the market. Most of

the large players use our models deeply in their business. We have over

400 models in a variety of geographies and across a variety of peril. A

peril for us is hurricane, a flood, an earthquake, a wildfire, and

geographies are generally country based. It can be even more specific.

Our models now are specific location based as we will show you.

We have deep science and the highest quality models in the market, and

this is recognised by our customers, by the market in general, by our

competitors. This is our claim to fame, and it matters more and more in

the market as the landscape of catastrophe models is changing. Our

business today in terms of our product line, we have those 400 models in

our portfolio, but we also have the data that underlies those models that

we sold just a bit of into the market.

We have software and platforms, and in the software side of the business

the way to think about it is that we have a core platform called RiskLink,

which those of you who've been following us for years know all about,

and RiskLink is the software platform that lives in the datacentres, or on

the premises, of our customers today that runs our models. Our models

10

don't work without RiskLink, and another product called RiskBrowser.

By the way, models are also built in software, I think you know that.

We also have a specific platform called MIU that is used to serve the

insurance linked securities, or the ILS, capital markets of catastrophic

bonds. We also have a Hosting Plus cloud service where some of our

customers host some of our models in the cloud already. For example, all

of our customers of a product we have called LifeRisk are in the cloud

already. Some of our RiskLink customers have outsourced that to us, and

we have some 80 customers in our hosting service today. So these are our

various software and platform offerings.

Finally, we have analytical services, which is simply this, in order to run

the model shops, in order to understand the inputs and outputs of the

models takes quite a bit of technical expertise, and some of our

customers have chosen to outsource that to us. We have hundreds of

people in India who are the outsource model shops for some 60 of our

customers today. How to think about how our revenue divides as you

would expect, the majority of our revenue comes from licensing the

models, and licensing the software that runs the models. We have an

eight figure data business, and we have an eight figure analytical services

business as well, but what really drives revenue today are the models and

the software underneath of it.

To put a finer point on it, our models come in families, and most of the

families of models are built around global climate and global earthquake

in various geographies around the world. As you can see we have

additional models, such as flood and wildfire for emerging natural

catastrophe risks, and we have a set of other models that have been

around for a little while that make up a minority of our revenue. Here

represented underneath of that you have two ways that you can use our

models. One, the on premise software platform of RiskLink, and two, our

new platform, Risk Intelligence, which is offered as software as a

service.

Now, whether you're buying our data products, or licensing our model

products, or licensing our software platform using our hosting services,

or even the majority of our analytical services, you are acquiring a

subscription. So, we have a subscription based model rather than a one-

time revenue model for all of our products, or most of our products, 95%.

Subscriptions can range from one to five years and the average on a

revenue basis is two and a half years per subscription.

11

Now, the customer renewal rate, I know we've talked about before, in

2018 it was 89% and we've seen an improvement in customer retention to

98% this year. The way I read that is a great vote of confidence in the

changes happening at RMS, and the market voting with its feet and with

its wallet to give us more wallet share, and we take retention, the way we

calculate it is like for like revenues. So, if you've licensed these 10 models

this year and you license the same 10 models next year and you're paying

us the same, that's one calculation. If you're paying us 5% more, that's

another, and then that balances out, if you're paying us 5% less then

that's a 95%. So, that's how we calculate it. It's not on a per customer

basis, it's really about revenue retention.

Our customer mix is predominantly the insurance sector, and we sell to

three segments of that market. One, the primary insurance carriers who

do property and casualty insurance. Second, the re-insurers in that

market, because risk is transferred at some point in time, often from a

primary insurer to re-insurer to balance out that portfolio and to mitigate

the risk is one strategy, and finally to the brokers who are also our

partners.

Revenue from other sectors, government, capital markets, the ILS funds

that I was talking about, as well as an enterprise or two along the way,

make up a small portion of our revenue today, and frankly we've never

pursued those markets with any gusto in the past. So just taking a step

back, the market that we serve, so you can get an idea of what's going on

for them, we'll talk about the size of that market, and a couple of the

trends that impact them because those trends then impact us in our

business, is that we serve the P&C business (Property & Casualty), which

is about $1.6 trillion in gross written premium today, and it's growing on

a non-adjusted basis of about 5% a year through the year 2030, but on an

adjusted for inflation basis it's seeing growth just under three and a half

percent CAGR.

Now, is there a correlation between the growth of that market and the

growth at RMS? Never has been, don't suspect there will be, but it's

important for you to just see the market dynamics that we're playing

into. About a third of that $1.6 trillion, or about a half of $1 trillion is

strictly for property insurance, and that is the market that we deeply play

in today, because obviously it's the property insurance sector of that

property and casualty or property and liability market that would need

our models to underwrite their property insurance policies.

12

There are some pockets of growth, there are a little bit more than others.

Asia-Pac is growing faster than the very mature North American market,

by about double actually. But it is what you would expect it to be, and

again, I would say there's no direct correlation between the growth rates

in this market and growth at RMS, and I'll show you why in a little bit.

You have quite a few companies that are in a segment for us-

You have quite a few companies that are in the segment for us to go after,

but most of these companies are owned by another company who is

owned by another company. There has been great consolidation in our

industry. So not all of these are standalone entities, but we look for who's

out there in the market using our stuff and off doing regulatory filings

with our models underlying their business. So before we go further, I'm

going to give a rather non-technical view of what the heck is a model

anyway. What does it do, and why would anybody care? Well, you would

think with claims history over property insurance that you'd be able to

collect all the claims, see what you paid out, what those losses were, and

use, apply some artificial intelligence and use that as a basis to model the

future risk of what's going to happen in your portfolio.

And if you thought that, you would be wrong, and I'll tell you why. There

aren't enough valid years going back to actually model this in a way that

can tell you the probabilities of what losses you're going to experience.

So in a model, we actually go back some hundred thousand years, and we

generate a synthetic history, if you will, with all of our scientists who

have specialties in earth sciences and so on, combined with our data

scientists, and they work for years on this stuff. And they come up with

the potential intensities and tracks of hurricanes, the potential seismic

effects of earthquakes and so on for all these possible stochastic events,

stochastic meaning randomised events because they are. And then they

apply the deep science to that and figure out all the different things that

might happen. And then you have to figure out what it might happen to.

And those are your exposures. So for those of you in the room, that

would be your house. That would be your car out front. For those of us in

this building, it would be this beautiful, god knows how expensive this

place was to fix up. It's so beautiful, this hotel. This is an exposure, right?

And so what you care about, if you're covering this in an insurance policy

is how much exposure do I have to these events that may or may not

happen. So I can estimate the loss of that. And the model does that too.

Now damage can be done in any number of ways. So these models get

even more complicated because you could have a hurricane where you

get damaged by the water damage and the flood. You could get damaged

13

by a storm surge. You could get damaged just by the wind. For a wildfire,

you can avoid the fire, but you can be in ruins over the smoke damage.

And so you have to figure out what hazards you want to model and how

one impacts the other. So again, the complexity of the models gets

deeper and deeper, and the science really matters. So then you come up

with parameters. You input vulnerabilities, another step in the model to

evaluate what a damage ratio would be to these related uncertainties. So

from all of these events, we generate a curve, and that gives you a

probability of the loss that may happen. And guess what you get to do

with that. Well, you get to price the risk because now you have a deeper

understanding of it, and you get to demonstrate to your Board and to the

regulators that you have capital adequacy to cover that risk. Or you're not

allowed to get into that business because that's how the insurance

business works here.

And so if most of the time our clients use our models for many things,

and I'll talk about that. But this is super important that they understand

this risk. So when I say we have 300 scientists that know this stuff cold,

it's really important when I say we have 400 models and the data to back

this up. It's an important launchpad for the other businesses that we'd

like to be in.

But that's what a model is. And that's what a model does, and that's why

it's kind of hard. So models make markets. All of these markets came

about because you were able to model. And model inadequacy is what

happens, say in 1992, when Hurricane Andrew hit, 10 leading insurance

companies went bankrupt to zero absent the proper model to understand

the risk to do all the things that we just talked about. So these markets

were enabled. I think it was our scientists here, Robert, who coined the

term ‘models make markets’, and it's true. So the whole insurance linked

securities market was made possible because you could model it. The

Bermuda reinsurance market was possible because you could model it.

Individual location risk pricing, which is a new innovation that we

brought to market over the last few years is now possible because it can

be modelled and so on.

So for an individual customer, let's take it down from the market and

down to an individual customer. Most important to our customers is that

you can bring new products to market and you can sustain your existing

property insurance products because you can confidently understand the

risk. Because if you can't confidently understand the risks, you certainly

can't explain it to the regulators or your Board, and you can't be in that

14

business. So to me that's the most important one, representing half a

trillion dollars of gross written premium. Beyond that, the rating and

regulatory approvals are super important because it's mandatory and

because you have to prove to them that you got it going on, you

understand your risk, you've priced it appropriately, and you're not going

to get in trouble, go bankrupt, walk out of the market without paying

your claims. That's not what the regulators want.

You have to demonstrate to your Board that you have a good portfolio of

risks that you built over time and that as it changes and that as exposures

get added, more houses along the coast. Right? Everybody likes to live on

the water and in climate change areas, and that gives you more

exposures, more risks over time. How has your portfolio changed?

Because it is dynamic and changes every year, and you have to prove that

to all of these folks. Efficient use of capital. Capital requirements are a

big deal in this industry. And finally, if you want to transfer your

portfolio of risk to somebody else, a reinsurer or an insurance linked

security, a hedge fund, basically, you have to show that that's a good bet

for them, and you have to show transparency and that models make all

this possible for individual insurers. So fundamentally we know a lot

about the business of risks, loss, exposure, portfolio, time. All these

things are in our core competence, and we'll be leveraging that in new

ways going forward.

Our customers go through a risk management lifestyle that is at the

centre of their business selecting risk after they identify it, pricing that

risk appropriately for all the reasons we just talked about. How do you

optimise the risks? That's another department in an insurance company.

How do I understand it, and how has this risk over here on property

related to this liability risks related to this DNL policy or E&O policy and

so on? Was I liable at PG&E for something, and therefore these things are

all attached? More and more, our clients want to understand the

interrelatedness of the risks that we assess with everything else that they

do. We're a part of an ecosystem, but we're not the centre of the

universe. Right? And so managing the risk, mitigating the risk, and

finally on the profit and return on investment. This underlies the

property insurance industry today.

And this is the risk life cycle that our customers talk to us about each and

every day that our models have enabled decisions in each of those but

somewhat indirectly, and I'll come back to that point. When things go

right, everybody's happy. Shareholders love it. Profitable growth, all the

boxes are checked. Our loss ratios that we measure quarter to quarter, so

15

manageable. We have profitable risk. We have efficient use of capital,

and we get rate agency upgrades. That's a good thing. We're in

compliance. We don't have to worry about that. Chief risk officers are

happy, stable, profitable growth. All signs are up. So when this goes well,

it goes very well, and it goes very well for a long time. When it goes

badly, it goes very wrong. It's catastrophic. If you look at these numbers,

a $200 billion catastrophe last year and it took 18 months to figure it out

in the rear view mirror. That's not good.

That's not good modelling, and that's not good risk management. You see

these losses, 4.7 billion catastrophe losses. You see the cuts of staff.

Thousands of people get pulled back. You see bankruptcies. So when

these headlines hit in a challenging year, it shows what happens when

you don't keep your eye on that risk management life cycle. We are super

important to our clients. So I don't know about you, but when I looked at

this coming in, I thought, "Great, I buy it. I believe it." So if this is so

important and if you get it wrong, it costs the industry billions of dollars

through bankruptcy, lawsuits, all kinds of things, why is RMS stuck at a

$300 million business? Why aren't we growing like a weed? Why aren't

we a much bigger and more important company? What is going on here?

And I will come back to that. But before I do, let's get into some of the

trends that the industry faces.

They have some headwinds, and they have some tailwinds. And there are

many I could talk to, but I'm just gonna pull out those that are most

impactful on the industry and those that impact us the most as a result.

So let's start with the bad news. The cost pressures are enormous. The

losses have been off the chain. And this is an industry that is legacy. It's

regulated. It's slow to change. It's slow to transform, even hearing about

this transformation for how many years now, right? And so on the left

you see ... like you think about a cost cutting in these firms. When Zurich

is talking about a $1.5 billion cost cutting plan, that's a B. That's not a

typo. That's something to give us pause. When you talk about job cutting,

not hundreds but thousands of employees to catch up and when you talk

about when the C-suite in these companies are talking about risk over

here and cost efficiencies over here, and if you sit on their calls. And I

know some of you do. That's all you're going to hear about right now and

you will.

That's all you will hear about from these guys for the next couple of years

because of this delta. Those that are optimised versus those that are not

optimised for this, there's a hundred percent difference in their cost

structures. Guess whose shareholders are happy, and guess whose

16

shareholders are not and are putting on this incredible cost pressure. So

the industry costs have been way too high for way too long. And now

with these losses coming in that were unanticipated. Now it's really the

perfect storm.

And so when you think about a lot of the complexity that's been added

has come from operations, which is being cut is the people costs, but

look at this: fragmented legacy IT systems. I cannot tell you how often it

is that I go and talk to the CEO of one of our clients who says to me, "Why

have I for years been building a product for one. I realise now my IT

shop has been building a product just for me. I'm the only user of it. It's

costly, and they didn't tell me that when I invested 5 million in it this

year, that I had to invest 5 million in it next year and the year after and

the year after to keep it alive." And these costs have now become quite

burdensome.

So internal IT custom projects are getting slashed back. Internal

operations and people are getting slashed back, and operations and IT

account for 61% of a PNC insurers’ total cost base. Think about that. So

that trend hurts us because the industry is under tremendous cost

pressures. So they really want everything in terms of pricing to go down,

and we're a big ticket item. So we feel that heat.

The second piece of that is that the market that we serve, the property

insurance market, has been good and soft for about 10 years now. And in

addition to the cost pressures that that's caused and the anticipated

losses that were experienced, you see a ton of M&A activity. As you can

see, last year was an enormous increase in M&A activity, over $40 billion

in transactions, a hundred transactions, and these aren't all nickel and

dime transactions. 10 of them were mega and the mega transactions

made up $30 billion of these. So we see our customers consolidating with

one another. We see our customers consolidate with non-customers, and

this impacts our business as they define synergies through that M&A

transaction. If they're successful, they need less from us because they've

consolidated. They may ... I have a customer who did a large M&A

transaction, and they decided to exit four markets entirely because they

were losing money in those markets. So they don't need models for those

markets anymore. So that impacts us directly in a negative way.

So those are some of the headwinds that we experience, that we see

every day in our market. So when I think about those and I compare it to

our 98% renewal rate, I'm pretty happy in the face of what I'm seeing in

the market as far as headwinds with how our market is responding to us.

17

Because you would have expected this to be a lot heavier for us, and it's

not.

We are seeing a ray of light there in terms of the beginning of the

hardening of a market that we serve. And that was in the beachhead,

which is really the Florida renewals. That's a big market. Obviously the

bay hurricane market supported by Bermuda and for the first time in a

long time we saw insurance premiums creep up. So I tested this of

course with Paul Zwillenberg, he's got a house in the Miami area, and

sure enough his premiums went up 11%. So right about the average,

Paul.

And we see premiums upward of 30% in some cases, and some aren't

going up at all. These well-written portfolios going into the reinsurance

market that our models underlied are not seeing these increases. They're

very balanced, nice portfolios, and riskier others are going up quite a bit.

The point being here that finally, I mean when, when you see these ... we

call them the ‘HIM’ hurricanes, Harvey, Ivan and Maria ... losses of

almost a hundred billion dollars in the market, didn't adjust premiums in

all of 2018. That was a little surprising. It was a little surprising, but it's

finally catching up in 2019 and beyond. So what is the industry doing to

manage its headwinds? And demonstrably they're moving to technology.

One, seven in 10 of the insurance companies are now doing something in

the cloud, and more interesting to me, almost 20% of the new stuff

they're building, they don't want it built in legacy old IT systems

anymore.

They want to build in the cloud, so they're going straight to the cloud.

And they're learning. The first reason you go to the cloud is you save

some money, and you save some time. The second reason that you stay

in the cloud is because you can build stuff faster and cheaper and better,

and that allows them to bring new products to market more rapidly. And

they're figuring that out. The second is artificial intelligence. And the

first move to artificial intelligence is to take a look in the rear view

mirror at where you've been. Did this customer campaign work? What

was the outcome of this thing I did in the past? What's more interesting

to me is what happens next is when you use artificial intelligence and

machine learning for real time stuff, real time decisions about your

business to make data driven decisions in real time, and the second is

then the next step of course, predictive, predictive analytics driven by

artificial intelligence.

18

The point is when you say 80% of them are there already or are going to

get there, 80% of them are not there yet. It is aspirational for a number of

them as I speak with them, but everybody knows this is coming.

Everyone knows the thought leaders are embracing this, and so they're

turning to this technology as well. And finally, digital transformation,

while their IT legacy systems are getting cut back and the waste is getting

cut back, they are investing. 25% of the R&D budgets in new stuff that is

transformational to their business, that will support their new business

models. And again, I started this this afternoon by saying this is the fun

part for me. Transformation and technology are converging at the right

moment of time because as you can see from the headwinds, these guys

can't sit back and wait anymore. They are compelled to move forward.

And so we're moving forward too and to help them with that. So the

impacts of those dynamics on us is that we can help our customers with

their cost issues through technology. And this is important, and we can

help them understand their risks better as well. But there's a broader

play for us that I find even more intriguing, which is how do we partner

with the industry for their future growth, new insurance risk products

and transformative business models that couldn't be supported with only

natural catastrophe models. And our customers are turning to us for

that. There's one other trend I want to cover off real quickly because it'll

put a fine point on things I hope, which is if you look at the last 30 years

of natural catastrophe losses experienced, it's been almost $900 billion

[in North America] over 30 years. These are big numbers, right? These

are losses, and remember losses are okay if you expect them and you

capitalise for them and you price for them and you can profit from them.

But losses are not okay when they're not.

So what happened in the last two years? $185 billion or 21% of those

losses were realised in two years alone. Is this a blip on the radar screen?

Well, no, it's not. Let's look at the global numbers. Same trend. 17% of all

global losses occurred in the last two years. This is a fundamental shift in

the core market that we're in, is that we now live in a world of constant

catastrophe. It's not going to play out and go quiet for five years, then we

have another hurricane. This is it. And we'll talk a little bit about why

later on when our model team comes up here. But this is it. So when I

talk to a CEO at any customer, you know what they're worried about? Of

all these events, seven, seven events, hurricanes and wildfire, seven

events only drove the vast majority of these losses.

So what I care about is where in the world is that eighth event? And am I

ready? Am I exposed? Do I understand my risk? Is there a liability policy

19

at a utility that I'm going to get hit for? Is there an E&O policy I'm gonna

get hit for? What about my property damage? What about business

interruption? And that's what they worry about. That's what keeps them

up at night. Where is that eighth event? And that's where we come in. So

how do you solve for that kind of problem? And there's only one way.

And that is through innovation. So this industry is poised for its

transformation. And if the people in it don't transform, they will be

disrupted. Robert's going to get up here later and talk about something

called the protection gap. The difference between what is lost and what

is the insured loss.

And that number is getting bigger and bigger and bigger. And as it grows,

it's a giant opportunity for somebody smarter about it to go close that

protection gap. And it leaves the existing incumbents vulnerable. And

the thought leaders know that, and they're doing a lot about that. But

they know it's one or the other, transform or be disrupted out. And I

think the Boards realise that as I go and talk to them. So what's been

going on and how does all this come back to RMS? Well, we went on a

customer roadshow, myself and all the new executive team, and we

talked to the vast majority of our customers over the last six months.

We've been traveling a lot. I think my United frequent flyer miles was up

to 750,000 points or something since I've joined RMS. Some of those are

bonus miles. Okay, I'll admit it, but nonetheless, it's a lot of travel.

We've done a lot of listening, and sentiment came up time and time

again. And this is capturing a notion that many customers talk to me

about is that you guys have been trapped in your platform problems for

years now, and the other guy didn't even bother to innovate. They sat

back on their laurels. In fact, the other guy was our main competitor

spent 3% of their money on R&D and said, "Yeah, let's just let it ride for a

while." Right? And so we haven't seen real innovation in the industry in

quite some time, is what they told me. Now, I could argue that our HD

models are quite innovative, but we didn't do a great job getting those

into the market and into the hands of our customers. We're changing

that now, but I think by and large, I agree with this sentiment.

So what did we do about it? We came here, and I realised that RMS had

done a whole lot of talking over the last six years. So we started doing

more of the doing than the talking and we went pretty radio silent on our

customers for a while. As a matter of fact, I was two months in before I

even talked to my first customer because I very poignantly wanted to

spend my time with the team, with the experts and figure ourselves out

to bring on a management team and to really spend the time there before

20

we went out in the wild and started seeing what was going on with our

customer base. And so a couple of things I'd like to share with you that

have happened over the last year. One is that we defined and launched a

new strategy for the business, and I'm so happy to say we recruited a

really top team that I'm having a ball working with. And we're getting a

lot of stuff done, and you'll have an opportunity to meet some of them

later this afternoon.

The renewal rate. Again, for me that is a great sign of confidence from

where we were a year or two ago to where we are today. Companies are

acting with their money, and that's always a good sign that what they are

saying to us, where we feel confident in you, and what they're doing,

giving us money, they match. Right? And then the third thing is that we

restructured how we work with a very important community that we

serve, which is the broker community, and we have secured new

agreements as a result. Now in the past our agreements caused a lot of

friction. We would hamper their ability to serve RMS customers really

broadly because we didn't want them to do this; we didn't want to do that.

On the other hand, they were able to take our intellectual property and

share it with non-customers in ways that just left things quite ambiguous

and not aligned in the market, and I believe partners need to be aligned.

So we were able to restructure how we work with the industry and two

beginning pieces of that is that we signed two long-term agreements with

two of the largest brokers in the world, Aon and Willis, and those

agreements together total a total contract value well into the nine

figures. So again, if I think about the vote of confidence that I was

looking for from our customer base that we're on the right track, that

certainly was one along with our renewal rates that I was pretty happy

about. And then going forward, as we move forward in the market, we'll

continue to get the other brokers that we work with in that community

around the same structure to align us and servicing the market for risk

more deeply than we have been together in the past.

So we promised a lot of things at our customer conference a year ago. I

was a couple months into my new job, and we said we'd deliver an open

modular platform because nobody wanted a closed monolithic one. And

that's what we had built. We promised that we would bring our new HD,

which is high definition, location level, really interesting technology to

market and our flood model. We would put that on the platform and

deliver it. We promised that we'd move our on premise models onto the

platform too, all of them, all 400 of them, that we would focus on price

performance because it's important if you're under cost pressure that we

21

give you cheaper ways to acquire our product and that we would have a

greater obsession with what our customers needed from us rather than

guessing, taking it to market and saying we got it wrong. RMS tended to

look at things from the inside out, not from what the market needed in.

And that was one of the reasons we stumbled. So we made all these

commitments, and I'm happy to say that when we went in front of our

customer base last May, two months ago now, that we showed them that

we delivered on each and every one of those commitments and more.

And we'll show you that this afternoon. So what's our future? What's our

new strategy? Everything before for us was a natural catastrophe model

with a couple of models like terrorism or marine around the edges, right?

And now we're looking at ourselves as a risk management solutions

provider. So that changes how we think about what we need to bring to

market. No longer point solutions model by model, by model, by model,

but really risk solutions that can be deployed across the businesses of the

customers that we serve as they deal with the emerging risks landscape

that they're dealing with.

We aspire to bring strategic value by delivering a next generation risk

platform along with a suite of models, a suite of products, a suite of

services that answer the needs of the broad customer base. And we're

committed, and this is super important. We're committed to the

beachhead that we own. We will always maintain our thought

leadership, our science leadership, our market leadership in the core

business of natural catastrophe and risk models. That's not going away,

and as you'll see in a minute, it is the springboard from which we

launched into these other areas. So if you think about yourself as a

platform company and as a risk solutions provider, you make decisions

differently than when you're the model vendor. And we've seen that over

the last year with what we brought to market. And I'm happy to say that

the market has been great in giving us really open feedback, and I'm

hearing from all the executives and the customers that we serve to a

person and endorsement of the new strategy and the direction that we're

headed.

A confidence in the team that we brought on to deliver on that strategy to

help their business. And in a couple of the quotes here, you see in one

case the CEO and another case, the chief underwriting officer at very

large firms, talking already about “We're gonna have touchless

underwriting. And we want solutions that will enable that new business

model.” Already they're thinking about us in that broad other category I

talked about beyond efficiency, understanding risk better, doing it at a

22

lower cost. What about the other part? What about new business models

coming to market? And you're seeing some ideas there. When a customer

says, "Hey, this is a means to respond to disruption in the industry where

transparency could fundamentally change the way business is done." So

those were fighting words, and I was very pleased that the reception we

got as we unveiled our strategy to our customers over the last six months,

one by one by one. And we spent a whole lot of time on the road

together, which is awesome.

Everybody's tired here. Look at them like ... So I promised to get back to

this point for you about if we're so important and what we do matters so

much, why are we in this slow grow $300 million revenue run rate

market wall? I'll tell you why. RMS never got out of this one box, and this

one box is the market opportunity that lives in the cost centres of some

insurance companies that do property, and it's called the natural

catastrophe modelling space.

And it is $500 million large, and you could argue it was at $475. is at $550?

it doesn't matter. It's not moving much, right? It doesn't have a bunch of

headroom because this is what happens, and this is just the model piece

of it. The one thing that surprised me when I started at RMS was that the

model scientists didn't really fully understand the value of the results of

the models and all the different ways that IT and other companies and

others had leveraged this great value into other solutions that were more

directively enabling data-driven risk decisions. We were sort of one step

removed from that. So we lived happily in this box. And did RMS aspire

to get out of it with RMS(one)? Sure. Did they succeed? They did not.

They did not. So we've lived in this box, And this box isn't getting much

bigger. So we're stuck in this, and we've made people fight really hard to

get the data-driven decision value out of the stuff that we have.

So we're not going to do that anymore. There's another market, the

property and casualty data market, and this serves the real estate

industry and the portion of it that serves insurance and that, that we

could address as a billion dollars. As I mentioned, we have a paltry eight

figure business there today. Should be much larger because people know

the quality of our models is owing in part to the quality of the data that

we put into our models and the quality of the outputs from our models.

Why don't we leverage that in the market? Because some customers want

data, and we've actually hidden it. Technically it was really hard to get to,

and we'll show you what we've done about that. So this is another market

that we opened up for us that is a total addressable market is about a

23

billion dollars, and then above that ... And this is where it gets spicy, is

the insurance analytics market.

And that really, I'll sort of define that in a couple of ways. And the piece

of that that we can most directly address very quickly is the risk analytics

market, that we can play quite broadly in insurance analytics as well.

And so let me talk a minute about insurance analytics. You have a lot of

horizontal players there. You have Microsoft. You have Oracle. We'll give

you some other names in your kit. You'll see, and basically they provide a

toolkit, an empty shell for you to put in your domain expertise, for you to

put in your presumptions, you to build more custom IT solutions from.

So that's legacy. That's legacy. I love walking into legacy markets that are

mature and underserved and do something fun, and we plan to do that

there. On the risk analytics side, you do have companies like Verisk, who

bought some things in the risk analytics area, and Guidewire is another

one who does that in the underwriting area. And they tend to focus on

operations, and they tend to focus on the rear view mirror.

What are the analytics of things that have happened before? And that's

interesting. That's really interesting to learn from. But what's more

interesting to us is how do you make near real time or real time data-

driven decisions by having the right information at your fingertips, the

right information turned to insights at the time you're making a risk

decision, whether you're selecting risk or transferring risk or optimising

risk or whatever you're doing with that risk? How do we do that? And that

becomes really interesting and quite potentially differentiating. And so

another way to think about the analytics market is where you're looking a

lot of analytics today is in the rear view mirror. You're looking at things

that happened in the past and you're trying to figure out what that means

to your future. But you're really looking in the past.

We have people who were exposure managers saying that their data's

three months behind. Well, I hope there's not a hurricane in hurricane

season in those three months or it's too late for you to understand what

you need to do, oh, with your $17 billion potential loss portfolio. These

are big deals. So one thing is to move from the past.

The second set of analytics is really in the present. How do I make data-

driven decisions today in real-time? And the third that will come next is

how does this predict my future and what should I be doing and why?

And that's where some really fun machine learning and artificial

intelligence can come into play, and that's a third layer.

24

Almost none of this business today is forward looking and predictive like

this in any useful way. They're quite rudimentary tools. So again, Legacy

market, lots of room for innovation there. And finally, emerging risks.

We have a few plays there and new insurance lines, and I don't mean

new insurance lines that the industry will invent, but we are only in

property today. What if we looked at liability and opened up that market

with these risk analytics? Might that be interesting? So that's what I mean

by that and I won't size those.

So what do we have today? When are we going to get into these great big

vast markets so that we can take our total addressable market from the

$500 million that it is today to the $8 billion that we aspire to? So as you

know, our core products, well-served, we continue to innovate there, we

continue to do R&D in this area, we continue to update our models, and

we will hold onto that leadership because it gives us the right to move on.

In the property data space, we do have a half a dozen data products, but

we kind of locked them up and made them very difficult to use usefully

when you're a consumer. We launched a new data product called

Location Intelligence, that we'll tell you more about this afternoon, in

June of 2019. So we already made an aggressive play into this market,

and in the insurance and the risk management analytics market, we

announced our first product, which we'll demo for you called Site IQ and

that launched just last month. We announced another project called

Exposure IQ and yet another project called Treaty IQ, all in the risk

analytics market, all of which will come out for our customers in 2019.

So those are our first three applications in the analytics market. So we're

not waiting, we're in that market now. What makes those possible and

what makes them differentiated is our model layer and our data layer

that feeds into that, and that will be more clear to you as you see the

demonstrations that we'll show you.

In terms of emerging risk, we announced in October of 2018 the latest

version of our cyber model because that is an emerging risk for us and it

is an insurable thing now. There's something called silent risks that

everyone's worried about, which is where your cyber payouts happen in

your property insurance, business interruption, E&O insurance, oops,

I'm liable for this. I made a mistake and I got a cyber attack and so on. So

insurance companies find that they're exposed in lots of hidden ways.

Our cyber model is the first to address not just direct cyber, but also what

we call the silent risk.

25

So what makes all this possible? Well RI, Risk Intelligence, is the

underlying platform without which we could not deliver these integrated

solutions to market, because the market is mired with point solution

after point solution after point solution, which has added complexity to

everything and cost. Those are all those IT costs and complexity costs

that are coming out, and has actually confused your view of risk, causing

stumbles in the market by some of the clients that were underserved. So

that's our plan in terms of new addressable markets.

Now, let's just take a look at the analytics market. I would give you as a

proof point, this market is growing at three times the rate of the property

insurance premium market. So this tells you that while I'm cutting costs,

while I'm cutting ops and while I'm cutting head and while I'm cutting

Legacy IT, I am not cutting here. This is an area of investment that's

working for me.

As you can see, the incumbents on the insurance analytics tend to be the

more horizontal players. The risk analytics tend to be the more vertical

players, and they're names that have been around a long time in this

space, and I really do love going after Legacy markets with new stuff, and

that's what we intend to do. Real-timedata-driven decisions, predictive

analytics is the space that we'll play in.

The other piece of the transition is to the cloud that I should address and

that is, we talked about people going to the cloud. Look at the growth rate

in that business. Amazon Web Services at Amazon is growing at about

50% a year. Lots of headroom there. Only 3% of IT is in the cloud so far,

so we're going to see numbers like that for a long time to come, and so

many, as I mentioned before, of the new applications being built by

insurance companies are cloud first.

By the way, all of our analytics are cloud first. We don't have to migrate.

We're building cloud first, which is very exciting to our customers. When

you hear a little bit more about that from Cihan, I think you'll understand

why.

So what happens when our customers go to the cloud? Well, they shut

down a data centre, they stop buying hardware, they don't have to hire

the people around it, they don't have to pay network costs, they don't

have to pay for storage and compute and so on. Instead we hosted in our

cloud, so even while we save them money, we get a larger share of the

customer budget. So we can expect that as customers go from on premise

to in the cloud, that will be a revenue opportunity for us at RMS.

26

So what's our focus over the next three years. First is to deliver on the

strategy and being that more strategic partner. Again, hearkening back

in my days of Oracle when we went from being the database platform

vendor, which was an important role and it was a great starting point,

but how do you take that and become a strategic partner and become

mission critical even more so at a higher level of value to your

customers? That's what we're doing here at RMS.

The only way you can deliver on this is with a platform. There isn't

another way to deliver this kind of value to customers. All the enterprise

grade that you would need to have, all the flexibility, security, data

analytics and so on, and the way to think about our platform is in two

giant buckets. One, really great model execution for risk modelling, and

two, really great data and data analytics all in one platform hasn't been

done before.

So, the next thing we started to ask ourselves is, we do get a premium for

our models because they are the highest quality, best science, and our

customers do love us for that, but the customers are under cost pressure.

So outside of anything else, what would happen if our competitors’

models were free? No license fee whatsoever, yet we could still deliver

the model solutions at a lower cost of ownership for each and every

customer. So we started to ask ourselves, how close to that could we get

to help them save some money? Because again, with a model, it's not

untypical in the software business that if you spend a dollar on a piece of

software, you might spend 75 cents to a $1.50 on the stuff that surrounds

that, the hardware underneath of it, the people supporting it and so on.

This is not uncommon.

So, what happens if we tackle that piece of the cost so that if our

competitors' stuff was free, we were still cheaper to own? So that drove

us to really work on price performance of the platform because that's

really key to our customers to have something efficient that is high

performance and lower cost.

And finally, how do we develop new partnerships? So we're going to

show you some stuff we've developed. We're going to show you a few

ideas about the future, but we're not going to tip our hand of all the stuff

under the hood, of all the ways that we're working with clients to bring

new products to market next year and the year after that. But we are

forming these partnerships as a strategic partner to our clients to bring

solutions to market beyond cat models. And you'll see three of those

today.

27

So, why a platform approach to market and why am I harping on that?

The kind of scale, the kind of flexibility, the kind of extensibility from

one line of insurance to another, from one kind of risk to another, and

the kind of agility and speed that you need as you transform your

business or you are going to be disrupted, cannot be delivered through

point solutions that are tied together through custom, very flaky code,

right? Because it's very hard to hold a system together like that, and our

customers are imploding from the weight of them, and our own Legacy

systems are imploding from the weight of them.

So platforms matter. I've named a few here, whether it's a consumer

platform like Amazon or their AWS business, whether it's a business

platform like Oracle, there are some common principles that each and

every successful platform in the world has followed, and I learned these

when I first worked on my first platform, which was the Oracle database.

30 years later, still market leader. So you can imagine one of them is

going to be future proof.

These are the five principles that I believe in as the underlying success

factors for every successful platform. So one, open, and what do I mean

by that? Well, open is important, that other people can come onto the

platform. If you can't host a custom solution from a big insurance

provider, if you can't let a broker come on with their cool stuff, if you

can't have open ways of doing that through application programming

interfaces, and by the way, open does drive a lot of technical

requirements on your platform, then you don't get to win, right?

The old RMS(one) notion was we're monolithic and we're closed. You

come onto our platform because you need us and we're going to lock you

up, coming on is hard and coming off is hard, but we got you. I love to

compete on having the best value to bring to a customer. We have a

couple of competitors. I'm used to having hundreds of competitors. I like

it. I like to compete. I like to win here. You have to be open to win.

Future proof is important. People make investments in you and you have

to get your architecture right. And if you don't, you have to change all the

time. And you see technology companies fail on architecture all the time.

We will not do that. We're future-proof.

Modular, you can't tell everybody that you want them to buy this bundle

of 400 things when they want one thing, right? So this whole value

proposition before from RMS(one), well, you got to take everything we

have. It literally was coded together and locked up. The data was stuck or

28

the application was stuck with the models. It was all one thing. So if you

wanted a piece of it, you couldn't have that. If you wanted to access a

piece of it to put in your ecosystem, you couldn't have it. Modularity is

important. Choice matters a ton.

Our customers have different ways of seeing risks, different ways they

want to enter the market, different products they want to bring to the

market. If you give them a platform where everybody looks the same and

they can't leverage in clever differentiated ways, you don't get to be the

platform of choice for any industry. So you have to allow them

differentiation, and that also speaks to a bunch of technical and

architectural requirements that you must implement in order to offer

that differentiated choice to your customers.

Finally, you don't get to be the industry leading platform until, unless

other people make gobs of money off of you. Not just your customers,

but you pull in partners, you create an ecosystem. We had thousands and

thousands and hundreds of thousands of partners. We had millions of

developers at Oracle that were building on our database platform. In

fact, if you think about the history there, we were this database platform,

we started building competitive applications to SAP. Well, we wanted to

be the platform for all the people we were competing with, and we ended

up, despite competing at the application there with SAP, being the

platform underlying 75% of their customer implementations.

So you must be open, you must have value, you must allow partnering.

SAP profited enormously from that, even though they competed with us

for financials and ERP applications, right? So it's important, it's

important that you allow partners on and third party solutions on. The

way I look at this is if you like our exposure manager product and our

underwriting insights, great, you should buy them. If you don't, you

should be able to plug it right into the platform, use our data layer, use

our models still, and then we just didn't win your business. Let us

compete on value.

RMS(one) was sunset because it didn't follow these five principles of

platform, and it failed in the market and it failed to deliver as an

architecture and it didn't have a fit. So we've been, over the last year and

four months, kind of changing the engine of the plane while we were

flying it, and we announced Risk Intelligence in May on which we have

live customers, we'll show you some demos today, that meets all of these

criteria of what it takes to be a successful platform.

29

So you guys, a lot of you in the room, I've read some bio's and whatnot,

have been around 10 years or more following us as investors or as

analysts. So I think we owe you a little explanation here about why

fundamentally RMS(one) failed. It's been quite a journey from 2014

onward. I pointed to some poor decisions that were made, some

judgment calls about platforms that were wrong, but why is that?

One of them, and I'll point to two, was really fundamentally the team.

While we had an amazing host of model scientists and while we had an

amazing team of domain experts who understand the insurance

business, committed, dedicated, loyal people, what we didn't have are

folks that understood how to bring a platform to market, and how you

run a platform company is different from how you run any other kind of

company in the world. How you build a platform is different than how

you build a piece of desktop, a software, or a model. So I know it's highly

unusual in a setting such as this to give you such extensive backgrounds

of the team, but because this was one of the two things that made us fail,

I thought it important to expose you to the backgrounds of the team that

we have brought together to deliver on this vision.

So Moe, if you would start. Moe runs engineering and platform with us as

a member of our executive committee. Moe?

MK: Sure. Hi, Moe Khosravy. I've been pretty much doing data powered

platforms and applications for pretty much my entire career. I very

much focused on the innovation there. I've got 90-plus patents that I’ve

been named on in that realm, and I've been at RMS for about a year now.

Most recently I joined from HP where I was running software for pretty

much the tens and hundreds of millions of connected clients and

platforms of power core businesses, including the applied machine

learning sections, data ingestions. Prior to HP, I was at Intel. Again, head

of applications, head of SDKs for all of our applied machine learning, AI

powered devices that really use sensor data, home automation, again

bringing unique value both to enterprise and consumer at scale. Before

that, VMware on again mobile cloud security investments, and before

that a number of years at Microsoft, everything from Windows, SQL

Server developer tools to search, and was one of the early members of

the Azure team working on the basics of cloud computing and data,

where I worked with Cihan and some of the other members that we'll

talk about.

KW: Thank you, Moe. Cihan runs product for us.

30

CB: Hi everybody, Cihan. I've spent the last decade in Silicon Valley with a

few start-ups that disrupted a couple of the established players in the

market. My most recent start-up was a company called Databricks,

which reached $2 billion of valuation in a short six years. There, we were

running some of the world's largest machine learning models, including

the ones from Apple, Uber, Airbnb, Lyft, Salesforce, and some of RMS's

newer models.

Prior to that I was in Seattle at Microsoft headquarters, worked on

Microsoft's data platform, including SQL Server there. My most recent

role with Microsoft was as part of the founding team of Microsoft's cloud

offering Azure, which is $20 billion worth of revenue for Microsoft today.

That's where I had a chance to work with Moe and some of the other

members of the team that you're about to meet. Thank you.

KW: Thank you. So you get the theme here, data platforms and each of them

has brought to market very market successful products over time, and

combining that with our core data science and model science team.

MR: Good afternoon. I'm Mohsen Rohnama. I've been with RMS more than 20

years. I have more than 30 years of experience in the catastrophe

modelling. I've worked in the industry extensively, built a model, my

team is responsible for creation of all the all models we're building and

bringing the science to the hands of the clients. I've worked with all the

primary broker reinsurance through the entire market. I graduated from

Stanford University many years ago when RMS started. I'm so excited

really having Cihan, Moe and Karen bringing the team of the technology

together. They can bring the science that was needed to the market.

Thank you.

KW: Thank you. And Robert, please. Robert is our chief research officer in the

model area.

RM-W: Good afternoon, everybody. I'm Robert Muir-Wood. I'm the Chief

Research Officer of RMS. My mission is to explore new applications of

risk modelling to identify new frontiers. We can model to extend the

application of models to new areas and promote those applications. I'm

also the chair of the High-Level Advisory Board of the OECD on its

function of looking at catastrophe risk across all the OECD countries. I'm

the author of seven books, most recently a book called ‘The Cure For

Catastrophe’ which, which promotes the whole idea of catastrophe

modelling to a general reader. I would recommend that book to you if

you don't know about this area.

31

KW: Your copy will be outside.

RM-W: I look forward to expanding our whole risk modelling applications in

RMS. Thank you.

KW: Thank you. I'm such a competitive person. I'm always a little jealous of

Moe's 90 patents, of your stint at Databricks and your work on Spark, of

your PhD, and of your seven books. I'll never get that, but I've created a

little bit of market cap along the way, so maybe that's how we can

compete.

I feel very good about the new management team. We also changed out

the business side of the management team, bringing in Neil Isford, who

has run $300 million to $7 billion sales organisations at IBM, as well as

FinTech startups. He worked at IBM Watson last, building out that

business. Marilyn Mersereau was actually recommended by the Cisco

folks, John Chambers and Charlie Giancarlo there, and has done a

beautiful job on branding and transformations. Really terrific

background. Reed was in our organisation and was promoted into CFO.

So I feel very good and confident that we now have the right team that

understands not just how to bring platforms to market, but how to really

be market-driven and how to work together as a team with the best of the

model science and the data science, along with this new aspect that we

brought into the business where we had previously failed and failed quite

badly.

It's not enough though to just keep it at the leadership team. So I'm going

to double click down one more into some key pillars, because those of

you who have tracked us for a long time will understand the pillars along

which RMS(one) and our software offerings had failed in the market. So

I'd like you to understand head on how we've addressed those. Moe,

could you talk a little bit about the members of this team? Thank you.

MK: Definitely. Quickly, kind of coming in, we wanted to assess how fast we

could actually rebuild the platforms, the applications, to just overcome

some of the problems of the past. What I did was really go in and really

take proven talent. One principle I like is when you're really trying to

make a change, don't go hire people that are familiar with the

technologies, go hire the people that have built the technologies.

So Rene Bouw joins us as both chief architect as well as the head of our

Seattle office, where we're actually doing AI and machine learning,

32

tapping into our old networks at Google, Amazon and Microsoft that

we've been collecting in. We've got a very, very powerful team. Again,

Rene brings a ton of experience around enterprise platforms that scale,

whether it's Power Pivot, BI, Microsoft Machine Learning Service, the

largest data market business for commercial and public brokerage

businesses. And he augments the team on the platform side.

On the experiences, we brought in Olga who has great experience at

Amazon, Intel, HP, but Olga brings both a design mindset in terms of

really sweating the productivity of who we're designing for and those

workloads. Again, if you have Amazon Alexa type scale where you're

handling 30, 40 million concurrent clients hitting your back end, you're

going to be able to handle the insurance and the reinsurance cases as

we'll show you.

That was the back end and the front end. If I can say it, we used to joke

that with RMS, our dates were a little bit of a Hail Mary, right? And you

hope you got the year right. It didn't really have the rigour of execution

of really being able to go the planning, the processing for it. What Subhas

brings from his experience at Intel and Nokia is, and this is what we're

used to, when you are projecting out a product that is going to come out

two years from now and you have to build a factory of fab hardware,

software, client systems, and you're building all of that and you're still

able to hit a plus or minus two-week timeline because partners are

delivering on that. Actually, depending on that, that's the kind of rigour

that Subhas is bringing to the organisation so that when we give a date to

customers, as you'll see, they can actually take that to the bank, and

along with standard deviation in terms of opportunities to fold that in

and again pivot on top of that.

So, that takes care of execution and the planning for it. But let's say we

built that, we're in a space where this is truly mission critical software.

What does it mean to be able to have real SLAs that we give to our

customers to where they can actually have best in class support from us?

Neil joins us, specifically I want to call out his experience from Microsoft

and the GFS team. This is not a well-known team externally, but this is

the small group that keeps Microsoft's multi-billion dollar businesses up

and running with the SLAs, including the third parties that are

generating tens of billions, hundreds of billions of dollars on top of

those.

I'm incredibly happy about this roster because as I collected my group of

10Xers, and again, 10Xer being a term for these are the high-powered

33

engineers and executives that are literally doing the work of 10, 20

people, they're bringing in their roster, so I'm really happy on the talent

that we've augmented and truly nobody else has this staff.

KW: So this very talented team is really intrigued by what happens next in the

emerging risk market and building a platform for that risk and to

manage that risk. So I feel very good, but what's more important, frankly,

is that our customers feel really good. You saw some of the quotes. There

were many, many people saying, "What are you guys doing, and what are

you up to next, and what do we don't know? Because you didn't just come

here to build just more business models, right?" So they're getting onto

that we want to provide higher order value to them.

This team is I think what's led to that increased renewal rate, what led to

the possibility of the broker deals we did and other proof points. We're

seeing in the market where the credibility, which you have to know had

waned after six years of no delivery of a very promised and important

product line, has started to swing the other way.

Is it completely out of the system yet? Absolutely not. We've been here a

year. All right, give it a little more time, but we're seeing that confidence

shifting in the right direction and a large part because of this team.

So that was one of the root causes of our failure, but there was a second

one. I've been in Silicon Valley now since 1993. I've run companies and

I've run funds, and one of the things I've learned is that great ideas are

everywhere out there. Taking a great idea, executing on it, and turning it

into a commercially viable product, and turning that into a commercially

viable long-standing business is really hard. There is a process. There is a

known way to do this that is incredibly customer and market focused,

that brings our customers into the process quite early, not as making us

order-takers: ‘I'll take the blue one of these and the green one of those’,

but really as thought leaders about truly the underlying essence of their

problems and opportunities, so that we can combine with our skills as

technologists and innovators and come up with exciting ways to meet

those challenges and those opportunities.

So we've codified that into a product development process, and we've put

the product group which used to reside all over RMS in many different

groups, which meant we only looked at a product like I'm building this

and you're building that, and how they come together at the customer,

we don't really need to care about, to really a unified risk management

solution offering. So all of those product management teams that were

34

scattered everywhere, that cared only about their one thing, are now

under Cihan in this new process looking at the market and customer

needs and seeing all that we bring to bear at RMS to satisfy those needs in

interesting and innovative ways.

So you will see some of the results of that after the break. So, that's my

cue. We are going to take about 15 minutes, ask you to come back here,

and then you'll meet the team and see a little bit more about the products

and the platform. I think there's coffee and all that outside. Thank you.

CB: All right. Hi, everybody. Welcome back from the break. Before the

break, Karen introduced Risk Intelligence, and talked about why, why

we've chosen the platform approach. In the second half, I'm going to

begin by talking about this journey that we've started with Risk

Intelligence.

Now the platform we're building is going to let our customers take

advantage of the model IP we've built over the last 30 years, but it will

also let us serve the risk focused enterprise in ways that we've never been

able to reach before. To paint that full picture, to give you the grand tour,

I'm going to start talking about, first, what our customers deeply care

about and I'll talk about how Risk Intelligence addresses those needs.

I talked about this in my intro. I've spent the last decade and a little more

building platforms that transformed a few industries. Every industry has

slightly different dynamics, but the script of the transformation remains

somewhat the same. If you slow these scenes down, you see a slow

motion video where the established player is tied down with some legacy

technology, while the new kids on the block show up with some new

tools and tech that seemed to defy gravity and then transform the

industry. I'm sure many of you can guess what that gravity defying tool

is. That's data and insights. That's how they do it.

Our industry is very much the same. Karen talked about the risk life

cycle and how our customers care deeply about that. Our customers also

know that data and insights is the way to unlock greater profitability and

greater return on their capital. Now this is super important, but our

customers greatly struggled with this. If you ask them, they describe a

couple of key central ingredients to how to become smarter about risk.

Now the first key ingredient they talk about is models, risk models. This

is why we serve them today. Now, without risk models, you can't really

35

select the type of risks or price the type of risk you want to work on. So

it's essential for them.

Another key ingredient, raw material for them, is data. The external data

set is important for them to augment their decisions along the risk life

cycle as they work on optimising and transferring that risk. It is key to be

able to drive those decisions. However, finding high quality data with

great coverage is quite a difficult proposition today for them. So they

struggle.

Another key ingredient is data integration. Now, our customers not only

have to organise their own data, but they have to put their data right next

to the third-party data that they actually use to augment their decisions,

to generate insights out of that. But they struggle here too. They struggle

here because their data is stuck in systems that silo that data.

So that makes it very difficult to get to those insights. Putting all of that

data together isn't all that useful, unless you can ask it questions. Our

customers, not surprisingly, struggle here too. They're stuck in legacy

systems that won't let them ask sophisticated questions to that data.

Finally, getting insights out of that data is fantastic, but unless you can

turn them into action, that insight doesn't really produce an outcome.

Not surprisingly, our customers struggle here, too. It's hard for them to

take the data and insights and deliver that to the users just in the right

moment, so that they can turn it into action. I'm happy to say that the

Risk Intelligence platform we're building is designed to remove all of

these frictions.

We've got amazing scientific rigour around our models. This is where we

lead the market today. We've got some amazing cooks in the kitchen that

can develop the most sophisticated models in the world that will give our

customers the most detailed view of risk that they're looking to take on.

But we know that this alone is not going to be enough. The Risk

Intelligence platform also puts the rich RMS data set right next to our

models. This is where our customers can tap into a rich set of attributes

to be able to augment their decisions. For example, they'll do things like

look up a property's age or a company's cyber defence score to tweak the

price of the contract that they build for them. These are super important

components, to be able to fine tune your capital's ability.

Another important ingredient is data integration. Now it is greatly

important that the customers are able to put all of this data together in a

36

unified repository. We're building a unified depository called the Risk

Data Lake. I'll talk about this concept of a data lake in just one second,

but imagine this to be a unified repository for all types of risks data. It's

built for the physics of big data that you want to work on.

Another key ingredient is powerful analytics. This is where customers

can ask sophisticated questions to the data. They can do things like

accumulation calculations or portfolio roll-ups. I'll dig into this a lot

more and talk about how they actually use it. This is another essential

component and the Risk Intelligence platform brings that to the surface.

Finally, instead of hard-to-use point solutions and legacy systems, the

Risk Intelligence platform comes with a suite of integrated applications

that simplify the life of our customers at key decision points. In the risk-

focused enterprise, folks like risk analysts or portfolio managers,

primary underwriters or treaty underwriters make critical decisions

along the risk life cycle. The Risk Intelligence applications are designed

to deliver the insights to these users just in the right moments to turn

them into action.

This paints the picture of what the Risk Intelligence platform does, but

let me dig a little deeper into how it actually delivers some of this. We

talked about this before, I said to get smarter about risk, one of the key

essential ingredients that you have to have is data integration. Our

customers are stuck in these siloed systems. The ability to be able to

analyse all that information is key for them. There's a very simple

universal law of analytics that drives all of this. The reason to put all of

the data together is very, very important, is driven by this simple,

universal law of analytics. That law is: the more data you put together,

the more interesting the analytics become, the better the insights

become.

It's very easy to observe this, because the more data you put together, the

more enrichment you can do on top of that data, the more data you put

together, the more interesting the correlations you can discover among

that data become. These are super critical to them, but as I said, they

struggle greatly. For example, today many of them have claims

management systems or portfolio administration systems or risk

management tools, but all of these systems silo the data into isolated

repositories. If you're an enterprise user, to get the full picture, you'll

have to access multiple systems and you have to understand the different

ways in which they represent data and different ways in which they

format data.

37

The majority of the user's energy is spent constantly transforming and

translating information between these systems. That's a huge waste.

Instead, we're building a unified repository, a Risk Data Lake. For some

of you who may not be familiar with this term, ‘data lake’, it's an industry

standard term that's commonly used across the analytics industry to

signify a repository for all data to reside together. This is different than

the database technologies or the data warehouses and data marks you've

actually heard about in the past.

The Risk Data Lake that we're building is designed to securely bring

RMS's data right next to our customer's data. That could be things like

policies and claims that they have from the past, as well as third-party

data. This could be things like crime statistics on a given neighbourhood

or it could be demographic information about the occupants of a

property. Having a repository like this unlocks a lot of enrichment and a

lot of analytics capabilities for customers. Here's some of the examples

of what they could do with this.

For example, they could bring their past claims data into this RMS Risk

Data Lake and validate whether their predictions about the future is

holding true or they could restructure a contract to be able to get better

use of their capital. They may use RMS's Risk Intelligence, the location

intelligence APIs, Moe will talk about this in great detail, to be able to tap

into some very rich set of attributes. They may be able to find what a roof

type of a property is or what the foundation age of a property is and be

able to tweak the pricing of the contract that they want to do.

They can even do bigger enrichment here using the third party data. For

example, they could look at the claims behaviour of a given property by

looking at the education or the income level of the occupants. Or they

could look at the crime statistics in that neighbourhood and try to

understand if that changes the claim behaviour of a customer. These are

amazing, really powerful scenarios for our customers.

To be able to do analytics at this scale, you need a scalable set of

technologies. The Risk Data Lake that we're building is designed for the

physics of big data. It can deliver analytics just in time, fast, dynamically

at a low cost to our customers.

On top of the Risk Data Lake, we have the microservices layer. This is

another industry standard term and I'll explain what that is.

Microservices essentially represent powerful analytics computations that

customers can do and turn them into modular components. Karen talked

38

about modularity and openness. This is another place where customers

are actually able to plug into our platform. The modular components,

these analytics calculations that they can actually do, are easy to

integrate into their existing solutions, if this is where they want to tap

into the Risk Intelligence platform. Instead of ripping and replacing what

they have, they can easily pull some of these capabilities directly into

applications they have built in-house or they've bought from other third-

parties. We use the risk API layer to be able to do this.

Customers can do amazing analytics using some of these functionalities,

too. Imagine, for example, a customer of ours, an insurance company,

which acquires McDonald's with hundreds of locations. They can use

these functions, accumulations for example, to understand how their

risk profile has changed by acquiring this giant customer into their

portfolio. They could restructure these contracts and look for ways to

maybe transfer that risk. They can do what-if analysis to be able to

understand ways in which they could actually use their capital better.

These are extremely powerful analytics that give them great returns on

their investments.

Finally, on top of the Risk Data Lake and the microservices layer are the

Risk Intelligence applications. Remember, these are the applications that

deliver the important insights directly to the users that need them. We

can take those insights and turn them into action. Now the model

execution application that you see to the left is the one that we target at a

persona called ‘risk analysts’. This is one of the key personas. Our

existing software RiskLink actually serves this persona today. But all the

other applications that you see, the exposure management, primary

underwriting, treaty underwriting applications are net new applications

and net new users that we've never been able to serve before. Remember

that red box in the market segmentation that Karen showed with risk

analytics? These applications are targeting exactly that area.

Before I move on, I'll touch on one more thing. I talked about siloed

legacy applications, point solutions. The Risk Intelligence applications

are different. They're designed to operate on this unified view of data

over the Risk Data Lake. The way that actually shows up is that, for

example, an exposure calculation that you make in one of these

applications is immediately available to all the other applications in the

system as the user. If you're an enterprise user, you're operating on the

single view of risk across all of your departments. This is where the user

no longer has to translate data and move data between different systems

inside the company.

39

I'll actually pull on this thread a little more. There's another reason why

the platform is super powerful and it is its data interoperability

capabilities. I alluded to this quite a few times, but this user constantly

having to move data in between siloed systems is quite a bit of an energy

waste. It's not only the internal systems that they have to move data in

between, they also have to move data in between organisations. They

might do that to transfer risk, for example, between organisations to

another company. As the data moves between systems and between

organisations, there is this constant need to cleanse, rinse and repeat on

top of this data, that's a huge burden on the entire industry. There are a

few standards that they try to use to actually solve this problem. In fact,

because of our leadership in this industry, there's a standard called EDM,

Exposure Data Model, which we actually brought to market about 20

years ago, that is the most popular data standard to exchange

information between systems. But EDM and some of its less popular

friends have not aged well and no longer is able to serve the new needs

of the industry.

We've been working on RDO, which we believe is going to help solve

some of these problems. RDO addresses many of the pain points that

EDM and some of the other less popular versions suffer from. Where

EDM, the Exposure Data Model, is rigid and is property-centric, RDO is

flexible, is able to support many lines of businesses. Where EDMs are

lossy and incomplete, RDO is complete and lossless. RDO is an open

standard that we'd like to bring to the market. RDO is not just on paper.

It's actually a key ingredient of the Risk Intelligence platform.

Remember those Risk Intelligence applications exchanging data? RDO is

in fact the way in which they exchange information between all of these

applications, so the enterprise users get one unified view of their entire

data set at all times up to date, up to the minute information across every

department.

We've been working on RDO and we'd like to make it an open standard

for the entire industry to solve this problem, but we know that without

customer validation, without customers using these technologies, it's not

going to become real. We've just announced RDO as an open standard

back in May at our conference. But even though it's just recently been

announced, I'm super happy to say that we've got unprecedented

participation from some of the biggest influencers in our industry. You

can see the logo wall up there. They're participating in our steering

committee. The steering committee is the committee that will be

validating and tweaking the standard before it actually goes out to the

rest of the market.

40

I've told you quite a bit about what the Risk Intelligence platform does,

how it actually does it, but I'm sure many of you want to get a closer look

at what this thing looks like. I'll invite my good friend over, Moe, to give

you guys a closer look at Risk Intelligence.

MK: Thanks, Cihan.

CB: Thank you.

MK: I think I was supposed to go over there, but I apologise.

CB: I'll jump.

MK: Again, what I want to do is build on what Cihan and Karen mentioned, to

focus on a couple of key things. One, areas in terms of what we delivered

in terms of commitments that we said we would actually make good on,

as well as investments that we're making that are in line with the strategy

that Karen noted, how we get into those growth opportunities. As we do

this, I'm going to poke on both the applications business, which is new

for us, as well as on the platform building blocks. But before that, I think

one of the most important pieces of our architecture and one of its

differentiating capabilities is this modularity. It's not just a term. I'll show

you exactly how this spells into business value for our customers, for us,

through a couple of examples.

I think the first example I want to show is interesting for a couple of

reasons, because it ties a lot of things together. By the way, this will be a

theme that I go through, I'll highlight some of the applications, that we

call IQs, as well as, again, this is a logical architecture of it, but it's

highlighting how we're using these components in the platform to

actually bring these things into life. It will make more sense of how this

is accelerating our own internal roadmap for this.

The first example I'll show is interesting in the sense that it's going to

show you how easily we were able to use our existing crown jewels, the

data, the model science that we've been sitting on and unleashing that to

get into, again, going from selling into a cost centre and going into a

profit centre to attack that big $1.6 billion TAM of risk analytics

underwriting. You'll see how quickly we were able to do that, something

that wasn't on our roadmap, then something that's actually able to be

delivered here. The second, hopefully it will be obvious in terms of the

platform underneath, the team that we've put together, that's what's both

41

the applications user experiences, as well as the delivery of the core

experience itself.

Let me jump in and kind of set the stage for it a little bit. SiteIQ is a new

application. This is really for that risk decision maker, whether you're

the underwriter or whether you are the MGA or the cover holder that

they work with. What it does is it takes something that was a super

analogue, error prone, something that basically you're a quarter or two

behind in terms of the rules that you're setting, and trying to figure out

what your underwriters and MGAs are actually adhering to, and it turns

it into a digital process to where you can take action before it's too late.

I'll show you a couple of examples of this one, but this is actually one of

those ones where it's a lot easier to actually demo than to talk about it.

I'm actually gonna show you guys a live demo here and I think it'll help

the stage. With that, if we could…, excellent.

Sure enough, this is SiteIQ at rms.com. It is commercially available and

I'm just going to show you the experience here. I'm hoping that the Wifi

is good, because Paul showed me what would happen if this did not

work. Apparently we used to hang people out there. I'm just going to

punch in just a good use case and we'll build up from there, just to show

you the ease of use. I'm going to use our corporate headquarters,

Gateway Boulevard. By the way, while I'm using a US address, I'll show

you how this is global data, because our model data is global data from

all these years.

Anyway, I'm punching in the address. Super fast, what you're able to see,

and by the way, this experience runs on an iPhone. It runs on an iPad. Of

course I'm here, I'm showing it on a desktop. Instantly what I'm able to

do is get all kinds of information about this particular risk. I see

information from ESDB, a differentiated core data set that that tells you

exactly the attributes of this. We're showing some of it. You see a pretty

scary earthquake risk. You see FEMA, climate, we'll just gloss over that

one, temperature, but I want to show you a couple of things. Obviously

one, check out the performance. This is not just using Google Maps or

Bing Maps and trying to get the performance when you're putting these

overlays that no other system can handle when you actually start putting

flood data, wildfire data on top of that.

Let me not jump ahead. Let me actually show you, because that address

was wrong. That was actually putting it right in the middle of, the

geocoder put it right in the middle of the parking lot. Just by moving this

to get it more accurate, now suddenly you're able to see new risks that

42

are going to come into how you should think about this particular

exposure, just by jumping over the street. This is really highlighting a

couple of things. One, the value of why we're doing these HD models.

What happens when you actually have location specific information?

Imagine if this was a competition and it was zip-code level. You would

either be writing too much or too little. Again, this goes to the precision

of the model science, the data coming together.

One other thing I want to point out is obviously, these risks from first

party, from RMS, from third party, and in the future, as Cihan pointed

out, we're really bringing this data lake concept into all of these module

applications. It's going to be customer data actually coming into this as

well, because they may have information that nobody else has. It's the

first system of its kind that brings all of that together with these rules.

Here for instance, you can see for this location, we're giving very quick

basically red, yellow, green indicators for that underwriter, because time

is money. Here we're seeing a ‘refer’ based on the rules that were pushed

from the admin or this user. Go ahead and refer this risk as opposed to

just blindly accepting this. I'll come back to this to show how this is fully

customisable.

Let me actually just zoom out a little bit and just show you the

performance that, again, due to the platform. I'll try not to this time drag

the pin in the middle of the ocean, because bad things happen. Here I'm

going to Florida. As you can see, as fast as I'm able to move my mouse,

we're just retrieving information. The scale of this, when you think of the

number of concurrency, this is one single SaaS service that is able to

handle the load of pretty much the entire industry for it. It auto-scales.

Again, a testament of the platform, as well as the differentiated data that

RMS has been sitting on that now we're unlocking. Again, I can keep

going through this all day in terms of just moving the mouse around and

showing the various exposures.

Actually let me go back to this location itself and just show you a couple

of things. One, being able to tie additional sources. As a platform, are

you able to bring in these other applications that augment this? Here's

one example. Bringing in a total third party, a Google Maps into this one

to see, "Hey, is the visual information actually right about this particular

exposure?" Turns out here it is, so not too interesting. I'll jump back out.

Again, I mentioned this was a largely analogue process. This is

something where you may be one or two quarters behind, because you're

trying to do this over e-mail, Excel, etc., and it's just way too late. When

43

you've set a rule to say, right, no more business than X million off the

coast in Mexico, in London, Florida, etc.. Now, what we're able to do is

form a digital thread. This is really where the AI and ML come into play,

where when we're sitting in the middle and we know what kind of

information is coming in, what kind of actions the users are taking, from

what organisations, we're learning from this. What that means from a

customer perspective is productivity. You're getting the

recommendations, in due time, even automation for this to be able to

make it super, super productive for them.

I'll cancel out of this, but I promised being able to show one of the other

big differentiators that customers really get excited about is what

happens when you bring in, not just RMS data, which is remotely

interesting, not just public data, but when you bring your own. How do

you customise that view of risk? Because as Cihan pointed out, our

customers may be writing business where the actual demographics of

that location might matter a lot. Consumer spend, crime, etc., or

something completely different. We've made it fully configurable for

them to actually be able to write these rules and actually push that to

their clients. There's a back-end system, there's a front-end system that is

constantly learning to make the users more productive. Anyway, I won't

finish out with that one.

Of course the other types of tools, just being able to get the various

visualisations layer, being able to actually prove out the distance to coast

types of things, being able to synchronise portfolios. This is another

interesting part, where even though we're building these modular

applications, think of as a suite, that's what we're offering. You can buy

things modularly. If you see value in this and only this, fantastic, go

ahead and take it. But if you have Exposure Manager and you're actually

managing your portfolio there, it's this easy to actually bring those two

systems together. Again, the sum is more than the parts themselves.

We're actually attacking a couple of other really interesting things for

customers. Today, people in, actually in our system and everybody else's,

they have to go through, they have to log into various systems to be able

to handle event response capabilities, hurricanes coming, what's my

exposure on wildfire? Being able to bring soil maps, being able to bring

in demographics maps. It's as easy as just basically going through,

because, again, this is a SaaS service and we know what you're licensed

to, what you're entitled to. Quickly being able to bring those additional

layers, so they can make these decisions. Again, this is something like

everything else we’ve discussed, it's a platform within a platform. We've

44

architected this in a way that obviously scales with this one service that

handles the entire client base, as opposed to the old RMS way of it's a

point to point installation that's, again, costly, expensive to maintain and

so on.

Couple other things I just want to show. I can keep going with the

features, but I want to come back to one of the things that Karen pointed

out. In the old way of executing with RMS, we would have had the

concept, we would've coded this, we would have shipped out to the

customers and we would have likely been fairly far off in terms of what

their expectations were, what their workload was, and it would've been

expensive for both sides. When we actually started talking to a couple of

pilot customers as part of this process, because we're now designing

before we're writing a line of code.

As Karen pointed out, we may well go 60, 70 iterations, because of the

new muscle that we have inside of the company. We realise that instead

of one to a hundred locations for these underwriters, again, be it on

location, taking advantage of GPS, etc., or actually bringing in these

Excel workbooks, we designed something in a matter of weeks for them.

A super clean experience to let them not just go through the UI, but to fill

out this little CSV with an Excel that has their locations, drag and drop

that in and instantly be able to give them, with confidence, the view of

what did we import for them and be able to fly in internationally, look at

that content. Again, it's not a toy, being able to actually go through and

give them confidence of what we bring in, what do you want to filter, and

again, having all these systems connected.

It's a quick whirlwind approach of showing SiteIQ. Cihan, Paul, and

Karen know I can just keep going with this one. But again, I hope it

exemplified a couple of the key things that we're building on here. With

that, I'll stop the demo and we'll jump back into the content itself.

Oops. Let me know if it's synced up here. Perfect.

I do want to call out a couple of things. That was SiteIQ. Again, an

application that takes us into that rich underwriting space, taking

advantage of our past, our core business and dramatically leap-frogging

anything anybody else has. This was launched, it's commercially

available, again, as a SaaS solution. If you see value in this and only this,

fantastic, buy that. And of course it's powered by all the differentiated

content that we have that'll get richer and richer as we put this out,

45

everything from ESDB loss costs, financial information from the actual

hazards.

But I want to switch to something else. That was modularity at the

application layer. But what about the actual platform modularity, as

well? What if our customers wanted to build their own SiteIQ? Or more

likely said, "Actually, I have an underwriting desk. I'd love to use your

data. How do I integrate that inside of my workloads?"

Here, too, is a great example. Location Intelligence, as Cihan pointed

out, this is our offering for developers to be able to get access to really

the crown jewels from RMS. It's a super high-scale service that is no

longer buried inside of each tenant of the installations. Think of you

wouldn't build a Google for every single customer. You would build one

super high-scale service to where you can manage that profitably and

handle the scale of the world. That's how we're approaching these things

to where they're auto-scale. What it's able to do is it delivers, obviously

it's global information, but just trillions of data points, everything you

would want to know about these various exposures around the world to

be able to do, whether it's data science, whether to imagine completely

new experiences or to augment things that you already have.

This is Location Intelligence, we call it 2.0, high-scale service. Again, this

is another thing that launched just a few weeks ago. It is available for

customers to use as yet another modular application, whether you're

building a consumer application or again, as part of insurance,

reinsurance workloads.

I want to switch back a little bit in terms of, so a couple of key things

here. We had proof points of the modularity of the applications, on the

platform itself, and delivering what we said we would deliver. I want to

switch into what about our core business, model IP delivery. Here I'm

really excited that we at Exceedance, as Karen pointed out, we showed

another major proof point. We took the US flood model as well as the US

storm surge model and implemented those natively on top of the

platform and made it alongside of all the RiskLink models that we have.

One converged platform the customers can get that delivers all of the

core differentiators of the model, but we didn't just stop there. I think

that's another thing that we would have done in the past of saying, "Oh,

here's the IP. Good luck with it." We took out, as part of the application

that brings these things together, all of the pain points that were largely

manual, whether you're using RiskLink or really any of the others out

46

there. This is a differentiated way to, for instance, how do you want to

run scenarios and be able to do what-if with confidence? You can't do

that in other tools without having non-destructive edits with audit to be

able to know what what department did and being able to track that

throughout the system.

This is one of our first. We made it really easy to be able to get at those

insights to where you don't have to write a bunch of SQL code and get

other people involved. The main things that are important for customers

just pop up. These are gleams. The only way we could do that, as Cihan

pointed out, is we've architected a system underneath that truly no other

solution can do, because this is not a typical SQL server, client server

application that's trying to scale. It's a very different thing that you build

there. As well as some of the other really big differentiators on this work,

we talked about breaking down walls, letting customers access this data

using their third-party tools, which was ...

Customers access this data using their third-party tools, which was

extremely sought after for a number of years that RMS hadn't attacked,

and actually brought in things from the timelines, from the roadmap. We

now say, "When we give a date you can stand behind it." And we have...

with standard deviation of plus or minus quarters depending on that.

And it always used to be plus, all right, here's the date and it was plus.

Now we're actually able to show how we're accelerating things while

delivering things that weren't even on the roadmap while still addressing

the commitments. And it's a testament to the team. It's a testament to the

platform, and how we're working together as models and software and

product just as one group.

Now... Great. That was a lot of functionality inside of it. As Karen pointed

out, what about price performance? And here's another great proof

point. While we're rebuilding the engine as we're flying the plane, what

I'm showing here is running 70 million locations, which, obviously

RiskLink and the other solutions in the market were just kind of bar fat,

right? In terms of being able to handle that kind of load. But the same job

being able to run in 18 hours, we're now able to do... you can do the

math, it's almost five-X faster, right? So a dramatic boost in performance,

great. Our customers are more productive, they're able to write more

business in terms of what they've been able to do. Pretty much a first.

That's performance. What about cost? Cost, we also attacked. And so

what this is showing is... Think of these little logical boxes of... These are

cloud compute boxes, right? So from a customer's standpoint, how much

47

money do I have to invest to be able to get that kind of performance? And

so in that latest release of the Risk Intelligence, we're able to show that

you can get this kind of performance at a third of the cost, which...

Again, four by three, this is an order of magnitude of price performance.

So IP delivered, the application delivered, the productivity delivered, as

well as the price performance. And so we're just making good on the

roadmap. And again this is how, as Karen pointed to, we're winning the

customers back by showing them running the demos... Because of

course slides are cheap, right? This is something that the customers are

validating with us to really get that confidence.

Now that was our HD models. Right? And by the way, something I want

to point out on here, there's nothing flood-specific on this. These are

deep platform investments for everything I mentioned for any of the HD

models that are going to be running on top of the platform. So, as we talk

about wildfire, EU flood, and some of the other concepts on top of that,

they're just going to get these benefits, and we've just started in terms of

the optimisations for these. So that was great. That was the HD models

and again, how we've delivered on the whole converged platform. But as

we started thinking about, "What about the 400 other models that we

have that are based on RiskLink?" RiskLink's here to stay, right? We want

to be able to support our customers that are using this on-prem.

But I touched on how we want to modernise these as well. And so we

really challenged ourselves hard, as we started talking to customers in...

Through our own hosting business, we started living the pain of what

customers were going through, things that hadn't really been addressed.

And so what this is showing... Ignore the details on this, but we really

cracked open the architecture. And no surprise, RiskLink, like all the

other competition tools that were around this time frame... These are

legacy, right? They're showing their age in terms of the components that

they've picked. And if you wanted to rebuild RiskLink and some of these

other solutions today, knowing what you know about the tools that are

available, you would build it in a very, very different way. And so we had

a pretty interesting, honest set of questions on this.

We attacked all the customer pain points. So today if you want to run a

V18 model and a V17, you have to have separate clusters, which is time

and money, from a customer standpoint. Its total cost of ownership hits.

You have to actually... Then finally, once you have the clusters, you have

to install the software, right? And because these things aren't automated,

again, it's time, it's money, it's the maintenance for it. But what if you

wanted to have account-modelling capabilities, in addition to being able

48

to run the actual core analytics? Today you can see the other solutions.

They're all fragmented experiences that don't quite work well together.

We started looking at this in terms of, "Can those actually be converged

as well? Instead of having... Here, the complexity would explode.

Different applications, different clusters, different versions. You can see

the math in terms of what the customers would be facing. And of course

some of the other components underneath.

So because they weren't highly available, people weren't thinking with a

cloud-first mindset, right? If when a component went down, or when you

have performance-bottlenecks, that basically means they're not writing

business. So again, just attacking the business value in every part of this.

And I'm really happy to say that we took that challenge and what the

team did was we pretty much reimagined RiskLink and Risk Browser.

And I'm showing you guys a very early preview of something that we've

been working on that takes... It's pretty much a first. Imagine not having

to install anything, not having to deploy anything, not having to harden

anything. It doesn't matter how many versions of models you want to be

able to have side by side. You have one scalable SaaS service that is able

to take those, bring the best of account model and capabilities of our

core products that people are buying, bring that with RiskLink, and

instantly make them productive on top of this. And it's built natively on

top of the Risk Intelligence platform.

So it serves a couple of things. One, we're using our legacy as an

advantage. Customers are able to say, "The things I used to be able to do

20 years from now, I'm able to do in this, on top of the platform that gives

me access to all the applications that Cihan pointed out." But I'm able to

do this with total cost of ownership improvements, performance

improvements, maintenance improvements, and as I'll show, all of the

feedback that we were getting around these gap closures, right, "You

don't have x," "You don't have y, " we've actually addressed.

So really quick whirlwind demo of this, of how this is not... We say, "It's

not your dad's RiskLink on this," in the sense that it's portfolios, it's

accounts, it's point editing, it's bulk editing, it's SQL data access, it's bulk

data access... Everything that people were expecting from the old

platform that we were kind of holding hostage, we just basically broke

down the walls. So if this has value for customers, they can instantly use

this and only this, but as a side effect of saving money, being more

productive on top of the platform, they're actually able to jump on top of

the Risk Intelligence platform to then automatically be able to, for

49

instance, get access to SiteIQ without having to re-import all of those

things. To be able to write these losses, to be able to get to exposure

managers.

So this is something we're very, very excited by, and I just can't help with

the show this... Without even focusing on the improvements in terms of

performance, right? Because we wanted to do this as both a productivity

enabler as well as the total cost of ownership enabler, so the customers

can look at our products and go, "I'm getting more value out of this." But

as a nice side effect of how we've re-architected this... I point out NAHU

[North Atlantic Hurricane], but it's really pretty much every RiskLink

model that we're running through this, we're seeing dramatic gains,

which means they're able to write more business, they actually see RMS

as a much more powerful tool in terms of how they do their business.

And we're just getting started on this.

So a quick preview of this... And this is something that we're going to be

offering, again, as a beta, as a preview for customers in our Q4

timeframe. So that was RiskLink as a service.

I want to switch gears as well in terms of talking about a couple of the

other applications that we're building. And again, we're going from proof

points in terms of things that customers were waiting for, delighters to...

Again, really making people more productive in these personas. So

exposure management's the next area. ExposureIQ is effectively the

next-generation exposure management tool from RMS. And I should

point out another thing. As Cihan pointed out, again, this was something

that was buried inside of the risk modeler application where... Again, as

a customer, it was trying be everything to everyone. So what we're doing

is having this in a couple interesting passes. In the Q3 timeframe, we're

going to be shifting this as yet another modular application on top of the

platform that customers can use immediately and be productive.

And what it's going to do is it's not going to just be able to do, "what are

your hotspots," "What are your cold spots," in terms of your portfolio, but

it's going to let you take actions on top of those because again, we're able

to do things a lot faster that you just couldn't do before. And immediately

this, like the other IQs, gets its own cadence and its own shift vehicle.

Because we've separated the execution of our platform teams in our

applications, it's no longer this sequential execution that's been plaguing

RMS for the past four or five years. We're actually able to see, hopefully,

these are proof points, we're doing things in parallel. And so Q3 is going

to have not only these capabilities, but with some of the highly sought

50

after event response capabilities, of, "There's an event, how do I make

sure that I take that event's information and immediately bring it into my

portfolio to see what's happening?" Again, another first, another easy

delighter based on how we built this.

And in 2020, really going after some of the interesting cases around how

do they do clash? These really high-value use cases that are actually

enabled by the RDO model because they're actually now able to do things

that are... Lines of businesses that you can't represent in EDM. So you

can see the strategies coming together. The applications can be used in

isolation, but they get better as you have more and more of these things.

But you don't have to buy everything all at once to get value. And I, I

would say finally and again... Sorry, and one other thing on ExposureIQ,

again, the dates are there another application that's coming out, and

again, we're at another release cadence. It gets better and better over

time.

And finally, another target persona for us... Because we're realising these

are different people. As Cihan pointed out, when we say "persona driven

experiences," that means we deeply understand who we're designing for

and what makes them productive. That's simply what it means. And so

we understand the treaty underwriting use cases that were, again, buried

inside of the risk modeller application, is a very different persona to

attack. These are people that time is money. They want to make very,

very fast decisions. Latency matters, not deep analytics and... Been

writing a bunch of SQL and trying to program things. And so again,

another persona here. And a little less details on this one because I'm

sure the competition is also realising, listening, and in terms of what our

core differentiators are here, but I'll touch on... This is a very

differentiated application for us.

We're really looking at what is... What are the pain points for that treaty

underwriter when you have thousands of these reinsurance programs

that you're looking at? And instead of doing... I would say what... I don't

want to talk about the competition, right? But in terms of... It's a very,

very different way of attacking this that is highly intuitive, very visual,

yet takes advantage of the scale that we've delivered as part of our

applications to really have pretty breakthrough experiences. And so

TreatyIQ is another one. We're going to have a preview of this in Q4,

commercial availability in Q1, and and again yet another example of

these module applications coming together.

51

So with that I just want to close on hopefully a couple of things that we've

illustrated through this. Modularity, being able to deliver maximum

value to our customers in terms of what they see value in, making sure

they work better together as part of the suite, hopefully showing you that

we've fixed our execution muscles, even in this short timeframe, of being

able to do more than what we committed to but with a higher level of

quality, and actually adding capabilities and new businesses that we

weren't even selling to from cost centre, the growth centres. And again,

this notion of modularity is not just a thing for customers, but how it's

accelerating our own development of things that are not on our roadmap

yet, as we're not showing. So with that, thank you, and I will pass it off to

Mohsen to talk about our models. Thank you.

I'll steal that back.

I'll leave...

MR: Thanks Cihan and Karen for a really wonderful presentation giving really

lots of excitement about the technology. So what I really want to do, what

we have built over the last 30 years. If you take a look at this map, this

map shows RMS’s global model offering by peril, by region. In the last

two years, the development team at RMS made a significant progress to

building new models, updating existing models and also looking at the

new geography, new region, and new peril model as well as looking at

the emerging markets. The recent events in the last two years,

supplemented with the loss of data recently see demand for more

accurate modeling. Models need to be more accurate in order for clients

to take advantage of that. So as Moe demonstrated here, the model has a

really high computational power needed in order to get the right results.

So in the last two years, we’ve developed high-definition HD models for

US Flood, Wildfire model, which I'm going to talk about the briefly the

next few slides. At the same time, we developed a Euro Flood model

within 15 countries. We completed Japan Typhoon, Japan Earthquake, as

well as the New Zealand Earthquake. These are very complicated models

and required a fair amount of computational power. Legacy type of the

application is not able to really provide a solution.

So if you look at it actually in the last two years... And Karen mentioned

Atlantic Hurricane 2017 and ‘18 caused $185 billion loss. And if you look

at actually two model, Atlantic Hurricane and Wildfire contributed more

than... I can say 7 events more than 95% of those $185 billion loss. So if

you look at actually Hurricane Harvey, Irma, and Maria in 2017

52

contributed $92 billion loss. 2018 Hurricane Florence and Michael

contributed $16 billion, and then we have a wildfire in California

combined created $27 billion loss. All of this... actually for our customer

is really important to understand exactly what they need to really

provide that. So we need the model can really holistically provided the

right view, and that's why the average model in a market is not going to

be supported. The growth for the client and for RMS. That's why I'm

going to walk through the video briefly about the two models to see why

we need these technologies in order to get us to the next level.

US Flood model was one of the more complex and ambitious models we

ever built at RMS. And this model, from hazard perspective, provide a

comprehensive... A stochastic type of the event, including the tail event,

like Hurricane Harvey, is included to these type of events. You can

capture the result very systematically. And I step back and look at, "Okay,

well, what the clients they need to have regarding the information from

vulnerability?" They need to know basement, they need to know first

floor elevation, they need to know foundation, where the levies are,

where the flood defences are. These are all very complicated

information, which they needed to get to the model in order to build it.

So to achieve that goal in a vulnerability perspective, that's why we need

such a sophisticated application to be able to do that. And now we are

building a model, bringing it to the market. And right now it's available

in 2019, and it's farther differentiate us versus all other model in the

market. There are some model in market, maybe they can use it. But

future is going to be very different. Flood is a high-grading hazard and

required a very systematic approach to do that and quantify it. So in

order for us to grow, we need to really have the right technology.

How much government can borrow money to support residential losses?

At one point FEMA is going to tell, "Okay well, private insurance market

is opportunity for them to start to grow." That's why it's important for us.

However, in the commercial industrial on the client, they are running

the Flood model, but because they don't really know application is not

there, they put a limit on deductible. They don't really control it, and

that's why it's very important for us to provide a sophisticated model that

can really provide at the location level, not aggregate, at the location

level the solution needs to be provided. That's why the RMS model, last

two weeks ago, won the Trading Risk Award for the best Flood model in

the market.

53

So if you look at... The next one is a Wildfire model. It's another high-

grading hazard model. There is no reliable... I'm repeating. There is no

reliable model in the market that can provide you very holistically view

of the risk that can be used systematically. Very calibrated, with 2017 and

’18, we have spent a significant amount of time to build this model. We

look at all elements that contributed the fire, from embers, from smoke,

from all the aspects of the science. And that's why computationally, it's

very important to have the right technology in order to be able to do that.

So if you take a look at this slide, right? So if I... This is the reflection of

2017. On the dot point here shows the fire started. So a key innovation... I

stood back and look at, "Okay, science matters. Why science is

important?" Right? You can build a modern and average way. But if you

take a look at... We have a direction of the wind, we look at the wind

speed, every single dimension is coming. We look at how this is going to

be impacted. All the red dot point in this slide shows ember flight by

wind speed, and after 90 minutes, thousands of the building got to the

fire. That's why we need to have the right technology and right model to

really differentiate. The word "differentiate"... There are average model

in the market. But what is important for us to really build a model that

can support the technology moving forward, helping our client to grow.

So the other element is climate change. If you look at this, a slight

climate change, you can see more frequent flood, you can see severe

hurricane, you can see drought, cyclone, typhoon, on and on. And

yesterday you saw a tremendous amount of hail in Mexico, and there is

enormous amount of changes is really happening. So climate change is

really important. And that's why clients to reach out to us to look at this

problem systematically. We are scientists, we build the model, we

provide the solution to the client. The last 30 years we build sophisticated

models. And we pushing boundary, we're moving to next level. And that's

why technology is important for us to really have that dimension.

So we did three decades of building sophisticated models over the years.

We are engaging with the client to attack his problem. And besides the

climate change, the other element which is really important is exposure.

Exposure is on the coastal region, they are at the element of this risk, due

to the sea level rise. And clients, they need to understand that, and how

to address that. "Do I need to worry about the next five years? Do I need

to worry about the next 10 years? How do I really address a question?"

And that's why it's very important for us to really look at that one

holistically.

54

So therefore the requirement for models is continuously changing. And

we need to build a model. And that's why in the future there is not only

risk... And I can say insurance market needs to have a model. This is

going to be expanded in a new area: bank chain investment. They want

to look at, "Exactly how can I manage my mortgage to make sure they are

protected against the flood?" Corporations, they want to really

understand where they can optimise their operation in terms of

providing supply chain with the amount of raw material is coming. And

also in terms of the countries and cities. They want to provide the

protection and also mitigation.

Now I'm talking about the next slide about taking it to the future. In the

last few decades we provided models. And clients, they use it and we

provide that help for them. But what's important for us, you can see the

landscape up there is continuously changing, and we need to really

provide and use the state-of-the-art technology to optimise the operation

because this is really... Again the question is not only the model, it's the

operation of the insurance process that's really need to be... Really look

at it very systematically. And that's why we are really building that

foundation, as my colleagues talk about it, and is going to really provide

that in future of the risks. Not about only model, about everything else

needed to be there in order to quantify the risk systematically.

So if I take a look at this slide, I call it the real risk platform. I'm

emphasising the word "real" again. It has to be scalable performance-

wise, use the right technology supported by powerful application that

Moe demonstrated, data and analytics that Karen, Moe, and Cihan talked

about. It takes us to the next level. We are really modernising the risk-

management operation by increasing efficiency. Increasing efficiency.

You saw the... Performance-wise. You look at the technology, we bring

highly sophisticated models much cheaper, much faster, and that

actually include the accuracy, and so on. So the benefit our clients,

they're going to get, and market is going to grow with us, it's not only

about the models. It's about the technology combined together.

So that's why it's very exciting for us to create a foundation that actually

the entire industry is going to be using in the future. So let's talk about

future. I'm going to hand over to my colleague, Robert, who will talk

about the other future risks. Thank you.

RM-W: Thank you everybody.

55

So catastrophe-modelling is really the original InsurTech. So, it was the

first wave of InsurTech was really catastrophe modelling when it came

30 years ago. It helped disrupt markets. It created the Bermuda

reinsurance market, it created the ILS market, and every time we have

this situation where models give confidence to how you can establish a

new market, that market grows, it requires more application of

modelling. So, we have great interest in actually how models can drive

markets, whether we are focused on earthquake, flood or wildfire. But

we also see high quality RMS models have brought confidence not only

in natural catastrophe risk and transfer, but also in terrorism risk, and

most recently in cyber insurance, too. So markets need confidence, and

models provide that confidence. And now we have new hedging

instruments for offsetting some of the potential costs of future climate

change, or even measuring the benefits of alternative interventions,

adaptation around climate change. So for all of this, they need

catastrophe modelling.

I've got to give a talk in Singapore at the start of August, which is going to

be on 10 applications of catastrophe modelling beyond insurance. It's

really opening up right now. A key driver of additional demand for

modelling comes from the ambition to reduce what is called the

protection gap, as Karen mentioned earlier. The protection gap is the

difference between the actual costs of catastrophes, and that part which

is insured. According to Swiss Re, over the past 10 years, the protection

gap has been running at an average of $130 billion a year. Now, if that

protection gap was filled, so if insurance could fill the protection gap, the

size of the catastrophe modelling market would be three times as big as it

currently is. Just to give you a sense of of how important that is... If you

like, the protection gap is missing claims payments to speed recovery.

It's lost premium for insurers and the lost demand for modelling which

will be needed to support writing the business.

Now people talk about a single protection gap, but I would argue there

are really three distinct protection gaps each requiring a different

diagnosis and range of interventions. First, there is the hyper... The high-

risk protection gap that we find in wealthy countries like the US, when

insurers have deliberately avoided writing too much of the risk because

they fear how to manage and reinsure it. We have the example of

California earthquake, where only 10% of homeowners have earthquake

insurance. Then we have the emerging markets protection gap, not only

found in developing countries, it also exists in Italy, for example, but one

could simply wait for a middle class to emerge who will buy standard

insurance products. But much can be done to offer micro-insurance or

56

sovereign risk-transfer, all of which will require risk models. The third

protection gap concerns intangible risks.

In the 1970s, S&P analysed the companies in the S&P 500, and found 85%

of their value came from physical exposures from factories, ports, power

plants, pipelines, refineries. By 2015 they did the same exercise again

and they found that the proportion had reversed. Today, almost 90% of

the value of S&P 500 companies is intangible. Intangible includes IP,

reputation, brand value, business continuity, customer information.

Think about Facebook and how much of its value is in hardcore

factories, for example, not very much. The... All much more difficult for

insurers to devise products to cover. Like California earthquake, US flood

manifests a high protection gap. Even in the mandatory for a mortgage,

one in 100 year coastal flood zone, 30% of homeowners don't have flood

insurance. The gap rises to 60% of the risk cost across the whole US, and

70%, higher even than average, for the 160,000 properties flooded by

Harvey in Houston. Around 97% of homeowners' flood insurance is

currently provided by a federal scheme that is continually running out of

money and has had to borrow $25 billion from the US government.

Plans to prioritise and risk-rate the market are all dependent on the

availability of reliable high-resolution risk modelling. The newly

released 2018 RMS US Flood HD model has the potential to enable this

transformation. Then there is the Emerging Markets Protection Gap, for

which models are critical to expanding insurance solutions and creating

new forms of risk-transfer. Recent releases by RMS of Earthquake and

Flood models in India, Earthquake and Typhoon models in Philippines,

and Earthquake models in Indonesia, are all helping reduce the

Emerging Markets Protection Gap. And then we come to the most elusive

of the protection gaps: intangible risks to corporates.

As one component of the gap, large commercial losses have been

shifting away from physical damage to business interruption. 23% of the

claim in the early 2000s was business interruption. Transforming to 58%

of the claim in 2016 and 2017. increasingly, insurers are being asked to

write non damaged business interruption covers triggered by

interruptions in supplies of oil or electricity, or automotive supplies, or

cloud computing. And that will require in every case modeling solutions.

Among the most significant intangible risks are interruptions to global

supply chains. Recall in autumn 2011 when a series of vast industrial

parks built in the Chao Phraya floodplain north of Bangkok were all

flooded in the same incident which lasted for up to nine weeks. The

flooding took out some of the principle manufacturers of hard drives

57

among other industry sectors. Western digital had production

interrupted for six months. Hard disk prices spiked and shortages put a

big dent in the profit of global PC chip and memory companies.

No one has ever added up the total cost for businesses that ran into tens

of billions of dollars. One of the key trends for our modelling agenda

concerns the way in which catastrophes are becoming increasingly

manufactured. To avoid the wind blowing down the overhead wires in

lower Manhattan, they ran the cables underground. Along came

superstorm Sandy in 2012. when the floodwaters short-circuited the

underground cables and caused long lasting blackouts with significant

human and financial consequences.

With our increasing dependency on electricity, extensive long lasting

blackouts whether driven by a hurricane, a solar storm or a cyber attack,

will be increasingly disruptive and expensive. Another risk modelling

opportunity. Increasingly we find disasters to be human triggered, in

which the insurance claims switched from Acts of God falling on

property insurance classes, to Acts Of Man with the potential to be

picked up on the casualty liability policies. This leads to questions about

the risk to liability insurance covers of a utility whose equipment

triggered a devastating wildfire. A situation that has happened a few

times since the year 2000 in California.

Only now with the 2018 Camp Fire, the costs of around $12 billion have

led to the utility, Pacific Gas and Electric, filing for bankruptcy. In the

future, the utilities will be using our wildfire model to highlight where

they have the greatest concentration of risk, and the greatest need for

equipment hardening. Another situation of this kind concerns

earthquakes triggered by the deep disposal of wastewater from fracking

or oil extraction in Oklahoma. So far, this activity has triggered

earthquakes up to magnitude 5.8, causing significant damage to some

rural towns. A number of lawsuits are working to move that cost back to

the liability coverages of the waste water disposal companies. Now

scientists from the US Geological Survey have identified that some of the

largest earthquakes that affected Southern California in the early

decades of the 20th century were likely triggered by oil production.

Many of these fields are still operating. After the next earthquake in

Southern California, liability insurance coverage could be forced to pay

out for some of the uninsured earthquake claims. Here there is another

risk modelling agenda for us to pursue. Over the past decade, we have

seen what we can call blue chip catastrophes, calls to some of the largest

58

companies in the world taking deliberate shortcuts that don't work out.

Whether it was BP choosing not to test the concrete lining of their

Macondo well, or VW writing cheat software that knew when an

emissions test was being performed, or Boeing writing software to

attempt to correct the problems introduced by fitting larger diameter,

more economical engines on the low slung 737 air frame. Risk continues

to expand and with it the opportunity for additional areas of risk

modelling. Thank you.

KW: I'm going to wrap us up here in just a few moments. We'll take a short

break and then open up for your questions. Every time I look at Robert's

numbers I'm astounded. When you look at the billions of dollars of these

blue chip catastrophes and what's at stake, and it just hearkens me back

to one of my first points, which is sitting in these board rooms, what we

talk about is risk, risk and risk. So we're excited about bringing the

world's risk platform to market and expanding our business. So a lot of

questions I'm sure you have are around our investments and support of

future growth. What are we doing? I am not here to give forward looking

guidance on 2020 or 2021, and precisely how many dollars and cents, in

which years in which margins. But I do want to give you broad strokes

about how we're thinking about the business over the last three years.

I see you all smiling. Thinking of the follow-up questions you're going to

ask me about that. Good for you. When I came into the business, the first

thing I did, I took a look backward before we can take a look forward.

And what you're looking at here in the green bars is, if you've reviewed

RMS numbers of late, those were the reported adjusted profit operating

margins as were reported. As you'll recall for some period of time RMS

was capitalising as research and development expense, and you see in

the green bars the results of that capitalisation and amortisation. I'll

show you a more detailed slide in just a sec. But the first thing I do,

because I think capitalisation of software development in R&D is the

devil, is take that out. Because we iterate our products and we update our

products so frequently in the software business, that I have not in my

businesses capitalised software.

So I like to look at it, and I always keep my eye on the cash line and

money out, out the door is money out the door, right? And so the blue

bars represent the adjustment in margin. If you account for taking out

the R&D capitalisation, and you account for taking out the subsequent

amortisation of those costs, right? And so this, so that you don't say,

"Where's her math and what's going on here?" is in your kits to just show

you what I like to look at. Cash in, cash out, what are we spending, what

59

are we getting? So what you can see is that on average from 2010 to 2018,

the margin has ranged from 29.5% to as little as 1% on this adjusted

basis. Taking out again that capitalisation, that amortisation. And the

average margin, I call that the pro-forma, of that period of time for eight

years was about 17.6 or about 18%. You see the margins were higher in

the first three years, the margins got lower and the next years as those

investments were made and so on. But I think it's good to level set us in

the past before we talk about the future and the investments that we're

making.

So we've not shown you this before, and I have some explaining to do

here, but in terms of just a broad stroke breakdown of our expenses, like

most companies that Robert just talked about, whose assets are

intangibles and intellectual property such as ours, you won't be surprised

to find that our head count costs, and our talent costs are the vast

proportion of our spend. This will map to other technology companies

that you follow. About 74% of our expenses are employee and employee

related. The way we break that down, and this is for the full year 2019

forecast, is that you see an R&D expense about half. R&D isn't just new

product research and development, so let me help you understand what's

in that budget.

All the models science, new and in the future, updated and upgraded. All

the support of our portfolio of those 400 models that our customers use

every day in terms of updates and upgrades, which happened quite

frequently, are covered in that cost. All the new things that you saw in

terms of the Risk Intelligence platform (RI), the Risk Data Objects, some

of the facilitating technologies that we've been building, that we

accelerated our development on. Because we were as we like to say,

changing the engine while flying the plane, sunsetting our RMS(one)

product while we were bringing RI to the market.

All the applications you saw being developed and the constant updates

and quarterly upgrades of our software, our on premise RiskLink

platform as well as our cloud stuff, old and new. On top of that is the

whole product function. That sort of secret sauce we were talking about,

that we consolidated under one team. So the folks that go out in the

world and figure out the underlying problems and challenges that the

market faces, so that we can come up with product and technology and

interesting new solutions to meet those challenges and opportunities that

the market faces, although they're not writing code and building models,

that is included in our R&D budget as well.

60

Mostly people costs, with some development environments that are

pretty expensive for us to run as well. We have very reasonable sales and

marketing costs, because our product is quite sticky and our market is

quite targeted today. Then our cogs [cost of goods sold] again are very

reasonable. Our gross margin is nice for a business like we are.

Remember we have the analytical services business, an eight figure

business out of India. So we have people behind the services that we

provide that drives some of the cogs, and we have cloud services that

drives the costs of goods sold there, and our gross margin. So we keep a

close eye on gross margins. Ours are quite nice for a business of our size

and shape. You'll see companies like Salesforce and others in the sixties

or seventies, this is considered very nice margins for a comparative basis

of our business, something I keep an eye on.

So what's going to drive profitable growth at RMS and what's the

timeline? I will say it again and probably again in our Q&A, I’m not here

to provide forward looking guidance. But here to give you broad strokes

on what to expect, and what are the knobs and dials that we're having

our eyes on right now and playing around with. So what are our growth

opportunities in our current markets? Well, one, we can grow share in

our current model market. Having lost some credibility with the lack of

the platform deliverable, we earned ourselves out of some accounts and

we lost some credibility where customers were waiting and seeing before

making more investments in us on our core model products. They

thought we might get too distracted by the platform, and so I think

there's a nice opportunity to grow share in the current model market.

That said, in that market, we have a head room problem. It is about a 500

million dollar market. It is not the $7 billion dollar market. So, but we do

think near term we can grow some share there. Secondly, we've never

had a real go to market in these other sectors. So we barely call up. We

basically answer the phone when a utility calls us, or when a government

agency calls us, and capital markets we certainly go to market with, but

expect us to get a little bit bigger there. So the revenue from other sectors

with the products that we have today, you can expect that to grow. But

notice the pie slice isn't growing that big. Our primary focus will remain

the insurance sector in the mid-term, with some forays into these other

markets. You can imagine that if you are a private bank and have a multi-

billion dollar portfolio of mortgages in California along the San Andreas

fault, that you don't resell into the government, that you might care

about the liability you're carrying since only 10% of your customers have

earthquake insurance.

61

I have earthquake insurance, but I will not let them run SiteIQ on my

house because I know I am smack on the San Andreas fault. I don't want

to know. So, the other thing we talked a lot about, which is the broader

growth opportunity for RMS, and it's simply this. To leverage our

position in our core business to enter into more robustly the data market,

in a very pinpointed sniper shot way. To enter the risk management

analytics market followed by the insurance analytics market. And we're

not going to talk about which emerging risks or new insurance lines

we're really thinking about going after. Robert alluded to some of the

thought leadership we have there and some things that are of interest to

our clients, but we did a lot of telling and not a lot of showing in the past.

So we're going to keep some things close to the chest, both competitively

and until we're closer to market. So talking of a few points to wrap up.

New products, we'll continue to launch. We've talked about three new

high definition location level based models, that we brought to market

this year. We had 10 additional model updates that we already

announced in April. We brought our first three analytics products to

market. One is available today, the other two are available later this year.

SiteIQ, ExposureIQ and TreatyIQ. The new data products underneath of

this, from geo coding to hazard layer, to exposure, have all been

enhanced. All the way through our lost cost data which you saw

surfacing through our location intelligence API, which was being used by

our SiteIQ applications. So you can begin to see how all this is coming

together in our data products. Finally, new RiskLink as a service, which

will be a very important strategic move for RMS.

Bringing all those model analytics that we've had on premise into the

cloud in what we call a Preview, also known as Beta software in this year,

in the fourth quarter. And then available commercially to the market in

2020. Now it's important that we will continue to support our on premise

offering. So when we push out updates to RiskLink, we will do it equally

and on the same day to the on premise as well as to the cloud solutions,

because customers want choice. A smaller group of customers are

always the early adopters and then it moves into other markets, and

they've self-identified, I believe. Believe me when they've told me, "I

want to be first, second or third to the platform." So customer choice is

key.

So, RMS development product life cycles. Just to give you an idea of

timing and what the drivers of timing of ultimate revenue and

profitability are, in these new businesses is that a new model can be in

development for one to four years, before we bring it to market.

62

However, on the new data products and the new software products, we

can bring a product to market in 6-12 months because of the of the

platform that you saw, we're able to very rapidly bring these things to

market. And then every quarter after that we iterate new features and

new functions. So SiteIQ version one, for the customer that doesn't want

to customise it with their own data, is ready here and now for those

customers. If you want to customise it with your own data, then that's a

future version of what we call a ‘fast follow’ three months later, six

months later, and so on. So we open up broader parts of the market as

we continue to build out these offerings. You're familiar with that, with

your consumer offerings, whenever you get on your iTunes or your

Amazon, you see new things showing up all the time and we tend to

iterate every quarter.

So that'll help you understand the cadence of us bringing products to

market. Then so what that means for us is that our investments in both

R&D, that big bucket of R&D, as well as our go to market, to expand our

go to market will continue. So finally, on premise to cloud transitions, if

you follow this in any of the companies that you follow, companies

who've been traditional on prime business, moving to the cloud typically

takes three to four years to complete the transition. Now sometimes

these companies, especially those that have a perpetual license model,

will do one of these with revenue and profits [draws a U-shaped curve in

the air]. They'll do a ‘U’. The reason why is they're not like us, they're not

subscription based. So let's say it's the last day of a quarter and I book,

I'm an Oracle and I'm a perpetual license model. I book a $12 million

deal.

Well I get $12 million worth of revenue right then and in that quarter,

now I'm going to book a $12 million deal, but it's going to be on a three

year term license? It's a three year subscription. I booked that same deal

the last of the quarter and I get zero revenue for the fiscal year. That

revenue is then ratably recognised over the three year subscription. We

do not have that adjustment to make because as you'll recall, 95% of our

business is subscription today. So we are not going to suffer that same

deep ‘U’ that some of these companies you've seen who go to the cloud

face. The other issue is for us, we have our models to move to the cloud,

which is quite an effort to do it right and with elasticity and performance

and TCO that Moe and Cihan discussed. However, all the other things

we're building are cloud first. So we don't have to move our analytics, we

don't have to move our data, that is there today.

63

So that's a little shortcut for us as well. Also our HD models are cloud

first. We don't have an on premise version of them that we're offering. So

I just wanted to give you the shape of that and how you might think about

it in the context of other businesses that are going to the SaaS model, and

some of the advantages that I think of that we have, and what you can

expect and not from that. The next point I would say is that customers

can do whatever they want. So those that are under a lot of cost pressure

might say, "Boy, I'm going to be an early adopter because I want to shut

down my RiskLink on premise because it's costing me X-millions of

dollars, and I want to go to the cloud sooner rather than later."

Others want to take their time. "I bought this hardware, it's sitting

around, I've already capitalised it and amortised it. I I'm fine for another

few years." Great. Let them. If you just want our flood model as some

customers have, you can come on to our cloud just on flood, And we have

customers who've done that. If you want our RiskLink models and you

don't want flood or wildfire, you can do that too. If you just like SiteIQ

and you've never ever had one of our models before and that's all you

want, great. We'll take your business. What about any of the other

analytics applications? Sure. Happy to have your business as well, so

they're not linked anymore. If you just want location intelligence because

you want the multiple trillions of pieces of data to feed right into your

underwriting system, great. Happy to sell you that as well. Don't even

need our model. So all these things play together, but you can come onto

the platform in any way that you like.

So the old notion of when are they going to migrate each and every thing

that they have onto the platform, doesn't exist for us anymore. It really

is, come take what you want. Some will take all, some will take few, and

we'll go from there. So go to market investments will also include making

sure our customers are really happy in dealing with this change

management. We've talked a lot about data, and not to get too technical

but consuming new models is a thing. It takes effort. Some of our

customers take 6-12 months to consume a new model. To understands its

nuance, to explain it to its regulators and so on. Because big important

decisions are made around this. So we are investing more and making

sure that our customer support is excellent through these changes that

our customers are going through, and that we understand the buying

journey, and that we're easier to do business with than we have been in

the past.

So what does that mean for broad stroke timelines? Well, if you think

about it, our customers take a long time to buy things because they're

64

regulated. So the insurance industry generally takes 6-12 months to buy a

new thing once you show it to them. So we launch a product, 6-12

months later, we start to get our sales from that product. We have some

tire kickers, and some early adopters along the way, and it grows from

there. Then we get that booking in my early example of that $12 million,

and that booking still isn't revenue. Even though we built the product and

we've waited here, and we've taken the order and we have the total

contract value, now we wait for it to ratably be recognised over the one,

two or three year subscription of that. That's why when you think about

the mid-point that we talk about, why it takes a little while for the stuff

that you saw today to convert into revenue and profitable growth.

But we're always keeping our eye on that prize. We're always keeping our

eye on cogs to make sure that our costs are in line with our future

business. So how to think about medium term based on all of that,

modest revenue growth for the next couple of years. With revenue

growth accelerating as the rubber hits the road, and as we gain traction

in that market, and as those sales become bookings, become revenue.

And that starts to build in the areas of our existing business, as well as

the new areas that we talked about. We did invest as we announced

earlier, through DMGT about $20 million additional in this fiscal year,

2019, than we did in 2018. Investments will continue, and I'm sure you'd

like a dollar amount and how that'll affect margin next year. We're not

going to give very particulars around that, but what I'm indicating here is

it's not just one of these, like a one-time surge shot. But a very

demonstrable and deliberate view of getting into a $7 billion totally

addressable market, instead of a $500 million totally addressable market.

We'll take a few dollars and make a few bets very judiciously, very

thoughtfully as we go, with our eye on the long-term prize. When we

have a trade-off to make, between a short-term goose of revenue that

could cost us a long-term relationship, or a long-term position in the

market, we will choose the latter each and every time. Particularly in a

market like this that is incredibly sticky. It's partner oriented. You must

earn that partnership, and we're beginning to earn that back with our

clients after many years of non-delivery, and we'll do nothing to

jeopardise that in terms of the coming mid-term or long-term years, for

the sake of short term revenue. You can expect as these new markets,

more data, more analytics, more risk analytics, emerge, that the balance

of our revenue will also shift. So that we're not as reliant solely on the

model revenue, but you can count on us to continue to invest in our

thought leadership, our science and data leadership, with the models,

and our clients demand that of us and want that of us.

65

So clear focus always on driving enterprise value, profitable growth and

healthy margins in the medium term. So with that I'd like to end on a

personal note. Thank you for being here all afternoon and taking in so

many details. It's been a great journey for me the last year and I'm

having, on a personal level, a ball. This is an interesting and challenging

and innovative market, and I am so intrigued by the notion of risk, and

how that emerging risk market that we're seeing takes us to the next

level. I do believe we're at one of those moments in the risk market

where the transformation is inevitable, and we have all of these toolkits

of emerging technologies that we can apply to that in ways that simply

weren't possible before. We're all excited to work on this. I'm very proud

of the team that we've assembled beyond the folks in this room.

There have been another hundred, now I think we're up to 120, very

talented technologists that have come into the business over this period

of time. I have had the privilege over my career of working with some

exceptional teams, and I would like to tell you this team here and the

folks that they brought in, is the best team I've had since my class of, at

Oracle. And that was a famously, a wonderful team that we had a lot of

fun and created a lot of value along the way. So with that I'll just leave

you. I am terrifically confident over the long-term, that we have an

opportunity here to build an important company in the area of risk. I'm

looking forward to coming back when you invite us back sometime down

the road to show you what we've done next. With that I'll end and we'll

take a five minute break and then come back for the Q&A. Thank you so

much for your time today.

PZ: Thank you very much. We have set aside a bit of time for with the whole

team, so please fire away. One or two questions will be preferable than

three or four, since they're not used to the rapid-fire approach. But let's

see who has, who's ready with the first?

MW: Thanks a lot. It's Matthew Walker from Credit Suisse. Two questions

please. The first is just a question on the renewal rates. It looked like it

had a pretty meteoric rise in a very short space of time, up to 98% or

whatever. Are you surprised at the revenue then also doesn't spike if you

have such a massive increase in your renewal rate, what's happening

there? Are you getting some special offers to get people on board? If you

could just explain a little bit about that. Then the second question is on

the incremental margins. So as a revenue comes through, given your cost

base, the way you described it, it looked to be, I don't know, 70% fixed.

What are the incremental margins? What's the drop through on each

66

dollar of revenue, on each incremental dollar of revenue, what's the

drop-through going forward that we should be modeling?

KW: Great. So I'll hit the first question. It's a great one. On the 98% renewal

rate. The way I take that is a market confidence in RMS. And the way

again we measure that is dollar for dollar. So if we had a million dollars

up for renewal, and some customers went down $200,000 and another

customer paid $200,000 more from that same product base, and that

same number of customers, that would be 100% renewal right there.

Right? So in that 98% number, there are actually some losses and some

gains in that. Very little logo movement. Most of that is about deals

shrinking and growing. So we had some customers who paid us

incrementally more, and some customers who paid us incrementally

less. If you think about that lag that I explained between bookings, when

bookings go down, downward revenue follows. When bookings go up,

upward revenue follows, but that's just one factor.

So for example, when you saw the 2018 revenue run rate, you'll recall

that there were some non-recurring revenue that made up, as we

explained it last year, a good deal of the growth that you saw in the FY '18

revenue number that we did not pursue in FY '19. So what you're seeing

in the first half of FY '19 is that went away as well as the residual impact

of the lower renewal rate that we experienced in FY '18, coming into play

in our first half of FY '19.

Now here's what I would like to caution you on, not to take that 98%

renewal rate, which I think is a great confidence show of us in the

market and isolate that from all the other factors impacting our business.

A couple of those would be this. Those M&A transactions that you saw

will continue to have an impact on us in a negative way, in the latter part

of 2019, and in some of 2020. And by the way, what I've said doesn't

change a bit of the guidance that DGMT gave you for FY '19. This is just a

fact of the coming and going of our business.

So I already know of some customers who have merged and now feel the

impact of that, as far as a reduced booking from those two combined

companies in FY '20. That will offset that lovely boost up in the 98% that

we received, which will be offset again by perhaps some uptake in the

new products that we're selling, which might be also a down a little bit

again. Because one of my customers decides in Q1 of 2020 to get out of

the business in Asia Pac, and they no longer need the X-millions of

dollars they buy for Asia Pac. So I would caution you to take that 98%

renewal rate as I do. It's a great sign. It's a good number. It's something

67

that I watch every day, but it is not in isolation an indicator of what's

going to happen in the following year. Our business is complex and there

are a lot of factors coming into play.

The second question about how all that falls to the bottom line. I'm really

not prepared to answer because you're asking for very specific guidance

on a dollar of revenue, minus X-margin equals a dollar of operating

profit. And we're not going to give any forward looking guidance on that.

What I will tell you is that, as you know, we've invested an additional

nearly $20 million in the business in 2019. Some of that you saw come out

in the first half. More of that, as Paul guided earlier, will come out in the

second half and then we're not giving guidance beyond that in terms of

the next fiscal year.

AM: Thanks Adam. It's Alex Mees here from JP Morgan. I'll keep it to two,

Paul. Firstly, just with regard to moving into the seven billion dollar total

addressable market, I wonder to what extent does that rely on displacing

people who are already providing these services, and how much of it is

actually growing the market? Then secondly, with regard to the current

penetration of customers who are on the cloud, what is that, and how

long do you think it will take before everybody is on the cloud?

KW: Thank you. Cihan, do you want to take the first one, and I'll take the

second?

CB: Yeah. So in terms of displacing existing versus growth, I think about the

analytics market in a couple of different ways. It's the red box they've

seen in the middle. One of that is there is analytics that is retrospective,

which looks back. Rear view mirror tells you about what has actually

happened in your business, how many sales did you have last month or

last quarter, versus analytics that is about now. That's about real time.

And analytics that's about the future, about predicting the future.

The parts where I don't expect a lot of growth is the retrospective part,

which is where some of the classic vendors play. But the parts where we

actually make a huge difference today with the models, if you think

about how the models operate, it's about predicting the future and about

real time as an event goes on, for example, on the Florida costs, the cone

of uncertainty, and understanding what that event is going to do to your

exposures, the real time impact, and predicting the next few days, the

next few weeks. Those are some of the things that are very, very

interesting for us. Obviously we operate in that space but we think

there's a lot more that we can do with the new applications to make it

68

simpler for people to get to the IP and the predictions that we can

generate.

KW: I think the first things I would anticipate from my experience displacing

are these custom legacy IT solutions that they spend many dollars on.

They're tired of investing in for a customer of one. Once you do that on

the cloud, it becomes a lot easier.

I think the second thing we'll displace over time are little point solutions

that only do one thing that IT has integrated in and integrated in and

integrated in and those are very fragile. The problem that our customers

are telling us they have with those systems is that they haven't scaled to

70, 100 million locations to the global businesses. So you might have a

multinational insurance company who bought, in one case, over 30

companies. Now they're looking at this and they can't see across their

business anymore. All 30 companies, all 30 legacy systems have bits and

pieces of point solutions. And the second thing, other than where is that

eighth event that's keeping them up at night is what is my cumulated risk

and do I have the capital to cover that? Am I capital adequate for that? So

that's the second nightmare for them.

I think in terms of journey to the cloud, a couple of ways to think about

it, we already threw out some numbers that we have over 80 customers

in our cloud today. But remember what I said is that in this industry

overall, in all industries combined, only 3% of IT has moved to the cloud.

So if you think about it, we have some customers that are in our cloud

only to run our life risk model. That's a really tiny piece of what they do

with us. We have other customers who have come to the cloud just to run

the flood model. We have another set of customers who are running

some RiskLink models in the cloud and so on.

So what you're asking is really when will everybody be in the cloud, and

I'm not smart enough to know that today in a meaningful way. What we

will watch for and what I expect to see is we'll see more activity in terms

of movement to the cloud in 2020 as the RiskLink as a service comes fully

to market. And we'll just continue to pay attention to that as we go.

But our growth and our business isn't reliant on everybody racing to the

cloud with everything. Because of the modularity and the incremental

solutions we have, we could have some very profitable on premise

customers today that we have large, eight figure agreements with that

then incrementally add something like SiteIQ or LocationIQ and that's

their first place on the cloud. So I no longer look at it as you used to think

69

about it, which is then you're going to take this and move the whole thing

over. Really the way we're doing it is how you see it done in other

industries, this more incremental approach, this more modular

approach, because frankly, that is a big principle of successful platforms.

PZ: Next question. Katherine?

KT: Hi. It's Katherine Tait from Goldman Sachs. Two questions from me as

well, please. First one, one of the things that really came through in your

presentation was how much valuable data there was that was locked into

the legacy business that had not been visible or valued by your customer

base. As you pull that out and I suppose re-productise that data, how are

you selling that back to the customer base? Are you seeing any pushback

in terms of them paying incrementally for products that they, in some

form, had previously? Can you just talk us through the mechanics of

that?

Then my second question is really on data quality. Clearly the models are

only ever as good as the data that feeds them. As you expand into some of

these new exciting and adjacent areas, can you talk about where you're

sourcing this data from, how you get confidence around that data and the

outcomes that it's producing and perhaps also give us an idea of how

does the cost of that data compare to your historic models, both in terms

of I suppose the actually going out and getting the data as well as the

expertise that it requires? Thank you.

KW: I imagine both Cihan and Moe are at the edge of their seat to answer that.

So I'm going to pass it to you, Cihan first, and ask you, Moe, to pile on.

CB: Okay. I'll start with the ... Let's start with the first question. How do we

deliver that data and unlock that in productise that, like you say? Moe

showed SiteIQ, which is one way in which we actually delivered that data

to a user that wasn't in the past sophisticated enough to be able to fiddle

with the model and be able to get to that information. That's one way

you've seen that powerful usability that we have. It's extremely simple

for users to use, for anyone to be able to interact with that information.

And that is an example of taking model information, high quality data,

predictions, the scores that you've actually seen are coming out of the

high quality models that we actually have. Location Intelligence API,

which was kind of the sister product, is another way in which you could

programmatically pull that into other systems that you already have. So

you don't have to rip and replace an existing system. You can basically be

a customer who has a solution already today and you can say I'm going to

70

augment that decision making with this new data set that I can get from

RMS and plug that directly into their system. That makes it much easier

for them to get to that information.

And both of these mechanisms really are examples of how we unlocked

quickly the information that you can get to.

Your second question, I don't know if you-

KW: Data quality.

CB: Data quality? Was ... So remember that we've actually been doing this for

30 years with our models and it's an accumulation of information that we

have that gives us an amazing edge. Over that period of time, the source

of that information, the collected information that we have, is really kind

of improved and expanded. And that's what makes it very hard to catch

up and be able to kind of start from scratch and be able to do this.

You're absolutely right that models are data and models are only as good

as the data itself. So collecting that information over 30 years is what

really makes us extremely unique and this makes it extremely strong for

us in terms of data.

KT: And future adjacencies where there isn't that track record that you have.

Can you talk about how you're going about ensuring you can replicate the

same level of expertise?

CB: Yeah. The competitors are going to be watching many of these, so we

don't want to give out the secret sauce, but getting to that information is

one of the important things. I got that question before as well. Is one of

the important things that we actually do that really kind of gives us the

edge here and in the future.

KW: I would speak without giving away anything, that our customers know

that our model data inputs are the best and that is one of the reasons our

models are the best. Our customers understand that the model output

such as lost costs are the most reliable and so they give us that in the

market.

With these guys, with all the years of experience that they have and the

300 guys behind them, we have that trusted factor and we do have a

process around that.

71

You asked a question about is there any pushback. No. When I came

here, I thought Location Intelligence was a great idea, but it wasn't

baked. We had customers who wanted it but it wouldn't scale and the

functionality, it didn't have the access. So we just waited to come to

market until we were good and ready with a scaled product that would be

a great experience.

So pricing is one of the functions that lives under Cihan in products. We

value base our pricing strategy and we haven't gotten pushback yet on

that. But we've just launched, and if we do, we'll adjust our business

model to what the market's asking us to do.

PZ: Patrick, you've got the microphone.

PW: Patrick Wellington from Morgan Stanley. A couple of questions.

The core models market, I think you described as being 500 million a

year. Seems quite small. Is it actually in decline at the moment?

KW: Is it in decline?

PW: Is the market declining in size?

KW: You know, I don't think so, but it could be. Some say it's dropping a few

points, some say it's raising a few points and it depends on what you put

in there. Some put cyber in there. I don't. Some put adjacencies in there.

I don't. So I look at that natural catastrophe model space as a pure play,

natural catastrophe. The global climate, the earthquake, flood, wildfire,

those are all the natural catastrophes and if you remember from our

chart underneath that, Patrick, was marine and terror and cyber. So I

don't include that in that 500 million measurable market.

Whether that market is going up 10 points over a few years or down 10

points over a few years, the fundamental issue is that it's got a headroom

issue. That market has a headroom problem because it is living in the

cost centre of a natural catastrophe modelling space within an insurance

company and it just has a natural headroom to it there.

The dollars for things like cyber that could affect liability policies, it

could affect E&O policies, different budget. The models like terror right

now are coming after NatCat. I imagine they're going to end up with

liabilities coming out of different budgets. So I do count those dollars

72

separately, Patrick. But for us, these emerging markets are different and

for us, new lines like liability would be different, but NatCat is what it is.

PW: So roughly what percentage of your current revenues are addressing that

500 million market?

KW: Right. We don't break it down that specifically.

PW: No, you don't. So ... Yeah, that's true.

KW: Thank you for playing. But-

PW: Is it a large percentage or ... is it a large percentage or a small

percentage?

KW: It's the majority of the revenue is addressing that market, yes. But not

just the models. It's the software, RiskLink on premise software and Risk

Browser software that comes with the models. So you could consider that

together as a package. That is the majority, meaning greater than half of

our revenue, and then we have as I mentioned the broad strokes that I

gave you was that we have an eight figure data business and we have an

eight figure analytical services business and we just don't provide greater

granularity than that in our revenue breakdown.

PW: Yes. Except you do in Paul's fifth slide where he says that the market, or

you have a market growing at 10% addressed by RMS of which you have

an 18% share. But you are talking about modest revenue growth for the

next few years. So my question might be for how long do you think you'll

be underperforming the risk market or how low do you think your share

will go in that market? Market growing at 10%, 18% share, you've got

modest growth.

AW: Just to clarify, the 18% is the share of DMGT's revenue.

KW: Yeah.

PZ: Yes.

PW: Oh, sorry. Beg your pardon. My mistake.

PZ: It can't be the case because we have roughly 300 million in revenue and

there's roughly a 500 million market.

73

PW: I was doing it the other way. So if we take the other side of it, which is the

market's growing at 10%, and you're going to be growing modestly over

the next few years...

PZ: So that's a broader market. It covers the various markets in the stack that

Karen talked about.

PW: Which is why I'm asking the question of how much of your business is

addressing that part of the market.

PZ: As we said, as you know we've been moving to provide less granular

guidance. We used to provide customer numbers. We don't do that

anymore, so we're not going to get drawn into minutia specifics about it

because then we end up in situations where a factory shuts down

somewhere or a show doesn't happen in one quarter and then it moves

into another fiscal year and we have to revise. All of our shareholders

have said actually higher level. So let's move onto another question.

There.

Then we can go down the row.

AMa: Hi, Annick Maas, Exane BNP Paribas. You've underlined the strength of

your new management team. I guess these people are very sought after.

So maybe could you give us an idea of the incentivisation structure and

KPIs?

KW: Sure. At a higher level, this team is here for a couple of reasons. One is

that you tend to find Silicon Valley company strong mission oriented. If

you don't love the mission and you don't feel it's important and it doesn't

resonate with you, you don't do it. So we are missionaries, not

mercenaries.

The second part of it is how we work together as a team is really

important to us. We spend a lot of time together, we feel like founders of

the business in a way and we take a founder mentality to how we run the

business. That's important to us to have the freedom to do that and how

we interplay and work together is the second reason people select in,

because they want to work on this team. And we've very quickly brought

in 120 new people around this team who also feel a part of that and they

have augmented the brilliant team that these guys have put together over

the year. So we feel good about that.

74

The third reason is that if we get this right and we build the important

company that we want to build, we can all benefit financially from that.

What that means is we're all focused on long-term enterprise value,

which ultimately translates, as you well know, into revenue and profit.

But we keep our head down and think about long-term enterprise value

as what drives our compensation in both the long-term and the executive

team has metrics along the lines of revenue and profit each year that

we're expected to meet in the short term as well as the longer term play.

Then as you can imagine, different teams underneath of each of these

guys have very particular metrics that we know will drive revenue and

profit if this team delivers the product on time, if this team delivers the

model on time, if this team manages its service business profitably and

so on. So you can imagine that a couple layers down, there are very

specific goals and metrics and we do believe in OKRs, objectives and key

results. John Doerr is going to be coming in talking to our whole team

about his book and spending a couple days with us. So we're excited

about that. So we are very metrics driven, but that's how to think about

our compensation and why the people around the table.

PZ: Just hand the microphone down. Thanks.

CC: Thank you. This is Chris Collett from Deutsche Bank. Just had two

questions. One was you talked about your move to a more modular

approach. Clearly the right thing to do, but just wondering, does that

result in the short term in a bit of revenue leakage as customers say we

just want to pay for the parts that we want, rather than a more bundled

approach previously.

Then second question was just could you give us some extent or an idea

of the extent to which your products are really being taken more as an

API form so they're getting really embedded into the customer workflow?

KW: Great questions. I'll take the first one, and Moe, I'll ask you to take the

second.

MK: Sure.

KW: On the modularity, no, our clients have always been able to buy the

models that they needed and wanted on a modular basis. So if you were

in the flood market, you could get our flood model. If you had wildfire

exposures in California, you can pick up that model. If you're in the

Florida hurricane model, you can do that.

75

That said, as you know, much of our industry has consolidated and what

drives a lot of the property and casualty insurance market are large

global multinational companies that really count on us for the whole

model portfolio as the primary view of risk. And so that's how some of

our customers buy, but you've always had choice.

I think when I visited the customers initially last summer, what I heard is

that they were staying away from making investments or looking at the

platform because they were scared of it, because it was monolithic and it

was an all or nothing proposition and that's not what they wanted. And it

wasn't API driven and it wasn't modular in nature and I heard some

choice words about that. Going forward, so I think the opposite is true.

We had never forecast revenue for all these things. The RMS(one)

revenue stream was quite small. So it is what that is.

So I think this will be more incremental over time as people decide I

want this, now I want this, and as Cihan I thought pointed out pretty

well, once you get SiteIQ and having ExposureIQ or if you decide to just

put your exposure management, custom tool, via our APIs, good for you,

now you get the benefits of the Data Lake, you get the advantages of that

instant integration and interoperability of the data in the process,

whether it's ours or yours. In either case, there's a revenue stream for us

at the API level, at the application level, at the data level, at the model

level. And customers coming and going as they please. I think actually it

opens up new markets, and we'll see how that evolves over the next

couple of years.

Do you want to take the API question, Moe?

MK: Yeah. We've really taken a different approach for the whole APIs as well.

Before we had APIs for that customer specific tenant installation, which

was great for their own internal productivity. But as we've shown the

approach of taking those things out that are adding costs for each one of

those tenants pulling it out and having one scalable service, and as we

move to self service for these things, just like any other ... we've built a

lot of data services, business, probably one of the ... Actually, no, we

built the world's largest data brokerage business for both public and

commercial data.

What's interesting is you cannot dream of what people are going to do

with that and they're paying per transaction, per subscription. And so if

anything, it's going to be net new growth on that aspect.

76

CC: Great.

ND: Yeah, hi. It's Nick Dempsey from Barclays. So I've got two questions left.

First one, you talked about AIR as your main competitor. I guess you

meant them investing not that much. Dig into the risk numbers, you look

at that it’s in been doing mid to high single digit organic. They talk about

that bit doing, probably more than that, they always talk about it warmly

in their little commentary.

So what have they been doing right over the last five or six years to be

doing high single, perhaps low double digit organic revenue growth

when you guys have been doing kind of a very low single? And second

question, you talked about Aon and Willis as two deals that you signed up

for multiple years. Are they paying less or more on an annual basis than

before?

KW: Good question. I'm trying to think if they would mind me saying. Yeah, I

don't think I'm going to give the figures if it's more or less. I can tell you,

just because they didn't tell me I could and I don't want to divulge

individual contracts of individual customers. These do represent five

year collective, between the two, nine figure agreements between those

two brokers and us and they are the standard operating procedure going

forward with other brokers and we're very pleased with the deals, we're

very pleased with the alignment that we feel we'll get into the market

with the broker community.

On AIR, I don't actually know that they break out the AIR revenues, per

se, because what I count when I count coming and going logos from

2017, I have gained more logos than I've lost in the evaluation that I've

done from 2017, 2018, and 2019. The various businesses are quite

different from ours because they are in predominantly the data business

and their engine for much of their revenue and growth comes from the

IOS where they've taken claims data from various insurance companies

and then resell it back in the aggregate to those companies. If you look at

the risk register data and the risks they identify always as number one is

that if one day insurance companies chose not to give them that data

anymore, then they'd be in trouble as a business.

The last data that I remember seeing, and you could be right, was single

digit growth on organic and the inorganic growth pushing them into the

double, but I haven't looked in a few quarters. So you could be more up

to date than I. But we do feel like AIR as a competitor, they did not

choose to innovate and I think that puts us in a good shot now.

77

They had a stumble with their earthquake model that they had to

publicly come out and chat about. We think again, it's just another factor

of our emphasis on the quality of our models and our science and their

go to market is a different approach than ours. But I can say looking at

my numbers, they're not picking up share from me.

PZ: Any more questions? Come on. Okay. Well, thank you very much for

joining us this afternoon. Thank you to the RMS team which I think

answered all of your questions. When we were in the process of the

Euromoney transaction, the thing that we heard loud and clear from our

analysts and from the shareholders was, given how important RMS is to

the future of DMGT, they wanted to hear more. I think hopefully you've

gotten everything you asked for and more today.

What you've seen is a team. What I hope you've taken away is what I've

been seeing in week in, week out as I've spent time with this team over

the last year plus. It's a team that's redefined its vision. It's renewed its

mission as a company. It's revitalised the culture of RMS. It's completely

different than the company of 18 months ago and it's rebuilding its

industry relationships and has a bright and prosperous future, not just

for the industry but for the team and for DMGT.

So thank you very much indeed. There will be well earned cocktails out

in the rotunda now. Thank you.

KW: Thank you very much for coming.