top of page

Digi Discussions

Building digital trust with the next level of data-driven risk models & engagements

Gail: Let me introduce Yannick Even, Global Analytics Partner for Swiss Re used to be based in Hong Kong and moved a few months ago to Zurich.

I mean there's no doubt that our members will find interest in what you have to say so I kind of put together a couple of quick questions that I think will stimulate their way of thinking and it could also open our minds with regard to what's happening in the global insurance market

There's a lot of talk from brokers and I think we previously spoke about it in a different session that we had like was everybody moving towards this digital age and in South Africa, we just recently received the POPIA Act, Protection of Personal Information Act.

The legislation was recently passed in alignment to protect clients’ personal information and I believe that's pretty much your ethos in you often say ethical data sharing and sharing the data with trust of your consumer…

What would you say is responsible AI that brokers or insurers could guide their clients to feel safe about sharing their data that it could be used in effective weight insurance?

Yannick: It's really important in order to build trust, and especially digital trust, to understand and share with your customer what you are doing with the data that you collect. How do you use the data for their advantage and not only the advantage of the broker and/or the insurer? So, find the common value that the data can bring. It can be either to better understand the risk at the personal level, risk a customer is exposed to and want to be protected against. Or it could be a better understanding of the customer itself. In terms of where is he in the life cycle, is he over-covered in some areas or are there gaps (to be filled)?

I think every insurer across the world is starting to realize that, in the world like 10 years ago where not much was digitalised, the trust was really built by the broker in their one-to-one relationship (direct with the customer), at the time of underwriting – ie. understand the risks and price it, or at the time of claim - which is usually the moment of truth for the insurer.

Now as insurers across the world are starting to digitalize those touch points, it's very important that insurers translate how this trust was built by brokers or the servicing agents into a digital world. This comes from a better understanding, transparency, fairness, ethics practice in order to explain to customers why we need to collect the data, what do we do with the data, and what's the value that customers get from automation, from personalization and from better prevention etc. If the customers do NOT see the value the insurer can provide, it will be hard (for insurer) to gain their digital trust.

So, that's what I found interesting in this data-driven focus of insurance transformation is that you're asking yourself the right questions on behalf of the customer and on behalf of the broker. What is exactly the value I'm bringing here? It could be just about cost efficiency, for my services to be less costly and provide fairer price. It could also be much more, it could be better personalization, better prevention, better prediction, that can help me basically ensure that my broker understand much better where the customer is in his life journey, what is his need at his personal level, what are his risk? maybe even (some protection) could be decreased if needed? I've never seen that, but it will be interesting to decrease a protection that is not really needed, based on personal risk in order to increase where on the other side it's really more needed.

I think as you do that, you really gain trust from the customer, because you need then to explain properly to the customer from all this data, we collect with his consent, we can go much deeper into this very granular conversation about risk.

Gail: It's so interesting - I mean we can talk about this for days.

Yannick: Yes, indeed, and then we could do all this looking at different risk pools. So, we could do this for motor, we could do this for life, we can do this for medical… So, there are different ways to do it - different kind of data we can collect. But overall, I think the idea is that we are NOW in in the time where we need to build digital trust for the brokers, for the customers, for the insurers, for the regulators - it makes a lot of sense to go to this next level of depth in our understanding of risk with data for the customer.

The benefits of a data-driven culture across the value chain & the embedded insurance opportunity.

Gail: What would you say are some of the biggest innovations in insurance industry that you've seen in the past five years?

Yannick: Well, there are several things… Innovation purely linked to technology advancement, such as ability to leverage cloud processing to transform big amount of data into insights - either customer or risk insights. Technology linked to connectivity such as API or privacy computing that allow you to build more embedded protection and embedded services within ecosystem and make it more relevant for the customer at a certain time in his journey. That is no more the journey of the insurer but the overall digital journey (of an individual), you could have for example on your mobile. So, ability to transform data and compute data into meaningful insights, APIs, operationalization AI that enables you to go to the next level of digital transformation.

Then looking at data analytics strategically to get rid of silos that currently exist in all our legacies including addressing at the people/culture element there - people within organization see data as a way to exercise power and it's difficult to switch to understand that you get actually much more value when you collaborate, share data across the organization, rather than keeping the data for yourself because you want to hide maybe certain things in your business. What we've seen is that when you're basically harvest full power of data that your organization have access to, when you start to catalogue all the data that you have access to, when you start to look at your data across all these operational silos, when you have a single view of the customer… then you basically start to really arrive in this data-driven insurance business model that that we aspired for. I think for the one who have who have already started this journey it's a no brainer, they will not come back. But it takes a while to start because I think a lot of people still thinking in the old world where data is seen as "your" data, where you don't really want to expose your data too much across talent of your organization and are a bit worried of what we will find out. It's a bit of a leap of faith that everyone needs to make. But what I've seen is that the organization that promotes data sharing, data usage (and governance) the responsible way, strategically as well, gets some benefit quite early and then it's easy to share the best practices and the stories (across), but it has to be done properly.

Matt: It's interesting you mentioned the APIs and sort of embedded insurance cause that's where kind of we specialize in. I think what I've learned is has to be relevant to the to the business, so whatever your core business is, you got to solve a problem that the business has with insurance, so it's got be closing efficiency somewhere that's what I found. But I think when you look at these embedded insurance solutions, I think the question comes down to you on bringing the cost into the customer or eaten in the platforms margin for example. Where are you seeing this going - platform looking more at absorbing the cost of the insurance instead of margin, give the benefits to the customers or do you see more passing costs to the customer?

Yannick: If you use embedded insurance to sell the product that you currently have - you will get it wrong, that's not what embedded insurance is about. It's about understanding the customer journey within a full ecosystem. What are my customers micro-segments, what are the different kind of journeys they do in this ecosystem? Look at it not as the insurer customer, but look at the customer as an ecosystem customer, with all your partners. Understand their journeys, understand their needs, and there is a particular time in this journey where protection will be in their mind and then, what it is? What kind of embedded services do you want to create to be relevant at this point of time with the data you have access to within the ecosystem, what is the right price and maybe automate (all this process). Most of the time when you do that, the solution will not be your current product. You need to create something different (to be relevant) and if you're able to do that, you can scale. If you just, try to use it as another distribution channel to sell your existing product, you will fail.

Data driven approaches for small brokers, managing data as an asset.

Gail: You know, I am fascinated by this, this is actually proof to what I was going to say. My next question is how can SME (Small to medium enterprises) can really develop AI or a digital presence in their own business without spending too much money?

Yannick: sorry are you speaking about SME business or commercial lines insurance/Insurance for SMEs?

Gail: yes, let's say an insurance broker, a small brokerage maybe 10 employees very small business how would he incorporate this into its business strategy?

Yannick: Well, first is to understand what data you get access to. Most of the time, as I said before, even in a small setting it's very siloed. What is the quality of the data you have, how to strategically use data for your decision-making process across all the borders and then what is the value of the data you have? Where are the gaps? How do you plan to fill these gaps? After you catalogue, you can quickly identify gaps and fill these gaps, may be from the insurer you work with, or from the customer, or maybe additional data you are not collecting today that you need to collect, and plan for that.

Then really see your data as asset. That allows you to solve some of the business or data challenges that you have. But the too often, I've seen even small companies start from "OK, let let's try to be data-driven"! But they don't even know what data they have! Where are the gaps, what is the quality of data they collect, how truthful is their data? Do they have consent to actually use this data the way they want to use, as there is a legal aspect as well to it. So, the first thing as I said is strategically look to your data as assets, and act as such. Have everyone into your organization upskill/upgraded about what data we have and why the right data is of importance for your business.

If you constantly make decisions out of your gut or out of the data from your competitor, then you're giving no incentive to make it right. If you show that most of your decision is driven by your data and also know when you don't have the right data, when you risk making the wrong decision, you have then more chance that people get concerned about it and do something about it. This should be in your KPI as well at the end of the year. For example, each broker has usually bonus based on the volume they sell, but what about the quality of the data they bring, what about the value of data they bring? How truthful is the data they bring?

So, this is when you look at your data as an asset, that then you can start to build this and you can align your KPIs to your brokers accordingly, and then as the KPI matters, we all know that there is a chance to happen.

Gail: I recently did a course and it was how to basically tell a story with data. A lot of it linked to how truthful is your data really to begin with. How true is the story that you're really going to tell cause the data can only tell so much based on how it is collected?

Yannick: And to complete then, once you get all your data right then you can start to look at AI and then you can start to look at responsible AI but don't look at AI when your data is wrong or when you have bad quality. Because if you build AI on top of bad data then you will have bad AI decisions and you will lose the trust. People will say it is not working, which is not linked to AI, it's actually based on the data that you used to create this prediction or this automation.

Data-driven Reinsurance perspectives

Matt: I was on a conference recently with I think one of your colleagues from Swiss Re. They had... they developed a tool for usage-based insurance for motor vehicles.

Yannick: Were those telematics based or what's the data they use for usage-based? Is it based on a box in the car or mobile App based?

Matt: Well I think it is App based, using the accelerometer and things like that to see when you drive and how fast you're driving... so, I just want to see from a reinsurance perspective, are you investing in sort of the innovations to pass on to the insurer? And any other sort of innovations you working on at the moment or are you kind of getting feedback from the insurance and seeing where you can help with on the data side?

Yannick: Yes - All what you said. We're doing all what you said. Maybe in terms of investments we tend to invest in mature Insurtech, data, big tech.. every kind of tech company that bring data or tech to enables us to better understand risk is a target. But we will not necessarily invest when they are too small, it needs to have a certain scale. Then we have also many partners we work with, that we don't invest in, but we have kind of common interest to scale and make it work. And again, it’s the same around big tech, data partners, all partners that can help us gain risk insights, or access different kind of customers that we would like to provide protection. Then after spending a bit of time with our partners to basically build and vet this risk model, then we will basically sell out this solution or this insights to the insurer. We have different kind of business models. First, we have the traditional reinsurance model. Second our business corporate solutions, that basically sell directly for commercial lines, we have our own brokers welling our own products. And we have also IptiQ by Swiss Re that is a digital insurer end to end.

Matt: Sorry, what is that name?

Yannick: IPTIQ by Swiss Re, it's a B2B digital insurance proposition, only digital, that by nature is looking specially to the embedded business models, looking at it with of course our risk management lenses.

So, what we bring at the end of the day it's data-driven risk models, especially for underwriting but we manage claims as well.

Building the next generation of talent for our industry with data driven approaches

Gail: If you could give any advice to young people starting their career in the insurance industry, in one or two sentences, short and simple, what would that be?

Yannick: Well two things, if you really want to scale tech and data-driven solutions, you need to learn two things: you need to learn (a bit) about actuarial science, and you need to learn (a lot) about data science. So, I think everyone needs to learn how to code or at least some key principles and this is up to the board level. Everyone needs to understand those concepts - what is machine learning about, ecosystem, embedded insurance... How does it work? What are the business models, new target operating models, and understand how all this tech and data will evolve naturally in the world that we lived in.

What I used to say to young students that have not yet selected the industry they want to work in is insurance is actually one of the most interesting industry as you touch so many different things on day-to-day life of a customer and of companies. I don't really know any another industry that is so interesting. My team, advanced analytics centre of expertise, is currently working on motor, on aviation, on ecommerce, on natural catastrophes, on climate change, on mortality, mobility, disability income protection, pet. Yes, we are even building a data-driven pet insurance product.

So, it's basically about risk (models), it's a touching everything in your life and it's also touching so many things in the commercial lines as well looking into infrastructure, global supply chain, renewable energy, it's like there is not anything that we don't touch and of course overall finance. How to finance all that, the financial risk, and the impact of the inflation on all this costs. So, I came across so many lines of business and it's making so interesting when you work for a big Reinsurer like Swiss Re that basically you can stay in the same company all your life and still have like thousands of different jobs and touching different things.

But the commonality is the understanding of risk, a data-driven understanding with technology. How can we enhance, make it cheaper, better, faster, and more relevant for the customer. And that's why you see the common need to make it happen. Understand actuarial science and understand data science and tech. If you understand a minimum on these two, you can thrive and you can have fantastic career and super interesting. A last important point for the new generation you can linked it all back to creating real benefits for society because all these topics are super important for making a better (more resilient) planet.

Exploring opportunity for space insurance as a new risk pool

Matt: I was saying I came across a new product line I've never heard of before; I got a client and she's working on a satellite that they're sending up to space with SpaceX. It’s called into-orbits insurance was a new one I've never heard.

Yannick: Yeah, so there is now a lot of new insurance specialty insurance – such as space exploration. it's starting to scale. I tell you there's so many different fields that there could be... A satellite collision, what will be the impact? the impact could be huge. Because suddenly a lot of services could be affected, because when we use satellites for everything today. All this GPS in our phone and an IoT devices that starting to make decision more and more autonomously for many things.

So, what happened when there is a collision or when there is a sun, I forget the name in English, flare from the sun these are big risks we're speaking about here that that needs to be modelized and (eventually) priced. I'm not an expert in that by the way.

Gail: Absolutely and if this is solar flare the questions also, how far is going to go? how much damage it’s going to do? it's almost unquantifiable and yet to insure it you must quantify it.

Yannick: Indeed, and there is the risk that we can see which is directly linked to that but then you have all the accumulation risk that will derive from that, and this is sometimes the risk that cost you the most.

Gail: Yes, with many structures that are damaged in these huge events consequential damage is often the higher damage than the damage of the insured item.

Insurance and Distributed Ledger/Blockchain technology

Matt: I'm not a big advocate of cryptocurrency with blockchain technology. I've seen some very innovative companies coming out, and I think insurance is one of the big industries that can be impacted by blockchain technology. Are you seeing a lot of innovation in that space or not so much yet?

Yannick: Yes, at Swiss Re we call it Distributed Ledger Technology, so it's more than blockchain. We are not necessarily looking much to the crypto itself but more to the technology - distributed Ledger technology - that can enable a lot of new use cases. A few years ago, we started small with a few POC's. I was actually the project manager for our first parametric health insurance product ever launched. This was in Singapore for gestational diabetes for the pregnant ladies. Basically, they have extra cost if they get diabetes during pregnancy, extra cost that is not very well covered by the insurance or by the national health programme.

So, we had the idea with MetLife LumenLab to create a product that, with the consent of the customer, will use the data from the electronic health record, five data points to underwrite the pregnant lady. Then if they're eligible, we basically create a smart contract in the blockchain that will basically look every day at the electronic health record (for claims triggers). So, the first benefit is paid automatically if she's diagnosed after a few weeks and the second benefit, which is much more, is paid automatically if you have complications at birth. For that we look through the electronic health record and in theory the smart contract to execute and pay the claims directly on the bank accounts of the pregnant lady.

What was super interesting is that the technology work, it was all secured, but where it doesn't work well is the human element of it… as always. So, we see that it takes some time for the doctors, or for most of the time it's the nurse in the hospital, she needs to go to a dedicated laptop, enter the information, she can make mistakes as well. So, by that time the customer already asking the claim to be paid, after 24 hours, they might complain why not yet paid. Well, we don't have the information yet because of all this this backlog or data mistake and it's taken a while for the information to be properly put into the machine etc.

It was interesting to see technology work but at the end of the day they are human limitations, especially around data input, and data quality as always, we are coming back to that! We start discussing what prevents actually this technology to really scale and provide value.

After this initial POC we moved very quickly into ecosystem at scale, solutions at scale with distributed ledgers and we have a dedicated team withing Swiss Re that is looking at that.

Currently, two main global initiatives as I recalled it but there may be more, that that we worked on. One is around Marine, all this Cargo vessels where we have now a good quality of data and we're using distributed ledgers to basically track and automate the understanding of this risk, the risk mitigation, prediction and the claims (automation). The other one is around the global supply chain, especially around food security. But my colleagues leading the Distributed Ledger Technology can speak about that more. I've not seen the latest update for a while. But there are now some new scenarios where DLT can be good especially in B2B , maybe not yet with customer, but in B2B such as marine, such as global logistics where we used it within big consortium. Distributed Ledger Technology to automate in an efficient way for risk management.

Insurance and facial recognition, microfacial expressions

Yannick: Ping the biggest insurance company in China

Gail: It’s crazy just the idea of how big it is, and they've even gone to the extent where they created it if I do remember right, reads the facial expressions of the agent.

Yannick: yeah, imagine when you have millions of employees (& customers) how do you ID them. So, instead of all the calls you receive because they lost the password, use voice recognition and face recognition to ID them. To reply to your question again, you don't necessarily need to build yourself all this tech, you can leverage vendors, you can open source I, there is a lot of AI that available open source and you can adapt it to your needs.

Gail: I recall you saying the app is so sensitive that it can dictate if the advisor is busy losing the customers interest and it could say ask another question in order to engage.

Yannick: yeah, so this is a going maybe a bit too far in terms of ethics for the western mind, but now looking into the different points (on your face) and your voice - you can basically see in real time your emotion, if people are interested (and engaged, if they are not, if they are bored, if they're doing something else etc. The way they tested first this technology was to use it during training of all their workforce because it's internal only. Then you're able to tell if they are not listening to the training, they need to do the training again for example. That's a way to test and then when it's good enough then you can bring it to the external world and use it for a lot of scenarios.


145 views0 comments

Recent Posts

See All


bottom of page