EP 110 - Rethinking insurance premiums with IoT data - Graham Gordon, Product & Strategy Director, Sapiens
|Dec 17, 2021|
In this episode, we discuss how the insurance industry is using IoT data to more accurately assess risk and to customise insurance premiums. We also explore the challenges of applying data analytics to areas related to consumer health due to privacy concerns and to the long tail of B2B use cases due to complexity.
Our guest today is Graham Gordon, product and strategy director of Sapiens. Sapiens is a leading global provider of software and data solutions for the insurance industry.
IoT ONE is an IoT focused research and advisory firm. We provide research to enable you to grow in the digital age. Our services include market research, competitor information, customer research, market entry, partner scouting, and innovation programs. For more information, please visit iotone.com
Erik: Welcome to the Industrial IoT Spotlight, your number one spot for insight from industrial IoT thought leaders who are transforming businesses today with your host, Erik Walenza.
Welcome back to the Industrial IoT Spotlight podcast. I'm your host, Erik Walenza, CEO of IoT ONE, the consultancy that specializes in supporting digital transformation of operations and businesses. Our guest today is Graham Gordon, product and strategy director of Sapiens. And Sapiens is a leading global provider of software and data solutions for the insurance industry. In this talk, we discussed how the insurance industry is using IoT data to more accurately assess risk and to customize insurance premiums. We also explored the challenges of applying data analytics to areas related to consumer health due to privacy concerns, and to the long tail of B2B use cases due to complexity.
If you find these conversations valuable, please leave us a comment and a five-star review. And if you'd like to share your company's story or recommend a speaker, please email us at team@IoTone.com. Thank you. Graham, thank you for joining me today.
Graham: Good morning. Thank you.
Erik: Before we get into insurance tech, I'm actually very curious how you ended up there, I think you have quite an interesting background. You were working with Masternaut, with Cybit, working with LexisNexis Risk. How did you find your way now to see if it's?
Graham: It's a curious evolution, I'm not quite sure if it was by design, or whether it was deliberate, one of those two things. So we could define that as moving IoT, so taking data from a commercial vehicle, because that's where it was available from and then putting it into a company's infrastructure so they can manage their business from.
If you think about transport, logistics routing, scheduling, the dispatch of a mobile engineer, and then that technology in around 2005, 2006, 2007 led itself quite nicely to consumer telematics. So that's when I started to get involved with insurance companies, and launching UBI programs. So I know it's a current theme at the moment, but these things have been around for an awful long time. I think one of the first commercial ones in Spain was around 2010, Germany 2012. The UK obviously had been leading a while for that one, mainly on the young driver side.
And then I joined LexisNexis, who are a data analytics company. So, if you think about absorbing, normalizing, contextualizing lots of data from multiple sources into something that's insurance usefulness or insurance useful, so that's the provision of data and scoring. And then Sapiens more recently is okay, how do you actually use that data? And Sapiens as a policy administration system or insurance software system.
So the pathway for me has been right the way from knowing sure tech right at the start through to big data analytics and then now, more recently, how do we actually take value from it and use it?
Erik: So you guys are not insurance company, you're selling technology solutions to insurance companies or to work to users of insurance products, who are your customers?
Graham: Our customers are insurance companies, and we supply the insurance software to them. So a way of describing it is how do you buy insurance yourself? Or you can go online and fill the forms in and then you can go to an agent or you can go direct to the insurance company. So we look after that whole end-to-end process that starts with you getting a quote from the organization.
And so, absorbing data from multiple sources, scoring it, looking at the consumer information, presenting the quote back to them, when they click by making sure their billings are all organized and all that kind of good stuff, the policy documents. And so everything that goes right the way through any midterm adjustments, and obviously, making a claim and the process and then the reinsurance at the very end of it.
And if I think about more recently, everything's moving online, so there's more analytics work coming into this. How do we help consumers choose the right cover and the right product, so big education piece, and flowing that through using machine learning and aggregation of large data sets to help the insurer automate as much of these kinds of processes as possible? You know, when you see an organization saying I can give you an insurance code in 60 seconds, we’re the technology that enables that.
So we go running around, pick up lots of information and help the insurer through that process of how to price it, cover it, send the documents to the consumer, and everything in between. We've got about 600 customers around the world, and it is very broadly. It’s a third from North America, a third in Europe, and a third the rest of the world. Rest of the World a horrible term, but Australia, Thailand, Singapore, lots in Africa, Southern Africa, in particular. So we get a quite a comprehensive and quite a macro picture of what's going on in the insurance world around. It's just because of the diversity of the customer base that we have.
Erik: I'm curious what digital looks like insurance right now. On the one hand, so I'm sitting here in Shanghai, China, so companies like Ping An that have thousands of data scientists and make a lot of investments in tech startups that are very tech savvy. But on the other hand, I have a few friends that are basically running around Shanghai selling insurance, the old way, knocking on doors, building relationships and personal networks.
So it feels like there must be some parts that are pushing the frontier of digitalization and there's other parts that are basically operating like they were 100 years ago. What is the industry look like right now?
Graham: Everyone says it depends. But let's take me and you as a consumer, we're really comfortable going online, filling the form and pressing the button: I don't want to speak to anybody, I want everything done just in a minute. Have you met my mother? Good luck getting her to do this, because she'll want that kind of face to face contact with the broker that she's been working with for the whole of her whole working life in the same company, who come and sit down talk to her, explain the cover, explain the options, and then we'll go away and get three, four quotes from an insurance company.
So they do a lot of the work and that one-to-one relationship with the brokers still important for a large segment of the population. Particularly if we start thinking about business insurance as well, so you've got SME, your sole traders are companies with less than 10 employees go online, get the covers suggested, take care of it very, very quickly. And then a more complex risk, multiple sites, multiple transport pubs, or multiple vehicles, or many, many employees that need someone to sit down, understand the business, collect all the documentation and negotiate and then again gives the end customer options and choices about which insurance company they want to use. So you have this kind of dichotomy of hell yeah, we could put everything online to remember there's a consumer at the end of this that wants the choice.
Erik: And then you've got the back end of this where you have to be assessing risk. I think you mentioned UBI, and maybe you can quickly explain what that means. I guess that's more from a consumer perspective. Then you get into how do you assess risk more precisely than we used to, we just put people in broad buckets and then…
Graham: Yeah, so exactly what it is. So let's pretend for a moment we're the same age, and we live in the same district and we both have different wives and family, me and you look the same. So it makes sense at the moment using the datasets that's available, combination of public records and your claim history, and that me and you would get the same prices. So, all these things that the insurance companies are pricing at the moment are proxies for risk. They look at the buckets of people like us and say right, your propensity to claim is X, therefore we are going to price you at Y. And what usage based insurance does is make it more dynamic and relevant and personalized to you as an individual.
So, simple ones that everyone's working on at the moment, mileage propositions. How many miles Am I driving? I might drive 10,000 a year. You might drive 4,000 a year. What does that mean for the exposure, the risk that you've got? Or even down to the more sophisticated solutions which are monitoring how you drive versus how I drive? So what's my acceleration braking speeding profile look like compared to you? That will tell me what my price should be and what your price should be.
And then adding contextualisation to that, so what type of roads are you driving on what time of the day you driving out. And again, that provides another level of insight that can separate me as a risk from yours or risk and give me a price that's very, very different from yours. So there's a growing trend now because consumers are beginning to understand some of this. And particularly with COVID when vehicles were sitting on the side of the garages, or in the street for days, and weeks, and months unend, they weren't being used, and then all of a sudden, you find out, you're paying the same insurance price as if you were using them.
So mileage programs are definitely starting to take off at the moment, and we're seeing a lot of activity around those. Then that naturally leads to Usage Based Insurance, or UBI, which is more about, okay, what do I actually do? And how does that affect my risk?
Erik: Yeah, I guess the central element of the solution there is data, and then you get into question of how do you get that data? Because you're talking about getting data about where I am, how fast I'm accelerating? So, there's the just the technical challenges of actually getting data from all these different sensors and sources. And then there's the privacy issue of how do you get permission to use this data to make commercial decisions? What's the approach here?
Graham: Let's do the privacy one first, that's relatively simple. So everyone we work with really, really transparent. Here's the data we're taking from third party sources. Here's what we're using it for. Do you consent? Yes or no. So that's a very, very simple way of making a complex situation look quite hard. So everything that the consumer is consenting to is articulated, and it's clear for them. And that's really, really important, because you need to trust what this is doing and understand the benefits. Some people like it, people don't like it. If you don't like it, just continue with the way that insurance has always worked. So it's all about consumer choice there.
Erik: It is there usually a bit of a carrier attached to say, hey, if you allow us access to more data, we might be able to reduce your premium?
Graham: Oh, significantly. As the carrier is definitely the safety and the premium reduction, it has a story here. So you have probably the most developed UBI, or Usage Based Insurance market is the UK, it started with young drivers because there was a problem with young driver fatalities and casualties. You know, when you're a kid, you get in the car, you drive very fast, you drive late at night, and you had a pretty high chance of causing a collision injuring yourself very, very badly or even death.
So, one of the first solutions to look at, if we say IoT exchanging data was for that young driver market. And over the course of the past 10 years, the casualty rate has fallen by 44%, which is more than any other technology or road design or speed cameras or anything else you can do. To have that drop, the only difference is telematics, and the telematics in the vehicle monitors the driver’s speed, time, location, all the things we talked about earlier.
On the flip side to that is if I am a driver with less than two years experience, I'd expect my motor premium in the UK to be something extraordinary, like £2,000. What telematics does is you understand the risk, you understand the client, you're clear to them about what you're allowed to do, don't drive fast, don't drive at night. You say if you do all these things, your insurance price is going to be £800. So, that's a pretty good incentive for these kinds of technologies. And that works very, very well when you've got this large premium to get down the next challenges how do you take that approach, and put it into the mass market because the benefits are clear.
Erik: I'm really, really interested in this because this is a topic a lot of other industries are also wrestling with. How do we use data to make better pricing decisions and so forth? How real time is that? So if I'm a teenager, I go out speeding, do I get a message from my insurance provider the next day, or maybe even an hour later that says, hey, your premium is going up next month because of the behavior? Or does it show up in my next bill? Or how is this actually show to the consumer?
Graham: There's a number of approaches. Young drivers with less than two years’ experience or a driver that's returning from a banner or a conviction, and there's a couple of approaches. So, one is the money approach. Okay. Your driving score is X, that means your premium this month is going to be Y. Again, going back to this variable pricing.
So if I'm driving like an idiot, and I'm speeding, I'm going to be paying more. And that's calculated monthly. The data for the consumer is available in an application nine times out of 10 and it shows them the score of the individual trip. So that data is aggregated on a trip level. And then you've got another approach is three strikes and you're out, and which is more draconian, but it's still quite effective.
So if you go over the speeding limit by 10% three times a month, we're going to cancel your insurance. I think young drivers start to understand this quite quickly from their peers and other people where this has happened to that you don't necessarily want your insurance canceled, because it makes it so much more expensive to get covered from the next insurer. So, there's a little bit of a stick approach there on the monthly premiums and then also on the cancellations.
And a lot of companies have tried to look at non fiscal motivations. If you drive nicely, we'll give you a Starbucks voucher or a coffee voucher, they don't tend to work so well. I think people are reacting more to money in their pocket than free coffees or even seeing organizations offering free call car servicing and things.
It's a really good example of just how far, I a call this mobile IoT has worked. There's a lot of information, there's a lot of technology out there, and it's nice to find one that lands, and has been very, very successful.
Erik: So you said that's the easy part, and then the challenging part is getting the data or what?
Graham: Getting the data is that multiple, multiple methods of doing, many manufacturers that are manufacturing these devices. And then you layer on top of that the embedded devices from the cars themselves, most of the cars manufactured today will have a telematics device in it that can send data from the vehicle to wherever it goes. All these devices operate at different frequencies. They cover things differently. They have different interpretations of what things are. So you have to normalize all of that. There's a large scope of work that goes on there as well.
And then cost is the other factor. So we talked about mass market, works very, very well when you have a massive premium difference, or a massive premium upfront cost. Motor premiums, let's assume we haven't had a claim in the past five years, they're actually quite low if you think about it, and then you start to say, well, actually, we're going to add another 10% onto that premium to cover the cost of the devices or the data extraction or whatever. It doesn't necessarily stack so well.
This cost issues the next one. And that's largely been solved now by smartphone only applications, whether you're using the smartphone accelerometer, you're using the smartphone GPS to give you enough data to score the driver on comes with its risks, because of course, the consumer can just turn it off. Again, going back to consumer segmentation, there's a large group of people that are quite comfortable with that because they understand the benefit of it, and the benefit is always going to be a better insurance price than they'll get by going their traditional route.
Erik: What about the companies that Tesla that are now maturing the vehicle into a computer? Do they have any incentive to provide this data in order to maybe tell if you drive a Tesla you might be able to get lower insurance because we have the data that would enable this? Or are they pretty much is that a no go for them?
Graham: That's an interesting one. And it's Tesla, so everyone thinks when they announced this for California, it was like, oh, my goodness, this is so innovative. The industry has been doing this since 2006. But because it's Tesla, it looks different or look shiny.
But what I will say and why it's important to Tesla and for most car manufacturers now that are selling electric vehicles, if you think about the whole life cost of your vehicle, you've got the maintenance, you've got the sticker price or the purchase price, you've got the insurance, you got the resale value, and that's pretty much about it.
If you've got an electric vehicle, all of a sudden that maintenance cost goes away and that fuel cost kind of goes away. It's still there, but it's hidden somewhere in your home electric bill or whatever it's going to be. So it's hidden and you're left with the sticker price and the insurance price. So all of a sudden the insurance starts to become more front of mind for the consumer because you're noticing it because you don't have those other costs that we're used to when we're driving petrol and diesel cars.
And because of the safety technology in the vehicles now, there's a whole body of work around the A-DOS technology so the safety features in the vehicle, so whether that's lane departure warning, autonomous braking, and cross traffic, alerting some of the stuff make sure you're staying awake. And as the car start to move more autonomous, this is all around the car is doing something better than the driver can itself. So an autonomous car equipped with autonomous braking can break faster than you or I can.
And you look at that and go well, actually, at the start, we talked about how all these vehicles are priced, or how our car insurance is priced at the moment, it's all based on me and you as individuals. And then we're looking at UBI, me and you as individuals and our driving styles, patterns, and times or days, and now the emphasis is changing to the actual vehicle itself and what technology is on the vehicle.
The first route in this path is how do you identify what's on the vehicle itself? And the second one is how do you get that data from the vehicle into the insurance infrastructure. And then the third one is, which we always get asked for, what happens if the consumer turns the safety feature off? So has this safety feature being turned on or off? You might get irritated by the beep every time you cross the center lane or something like that, so you turn it off.
So insurers do want to know this, but they need to go back and say there's actually benefits we're looking at the simplest stuff first. And then eventually, probably might not have been retired by this time, all these things are going to combine to look at product liability versus personal liability. If the car's got all these safety features, and if the car can drive itself as the car can act and be a better driver than I am, that emphasis is going to change to a product liability for your motor insurance, and which part of that's going to be personal. But I think that's still a long way in the distance.
Erik: These cases where part of that's been on an advanced autopilot, and it's crashed. Right now, it's that liability, the courts are basically still allocating that to individuals [inaudible 21:49]?
Graham: I go back over the history of time again GM they’re OnStar, which is one of the first insurance products that use this kind of data, progressive in this day have been using it with their snapshot program, multiple European examples where they package the insurance and the car pricing and everything together. They had do these, just add fuel programs when you're buying the car. And it kind of like was going in a reasonable trajectory, and then just stopped. And it stopped largely because of a couple of things that happened around financial crisis and all the rest of it. And how willing OEMs were to invest in these things, and also the technology side of it.
But terrified about the privacy implementations, it's great if I as a car manufacturer, I know who's driving the car. But once that goes into the secondary market, there's a whole raft of things that have to happen to make sure that I'm not collecting data on a consumer that hasn't consented to. And then what's happened over the past 18 months is suddenly just picked up. And again, it's coming to the front of mind again. And I think it's largely because some of those technology issues that we talked about and the privacy issues are solved. The car manufacturers have got more confidence in this and they want to start taking more of the consumer ownership, getting a payback for the investment they've made in this technology.
Erik: Let's go a little bit into Sapiens offering. So if I'm looking at just your data in digital solutions, and I see basically three components, the digital suite, which looks to me like different maybe SaaS applications, the analytics and BI, which seems to be around the data, the structure and offering and then emerge, which looks like a platform. Can you help us understand a bit more what is the technology behind these offerings actually look like?
Graham: The technology we built it on was so cloud enabled and was the first thing. So it can be deployed anywhere, anytime, operates 24/7. So digital suite is interesting because the insurance industry, if you think about the software that goes into managing millions of consumers, a lot of this technology is decades old, but it just works. But it doesn't work in the new digital environment.
So the whole point of some of the Sapiens work here is how do we allow these really cool consumer journeys that are SME journeys that can happen on an app or online or wherever the customer wants to be, how do you do that and still do that on legacy systems? So we put a digital layer on top of it, where we can use the API orchestration to extract data from the core policy database, how the insurance price on this old technology, drag it into the 21st century by putting a digital era on front of it. So that's a huge growth engine for us at the moment, because everyone just wants to buy quickly and simply, I suppose.
The platform side of it is more about distribution. So we talk about our consumers as being the clients. But a lot of cases, insurance companies, they've got agents, they've got brokers. So it's the kind of same approach, but I'm delivering that information to the broker or the agent for them then to have the conversation with the end consumer, like we talked about, that one-to-one relationship that's really, really important. So it's the same principle, but you're just delivering that into an environment that a broker or an agent can use. And that gets quite interesting and can be quite complicated because you're trying to present information for the broker.
The broker is always going to represent the client, not the insurance company. I think that's an important distinction to make. So quite often, they're presenting three or four different quotation packages. So we're allowing them to access that near real time and then make those decisions about what products and cover levels and costs they recommend to the insurer themselves.
And then data analytics, that's the buzzword and the value at the moment, two sides to this. Number one is helping the customer choose the right products coverages. And that can be done a number of ways, where we're looking at large data sets of what do customers that look like you buy? Well, they bought this, so we can recommend that. And also the education side of it. If you're doing this online as a consumer, I want to see what I'm covered for and what I'm not covered for very, very quickly. I don't want to go back to the days where I've got 16 pages of terms and conditions and cover limits to read through. So there's a huge body of work that goes into scanning these old documents, looking at them, and then trying to extract okay, these are the pertinent points this make this really, really easy for a consumer to understand.
And then the second part of that is the recommendation of products and cover limits cover lowers for the consumer, because we're not explaining anything to them face-to-face anymore, so we have to do that as part of the buying process. Claims is fascinating. If you make a claim, I know it's only going to be 10% of an insurance customers. We call them book of business, so it's not everybody. But how do we extract all of this data on a claim, verify it, put a fraud score on it, and then pay the person that's making the claim as quickly as possible?
There's a huge work and a lot of analytics. And particularly on the data extraction side of it, that goes into it to allow the insurance companies to make these decisions. Okay, these ones, I'm just going to pay straight away. I've got all the information. I trust the consumer to fill the form. And we've put it all online, we can scan the documents. We can scan the images. They're taken over the asset that's broken, or the insured risk that's broken. We can run that through verification: some of these UI techniques that we're using at the moment, and then we can pay them within minutes or hours of the claim being made.
And then the next one is how we route that through the insurance company's organization. So if it's complex, it goes to the claims adjuster that will manually review it. So there's a whole body of work that goes into that, because the insurance companies recognize if they can reduce the cost of claims processing, that's going to translate into lower premiums for the consumers. If you think about, what an insurance company does, it does two jobs. It takes money from you and pays you money if you have a claim.
So a lot of work that goes into understanding that and understanding how you can aggregate big datasets to give you information to make decisions where those decisions are automated, where you put algorithms into it to pay the consumer immediately or whether they route back into the insurance company's claims infrastructure. What needs a person looking at it to make a decision and what doesn't, I think that's a really easy way of describing it.
Erik: And this is often the case where an insurance company says hey, we want to be a financial service provider, so we're going to basically buy all the technology, and we're just going to handle the financial services and that you had or basically the tech element? Or do they have their own platform that might be in a legacy platform that they've built up internally and want to own and you plug in and fill particular gaps. Maybe it depends to some extent. But what's more common here?
Graham: It depends. We've got some customers where over a period, let's say 15-20 years, they've got their core database, but wouldn't even be able to describe what it's called, but people in industry know what they are, because it's only one company that provided them at the time. And then they build different software components to solve different problems. So we can do this claims thing over here. There's this new underwriting algorithm technology we can use this. And they end up with this network of hundreds of different software applications that solve different things.
The other thing about the big insurance companies is they buy other insurance companies who have repeated that process. So, yeah, a lot of the work is how do we orchestrate and replace a lot of that? You take all these different things and put them into a central location or a central thing, we call it the organizational brain. So we don't look at it like policy administration system or a database anymore? It's like, okay, this is the brain of the insurance company. How do we extract and exchange data of all these different places. It's got to go to ultimately deliver the right data to the right person, the right time and the right format, and then the compliance story for the right purpose. So there's a huge element of organizations where we're helping to do that.
And then on the flip side, organizations that have got cloud delivered software, they're on the right technology stack is how we can exchange data, again, through API orchestration, with some of the insurtech and just making these platforms far more open. So they can look at some of the insurtech technology, some of the investments that are made at the minute in really, really cool stuff. We talked about telematics a little bit there. We'll talk about home IOT in a minute. How does all that work so they can make decisions to on board these technologies without the expense of changing a whole legacy system?
Erik: But let's go into this home IOT. I'm just kind of trying to remember a stat that I read a couple years ago. But if you look at health care insurance costs, what some large majority is allocated to people above the age of 67 years old, so it tends to be in maybe the inverse of what it looks like for drivers. And now, of course, you have a lot of devices that are in the house, I've got a friend that's working on something that basically watches your grandparents and using machine learning to figure out, hey, grandma's walking a little bit slower, there might be some cognitive decline, we have to have somebody go check on it. What does this look like right now in terms of impact in actual insurance, maybe in some decision?
Graham: Well, this is really hard. This is the one that it's hard to unpack. And we'll talk about the kind of three different segments. The first one is if you read an analyst report, you see these 45 degree lines that go like this 45 degrees straight up. That's the compound annual growth rate. And that can be wearable, as it can be devices in the smart home devices. And you look at that, and you go, wow, this is amazing.
And I read a stat that was published by one of the analysts they're talking about now there's 20% of homes that are connected. So there's a hub in the home that can take data from multiple devices. And whether those devices are listening for water leaks, or monitoring temperature, or even looking at people’s movements, which I haven't looked at with an insurance company yet, but I'm sure we'll get to it. It's really, really interesting that so many use cases for this.
What I'm not seeing is too many actual individual use cases for it, where it's deployed for people and I can actually buy an insurance policy based on this data. And I think that's the next opportunity for the marketplaces. So it makes perfect sense, as a consumer living in a normal home the rest of it I can buy a device that sounds an alarm. If it hears the smoke alarm or the carbon monoxide alarm, I can have my front door with the cameras on it, I can have door and window sensors, I can put the device next to my water boiler so it can understand if there's a leak and temperature. This is whole body of work on code machine listening, it's not machine learning, machine listening, glass breaking noises, all these different things.
There's a database out there called Alexandria think it's called, that's got 30 million different sounds recorded classified into a sound library of 1,000 different attributes. So, all this stuff's there. And then you look at what's actually counted when they talk about the wearables the home devices, and it's really difficult to unpick. So where do we start with this one? Do we start with the fact that it's all there? Or do we start with the fact that no one's really sure how to use it yet? And I'm not quite sure which direction to go on this.
Erik: It feels like you need some kind of intermediary. You need like a Google or a Bloomberg. Because otherwise, you have all these OEMs, which are often small businesses or it's Sony or somebody. And they're selling products all over the world and you have all these insurance companies, and you can't have one to one relationships in negotiations between these, right? So you have to have an intermediary who say, I'm going to get access to all this data and then you can plug in and you can we can tap the data feed for three sets for consumer for the week or whatever. Have you seen anybody working on this problem? Or does that seem like potentially one of the solutions to this problem? How do you see it?
Graham: It is. But I haven't seen it in real life yet. We call these exchanges. Again, going back to that connected car, same problem. How do you connect multiple car manufacturers or multiple insurance companies? So you got this the rise of the exchange where the car manufacturers would feed the car data into a central database, and then the access to all that information is equitable for the insurance companies so everyone has got access to it. And that's the way it works. Private companies are doing it.
I was involved with some of the work there at my previous company doing this. And then it might be a legislation driven thing as well where the EU step in and say, actually, we want to have the central database that makes it equitable for everyone to share and use this data.
On the IoT home side of it, I'm sure there's work going on somewhere, but I haven't seen it. And again, we go back to who's actually selling and making these devices. If you look at the two big ones that everyone's familiar with Amazon, Google, that's like 75% of the sales combined. We talked about some of the services there. The technology is all out there at the minute. Machine listening, I'm a big fan of this, because I can see the value of it. But consumers get freaked out by it.
And it's only 2019, I think, it was when Amazon had to change their terms and conditions and say, well, actually, we're not going to listen to you anymore, because they were doing some of the anonymous listening to understand how to build up this library of sound do they can start to look at some of these services, and open them up to their third party developers. And then it all paused because the consumer ultimately got freaked out by it.
So what we're seeing in the home IOT space, if you look at consumers first, high net worth high value clients, and for insurance companies will invest in the technology. And that will be part of a precondition for their insurance. So if you or I owned a $5 million dollar home, and we had a car portfolio of another million dollars, as the insurance organization, I will come and I'll sit down with you on a one to one basis and I'll give you a technology roadmap that includes a lot of these services with the latest alarm technology, etc. And that will be a precondition for me insuring you and your contents and your cars.
So that's the use case for these IoT devices and the technology and always on monitoring of them at the moment. And that works really, really well. How the heck do you take that to mass market? Because now this is all about our home insurance might cost just $500. And as an insurance company, my profit margin on that $500 is a single digit, and now you're asking me to introduce technology that has a cost to it. So you have to be pretty clear on the benefits of it and forecasting the benefits of it.
So if I'm sticking a $50 device into that ecosystem, well, that's largely my profit margin. So I want to know down well that that's going to have a payback to me in reducing my claims costs further down the line, and that's very difficult to monitor and forecast. So for mass market, we're seeing a little bit of experimentation on it. We've got a couple of clients that are baking these kinds of sensor technologies into the policy so we go to it we buy it, we get given a sensor. And at the moment, we're aggregating that data, looking at it and trying to work out is that actually reducing claims in the way that the device manufacturers are saying? Is the accuracy of the devices working the way that's intended to, and trying to build those kinds of datasets that will allow these decisions to be made in the future?
And then obviously, I'm not certain about this, it's probably one of the only companies in the world that's doing this to every single consumer is the one in the US who’re public traded company, so we can see everything and how they're getting on whether they do home and contents insurance. And again, they provide the devices that monitors, what is it, windows opening, water leakage, door alarms, a reasonable service. And temperature as well, obviously, as well. So is the room too cold or too hot? If it's too hot, probably going to fire.
So you're able to play this out in real time and have a look at their loss ratios every quarter and see how they're progressing with it. Unfortunately, they're only writing in two states at the moment, and one of those states has a propensity to wildfires. So you have to like just think about this at the moment, I took a big hit last year. But you can see it. And I think that's really, really important. And that service that I told you about service wrapped around the high net worth clients. The big body of work now is look, how do we take this into the mass market, in the same way as what happened with the connected car story that we started with? And that's where it gets interesting.
And then finally, talking about the IoT devices and the technology there is look, if I'm a consumer, and I buy this, and my home is connected, and I've made a conscious decision to have all these services, what does that tell me about a consumer compared to someone that could not to buy all those services? So we're going back to this body of work that looks at trying to identify who's bought what and using the fact that I have something like this as the proxy for risks. If I've invested into it, if I've got smart home network, if I've got all our lighting, if I've got chime doorbell connected, smoke and carbon monoxide alarms, alright, I might not be connected to an insurance company. But the fact I've got these tells you something about me that is very, very different from the consumer that hasn't.
Erik: On the business side, are businesses at all interested in negotiating and saying, hey, we can certify that, we've got 10,000 employees across the world, and in the buildings where they're active, we can keep air quality levels above a certain degree? One of the things that we're looking at for one of our clients right now is a device that goes over chairs to measure your posture basically, because actually, back injuries are one of the leading causes of workplace injury, and it's often just a chronic, you sit too much, and you sit with the wrong posture and eventually you say, hey, my back hurts, I'm going to make a claim. So, things like this, are you seeing proactive conversations around this data?
Graham: Definitely. And we've got an awful lot of clients in the Nordics, say, Denmark, Sweden, Finland, they're social democracies, and they care for people an awful lot more. So they're using claims data to understand all of this, and then they can start to look at what you just described and lots of other different things to prevent the claim ever being made in the first place. So we're into this occupational health, we call it, and all those kinds of things like that, but it's based on evidence of claim history.
And that's a really good example of what you can do with accessing these large datasets. Now, as you start to see, and you go, actually, this is the biggest cost that we have, these are the biggest problem that we can solve. And then how do we prevent that from ever happening in the first place? And that's all around the use of the technology and the datasets to actually identify issues. And then sure, once you've identified them, you can start taking action.
Historically, all this data points have been in pockets of places and no organization has never been centralized where people can see it. You never get this overarching big picture. Once you've got the big picture, you start to understand that. And if you ever go and have a look at insurance company in the Nordics, huge focus for them and huge investment, they do a lot of group policies, I call it. I think this is where they started from with unions.
So, they have policies that they sell on a group level to individual unions, and then that drives a lot of the activity. Because if you're working standing up a machine all day, what's that going to do to you? Well, we can actually adjust your posture, we can start to practically come in and have a look at your employee health safety, air quality, all those things, and it's very, very much a proactive basis rather than okay, how do we process the claim as quickly as possible? And we're seeing that everywhere now, but they seem to be the leaders in it.
Erik: So we see a lot of interest in businesses that manage heavy assets figuring out how to [inaudible 45:48], let's say, so I build construction equipment and I sell it to a distributor, and then they sell it to somebody, and I have no idea where it is, but maybe I can sell this as a service over a platform and then somebody goes on the platform, and they log in, and they rent it for a few days. But then you have to allocate liability for that period of time. So you have this charge, where I sell manufacturing equipment, and instead I'm going to give it to you. So it's always in your factory, but you're going to pay me per use. But again, now that maybe the machine is not owned by that factory, it's owned by either me, the OEM, or it's owned by maybe a third-party financial firm that's [inaudible 46:27], but do you see much innovation around this in terms of how insurance is provided?
Graham: I see a lot of learning about, we call it Embedded Insurance, I suppose, is the overarching theme to this. So the product comes with the insurance embedded into it. In large businesses, because of the way they insure themselves, it's actually quite difficult to unpack this. So, a large business or self-insure up to a certain amount, and then if you start getting into these things, like insurance for individual equipment, components and insurance for this, that the other, you'll probably end up taking it because it's just a service so you might as well it's easy to buy. But the way, they're actually structured, the way they insure themselves is very, very different from you and I buying insurance. It's a difficult one to unpick how far that's going to go.
The other side of that is more for smaller medium organizations, so let's take that big piece of capital equipment that we're renting. At the moment, if I run a construction company, I'm ensuring myself as if I'm operating those machinery all day, every day. But the reality is, I'll probably use it on a project by project basis for a week at a time. So my insurance premium is sky high, because all I can say to the insurance company is I'm using capital equipment, I'm using diggers, I'm using machinery, I'm using this, that, the other and they go okay, that's make sense. Well, here's your price.
And getting down to that kind of level where you can go, well, today, I'm using this piece of equipment, tomorrow, I'm not. It takes quite a lot of discipline for the end customer to actually go on and say today, I'm using this, today, I'm using that. But again, you're going back to how do we make this a lot easier for the insurance company to understand what's being used, and when.
Again, most of these machines now have the technology embedded in them to extract the data from it. So they're coming equipped with communication, so you can technically see what's going on and how it's being used and where it's being used. Anything we've talked about today can work really, really well on a project by project basis. But for an insurance company and for us, a lot of this is volume-based. How do you make it addressable to everybody, not just a small, nice group that you can tactically manage very easily?
Erik: And then also, I'm sure that there's a certain percentage of small business owners who wouldn't be against kind of figure out how to turn off that sensor so it look like they were never using the equipment [inaudible 49:07] once a quarter.
Graham: You're renting this equipment on pay per use now. So financially, you're paying for the number of hours you use. So it would make natural sense that you're insuring for the number of hours you're using it. So there is an opportunity there. But like I say it's just because of the nuances of how businesses insure themselves, it is difficult to unpick.
Erik: Obviously, data is really at the crux of innovation in industry. Is there I think we've covered a fair bit of territory here, anything we haven't touched on it's either very interesting or very important for us to know?
Graham: When I speak at conferences, and everyone talks about the use of data technology, the rest of it, everyone always talks about Facebook and Instagram and Tesla and the rest of it. And I go, yeah, but to those organizations, I've got the sole purpose of selling you adverts. Just look what we're doing. We're protecting people. We're sharing that risk across multiple countries, industries, etc, etc and giving a consumer confidence to drive a car or small business confidence to invest and the insurance companies got their back if something goes wrong.
So everyone always talks about insurance as being this kind of like monolithic big old thing that hasn't changed for 200 years. And I go, well, that's not actually true. It's a really, really exciting space, and you can go anywhere with it. And guess what? My mother doesn't have Facebook, but she's got insurance going back to where we started. So it can actually touch every single consumer on the planet as well. Because of everything that goes on, there's a lot of technology investment that's being made. And it's probably at the forefront of some of the big themes that we've talked about today.
Erik: No, it's a fascinating space. And I know a lot of the tech startups that I'm talking to are keenly interested in understanding how they're relevant to insurance [inaudible 51:15]. Graham, I really appreciate you taking the time to chat with us. But I think for a lot of folks, this is just maybe touching the tip of the iceberg. If people are interested in reach out to you or getting in touch with Sapiens, what’s the right approach here?
Graham: My email address is Graham.email@example.com. That's relatively straightforward. And find us on the web and find us on LinkedIn, and in I'll be happy to talk. And for your insurtech organizations, we do get a lot of interaction with insurance companies, and they're looking for us to bring technology to them and ideas all the time. So we welcome it with open arms.
Erik: Graham, really appreciate that, we'll put those in the show notes for everybody. Thanks again, Graham.
Graham: Thank you.
Erik: Thanks for tuning into another edition of the IoT spotlight podcast. If you find these conversations valuable, please leave us a comment and a five-star review. And if you'd like to share your company's story or recommend a speaker, please email us at team@IoTone.com. Finally, if you have an IoT research, strategy, or training initiative that you would like to discuss, you can email me directly at erik.walenza@IoTone.com. Thank you.