Here at the SENSE Consortium, we are passionate about the topic of sensors and real-time data and its many benefits for the insurance and risk management industry.
What is our mission?
Using internet of things sensors and real-time data, to halve the cost of losses through prevention, thereby keeping companies, factories and manufacturing open and adding to the global economy - all while enabling insurance companies to offer better prices and services to their customers and improve ESG corporate responsibility.
We are seeing an increasing number of use cases in commercial properties and yet, the scaling and overall traction of this technology and data is still in its infancy.
The challenge
Imagine you are:
Head of Product Development at an insurance company,
And...
Your objective is ‘Implement IoT technology into Commercial Property Insurance’.
What are you going to do?
Benefits
For IoT and real-time data initiatives to be successful, any proof of concept or indeed scaling needs to include numerous (and typically) siloed functions as well as driving in-year and recurring benefits.
Before tacking where these benefits arise throughout the paper, it is worth highlighting the main areas to cover in any insurance focused business case:
Reduce the Loss Ratio
Grow the Commercial Property book
Improved Commercial Property book segmentation
Greater per customer and portfolio level real-time analytics allowing better decisions
Improved Customer Experience & getting closer to the customer (cross or upsell opportunities)
Reduced new acquisition costs (Decreased Loss Ratio and Retention => Lifetime Value)
Addressing the climate imperative (ESG)
Beyond the benefits for the insurance industry (Brokers, Carriers), the benefits are even greater for the customer themselves… but that’s a topic for a follow up paper.
Strategy
Before embarking on your IoT and real-time data journey, it is important to take the time to consider where you will see and add value from IoT and real-time data. Understanding the application of the data you expect to get will significantly impact how you will need to shape your IoT initiative.
One example: where on the value chain should you ‘play’ or explore IoT and real-time data? Generation (own and deploy the sensor), collection (capture all sensor events for one type or set), consolidation (capture of multiple or related events from different sources), consumption (using that data for decision making or insight). Taking it one step further, what would an ecosystem of data, insurance and services look like and how achievable is it?
As an insurer, it is unlikely to be in the generation or collection of data; it is more likely that you will consume data someone else generates, captures or consolidates. It is yet more likely that as an insurer you would orchestrate an ecosystem around data with new products, services and business models.
Depending on where you then see your role, this will dictate your focus, likely budget, partners and outcome hypotheses.
Approach
So where to start? Let’s look at the people, the process and how best to proceed.
Arguably, new and innovative projects start and either fail or succeed with people; the internal and external stakeholders that need to be actively involved and managed to stand a chance of success.
Internal Stakeholders: Buy-in is needed from a wide spectrum of (senior) internal stakeholders to have any chance of launching. The more time you spend understanding your client base, your business drivers and those of your internal stakeholders, the more valuable your business case will be. The key is to identify really clearly the outcome and decisions and business models you expect to support by having this technology and the real-time data for those stakeholders (and end clients).
You can then decide what are the most effective ways to achieve that goal - who is your sponsor, what is your governance framework, who are your influencers and so on.
A key success lever for IoT and real-time data is the skill needed to bring several siloed teams together to support a successful launch and scaling: the classic ‘what’s in it for them’ narrative. Fortunately, the use of sensors and real-time data have a lot of favourable and exciting aspects to a variety of stakeholders that have already been proven. Many aspects of such a launch have been de-risked through multiple use cases; it’s more a question of assessing specific organisational cultures and internal strategic imperatives and thereby which direction and options to take.
External Stakeholders: It is equally true for external stakeholders that you need buy-in. This time from both brokers and clients to pilot this (and a lot of clients for scaling!), alongside new ways of working in a partnership model with technology / data providers. Much preparation is needed with your internal stakeholders for these new types of relationship and knock-on impacts IoT creates with external stakeholders. With multiple partners in an ecosystem, the value proposition has to work collectively as well as for each individual external stakeholder. ‘What’s in it for them’ gets bigger with more networked touchpoints.
Building excitement and getting your IoT initiative up the prioritisation list! Prioritising is a tough one, especially during early scaling, in order to make IoT a priority over and above all of the other initiatives within a business - for both internal and external stakeholders. The imperative here is to have a clear strategy. You will have defined the ‘what’s in it for me’, a benefits case and engaged and managed your stakeholders often!
Decisions on which technology and what kind of clients and buildings to go for during the pilot and the early years of the launch, and preparing internal stakeholders for the launch (i.e. being ready for the type of data coming in and what we will do with it.) should all be aligned to your strategy.
Insurance Company - Internal Stakeholders
So, who are the main internal stakeholders or functions that need to be engaged, consulted, involved and managed?
The Pricing Team:
The most obvious objective of the pricing team in your IoT initiative will be to create risk models that include IoT rating factors. Counterintuitively, whilst real-time data is prolific and constant, your pricing team will still be of the mindset of having enough (historic!) data to create their models as they know how to do and which will be many years down the road (the initial IoT portfolio will be small, risk modelling uses several years’ data, time is needed beyond that for claims to develop).
Fortunately, there are some quick/mid-term wins in the pricing domain…
Commercial lines pricing has a relatively strong experience rating element (deviating the risk estimate away from the cross-client risk models, towards the client’s own experience). Therefore, once a client has a history of reduced losses as a result of their sensor data and ameliorated risk management behaviour, their premium reduces even in the absence of IoT specific discounts.
Having a straight discount for the take-up of an IoT policy is an option. This may be from a development budget, or may be justified from short-term, life-time, and/or cross-IoT-portfolio insurance risk savings.
Underwriters’ pricing override adjustments can be put into the pricing algorithm from launch. These will allow an underwriter to adjust the modelled estimate of risk for an improved best estimate of risk, based on say: a) underwriters’ knowledge that a sensor reduces risk, or b) feedback from a risk engineer that a data stream is likely to reduce risk, or c) an ad hoc pricing investigation (which can be performed faster and earlier than a full risk modelling exercise) that shows a reduction in risk for one client or one type of IoT data feed.
Take a lead from telematics in motor insurance where some products have their perils priced using ‘miles’ as the exposure measure as opposed to the traditional ‘vehicle year’. Apply this approach to commercial property and we could see fire split into electrical fire and non-electrical fire, with electrical fire using ‘Megawatt Hour’ as the exposure measure. Something similar may be possible with water consumption and the escape of water peril. Thereby increasing pricing accuracy, and also increasing the client and insurers’ ESG metrics.
Making clients and the wider internal and external organisation aware of these points will be of great benefit to the success of your IoT project. One for the marketing and PR teams!
The Underwriting Team:
Underwriters want to add value in the estimation of risk and new pricing options and models are likely to be a big IoT selling point to underwriters. Whilst top line may reduce, the impact will be felt on an improved bottom line and improved loss ratios. A reduced price (where it is justified) could also help get a deal over the line.
A real-time data feed means there is opportunity for underwriters to make decisions more frequently than once a year. However, to do so will require new thinking on how to accommodate this into the day-to-day for an underwriter and the tools and data feeds and insights necessary to make those decisions.
This real-time data feed with appropriate risk insights is invaluable for an underwriter at a single risk as well as a portfolio level. Analysis of this type of data at a portfolio level should provide interesting insights and trends for further investigation and risk improvement behaviours.
To take full advantage of real-time data with mid-term underwriting decisions, underwriting may want to consider adding new clauses into the policy to incentivise prompt action to mitigate or reduce risk. The client may otherwise be slow to implement a risk mitigating change much earlier than the renewal and if they are only incentivised by the promise of a more favourable renewal premium and/or where they carry only a small percentage of the risk in the form of a deductible.
Improving risk management behaviour in clients is a challenge with few clients implementing risk recommendations promptly based on annual surveys (if indeed a survey has taken place) [see our other white papers for the statistics!]. This new technology is one way to monitor improved risk management behaviours.
A learning from the use of telematics in motor insurance showed that the greatest benefit came from underwriting. Being able to quickly identify those drivers most likely to have large losses (there was a pattern of large losses being caused by drivers who exceeded the speed limit by >30% just days after passing their test!) and remove them from the book. While it is unlikely that you would want to cancel a policy of a pilot client when launching IoT in commercial property insurance, it may be worth considering whether the policy wording could include the mid-term removal of elements of risk that IoT highlighted and where the insured client could/would not put in place mitigative actions.
Making efficient use of IoT data will involve underwriters in a partnership approach: working more closely with clients, brokers, risk managers and other internal teams. How these interactions will work optimally in the real-time world needs to be considered.
We have noted above that through a reduction in loss ratio, there may be a reduction in premium for commercial property policies. Whilst this makes the book more profitable, consideration should also be given to:
New risk management / service / ecosystem offerings if this does not already exist. Complementing the technology with human risk management expertise, new services or an ecosystem of solutions at a single site or portfolio level.
Cross-sell opportunities, particularly in the area of Cyber if IoT devices are not effectively installed and securely. A future white paper will cover this topic.
The Claims Team:
Arguably the biggest impact of using IoT is the potential to reduce the frequency and severity of claims. Reasons include:
Better risk management (insured’s own data): For example, the insured receives real-time and rule-based alerts to changes in data (such as temperature fluctuations). Issues can be identified quickly and action taken to mitigate if required. Appropriate controls and response time service level agreements (SLAs) can be factored into determining an insurer's liability.
Better risk management (sharing of intelligence): In the longer term an insurer will accumulate data from similar sensors from a number of clients. Risk patterns may be detected that are not obvious from a single client’s data and can be fed-back for the benefit of multiple clients (as well as monetised).
Reduction in claims frequency as well as severity due to actions when a claim is imminent: e.g. Identifying power surge levels and durations that are certain or almost certain to cause a fire and instructing or automating a shut-down of the power when this occurs.
Reduction in claims severity due to actions during a claim event: e.g. Water flow devices on pipes can detect water flow patterns that are typical of a leak or overflow and automatically turn off the flow, thereby limiting the damage.
Faster claims notification: Having real-time data that includes patterns showing a claim event is (almost certainly) in progress means that many claims will be notified instantly and automatically.
Faster claims estimation: Similarly, real-time data will give an idea of the severity of a loss before the arrival of a loss-assessor and even before the initial communication with the client regarding the assessment. For smaller value claims, there is a potential cost saving for paying the claim based on the data alone rather than needing an in-person loss adjuster.
Because IoT makes it easy to quantify a loss, there may be a reduction in lengthy/costly negotiations between a loss adjuster and loss assessor (where appointed). Ultimately, that means the Insured is indemnified faster where a claim is valid.
Reserving and Capital Requirements:
Lower capital requirement on historic business
Faster development of reserving triangles: with faster claims notification and estimation, reserving actuaries can more accurately estimate a portfolio’s ultimate claims sooner.
...This means there will be lower volatility between predicted and ultimate claims on all ages of historic policies.
...And less uncertainty that the reserves are inadequate.
...Which gives the insurer a lower capital requirement (to cover this uncertainty).
Lower capital requirement on new business
The main benefit of IoT technology in insurance is to bring down claims
At the same time, and to a lesser extent, insurance expenses are likely to increase. e.g. cost of collecting/storing/analysing/acting-upon the real-time data. [Cost of supply and fit for the sensors is still up for debate as to who pays for this insured or insurer].
...So a lower proportion of the premium is for claims.
...So less uncertainty, as claims are more volatile than other expenses.
...Which gives the insurer a lower capital requirement.
IT, Information Security & Data Management Teams
As a general point on data capabilities; be honest in your data maturity as an organisation. For example, there is little point using IoT data for real time decision making if the bulk of your decision support capability uses excel. IoT has the potential to accumulate large amounts of data incredibly quickly, the more you store the more it costs you to store and to leverage so having awareness of what you have and on what basis is important for cost control (especially as most platforms are scalable in the cloud). When thinking about IoT initiative, like all data projects, what do I need, at what level and for how long are all valid questions
Buildings are inherently complex where often the same underlying technology is applied in different ways from building to building. The data is prevalent from both existing internal systems and new sensors installed to surface key metrics. It is the interpretation of that data that is key: identifying the key trends or drivers in the data as it relates to risk (as opposed to operational data). For example, increased humidity in the Air Handling Unit (potential escape of water risk) or an energy draw from the Fan Coil Unit in a HVAC system is a potential fire risk.
Building experts from broker or insurer risk engineering teams, or clients’ own facilities management teams can help to explain what the data is telling you. It really is a collaboration to understand the data - data scientists alone will not be able to correctly interpret this data without the building expertise input. In the short term, this insight can help identify the generic loss events that help deliver on the wider benefits of sensor technology and real-time data in commercial property. In the medium to long term, it is likely that machine learning and artificial intelligence will help automate some of that data interpretation.
So yes, new real-time data and technologies hold multiple benefits, and it is far from ‘an easy sell’. Mapping the journey of the value of the data; how it can be identified, captured and aligned to the risk management, underwriting, pricing and claims processing functions is key. Our experience at SENSE has started to break this down into the key areas of value for each function and where the value is and how it can be applied.
Other considerations… At this point, there are no mature standards for devices, data nor interoperability of devices. Therefore the ability to quickly on-board every type of data-feed from every sensor is unlikely to be a reality for some time. There are emerging data standards around commercial property e.g. BRICK.
IT/MI system builds usually require a very exact specification. The conflict between the wide world of IoT and the narrow specification of IT/MI will need careful consideration, management, and compromise. It will however be possible to consider the data feeds from all preferred sensors and consider the ease/difficulty of expanding systems from one to the other whilst deciding which clients and sensors are suitable for the IoT product.
Environment, Social, and Corporate Governance (ESG)—Senior Management:
ESG is on the agenda of every Board of Directors and Executive Committee up and down the country. For all the risk benefits of IoT, reduction in claims, new data insights, the oft overlooked benefit is that of ESG. The scope is vast from optimising energy usage, providing a platform to capture / measure carbon footprint to hygiene / chemical / air quality monitoring. This is a key lever in your business case for both the insured and the in
Senior Management - Core Message:
The core message to get across to the CEO is that by implementing IoT in commercial property insurance we can (measure in brackets):
Reduce the Loss Ratio (GLR)
Grow the Commercial Property book (GWP)
Improve Customer Experience (Retention)
Improve the Client Quality (GLR and Retention => Lifetime Value)
Addressing the climate imperative (ESG)
External Stakeholders
Ecosystems
When implementing IoT initiatives, the number and type of relationships changes. It’s no longer linear (client to broker to underwriter) and it’s no longer based on inception/ renewal, endorsement and claim interactions.
New partners come into play for example, sensor providers, insight and analytics partners, risk engineers, possibly maintenance partners to repair or be on hand when an event is triggered (obvious examples being plumbers or electricians). You begin to create an ecosystem where relationships and data are connected and where the appropriate permissions and security are paramount to support the shift from static to dynamic data and new insights. These new relationships need to be carefully managed and consideration of the ‘what’s in it for me’ for all the ecosystem players is critical.
As you progress on your IoT journey and build your own proof points, you may start to involve more than one sensor provider, more than one data set and that’s where it gets really fun - how to aggregate data effectively, what platforms are needed, which data is meaningful from a risk perspective and so on. This may also then shift to the ability to create a digital twin of your risk.
A small word of caution - it’s imperative that these ecosystems and platforms are built with scale in mind as it is clearly not cost effective to build something per client and neither could gain any portfolio insights on a per client platform basis.
Customer (Experience & Interactions):
As noted, IoT significantly increases the potential number of interactions with each customer. Today, Without IoT there may have been visits to a few sites just before NB or a Renewal year, some discussion about actions to reduce risk on a site, decisions taken and perhaps an endorsement to the policy. With IoT the discussions about reducing risk, and therefore reducing the customer’s premium (and of course likelihood of the emotional stress of a loss event), has the potential to happen any time there is something to communicate from the data. From the insurer's perspective risk can be reduced and persistence increased, but to do so the insurer needs to allocate more resources into relationship management and into the supply of IoT Information to those relationship managers.
Data and insights flow constantly, alerts based on data changes need to be sent and action taken so the interaction across the ecosystem greatly increases.
Additionally, previous research undertaken by Intelligent AI from over 5,000 risk engineering visits showed that for many insurers fewer than 1 in 7 major risks identified through the visits are resolved within 5 years. To find out more, read Intelligent AI's survey on the need for real-time risk data in Commercial Property Insurance.
Regulator:
The insurer will be actively managing risk, which will be looked upon favourably by the regulator. There could be an argument for lower capital requirement on the IoT property portfolio.
Public Reputation:
Being seen as an innovator is good advertising and therefore boosts the insurance company’s entire portfolio. Specifically to the IoT.
Prioritising – Where to focus IoT efforts
Every client will have a different risk profile, a different weighting of perils and coverages that they are exposed to, a different commercial building they are in (often of a type correlated to their industry), and therefore different sensors that are most useful to them. Some may have a number of sensors already installed and perhaps be experts in the data these sensors produce (though they may have only viewed the data from a production perspective; not an insurance risk one.)
A good choice of client and sensors for the pilot is not just defined by the two being most likely combination to demonstrate a reduction in risk; they also need to be representative of a broad section of the commercial property insurance market, to push the scalability of IoT use in commercial property insurance and show the project can be successful in the longer term. The people involved themselves also need to be comfortable not knowing all the answers and be willing to collaborate and learn from each other. Many a project has failed simply because of this people element alone!!
To solve the question of how to prioritise and put the client/ sensor combinations in order of suitability for piloting, a good approach is to first categorise and prioritise the various client and risk features.
Perils – Gross Value of Claims Incurred:
In terms of gross value of claims incurred (as opposed to number of claims notified) Fire is the biggest peril in Commercial Property. There are also a number of sensors relevant to Fire: power surge detection, smoke detection, heat detection as examples.
Catastrophe weather events in combination account for a significant proportion of claims. However, there are several different catastrophe perils; IoT sensors in a building are not as relevant to detecting external weather events as they are events inside a building, and even if you can detect an external event starting to happen, you are generally less likely to be able to reduce its impact as much as is possible with internal events. For these reasons, catastrophe perils should be the lowest in the priority order for the deployment of IoT sensors. Though that does not mean they should be ignored entirely, advanced detection of a flood could mean being able to move equipment for example.
Of the non-catastrophe perils, Escape of Water (EoW) is the second largest. This includes Freeze claims as the detection and action taken to mitigate it is the same as non-Freeze EoW. Sensors that are relevant to EoW include: detection of water flow rates in pipes, pipe temperature devices, smart sprinkler systems, actual dampness/ leakage detectors in places liable to flood.
Although Theft is in third place by non-catastrophe claim amount, there are plenty of relevant data sources and many are widely used and/ or widely applicable: intruder sensors, security cameras, smart locks, clocking in/ out, counting of produce. However, technology to reduce Theft has been around longer than technology to reduce EoW, and thieves fight new technology more intelligently than water does, so Theft is on average likely to be in third place in terms of how much IoT sensors can reduce the gross value of claims.
In summary the suggested ordering of perils for the reduction of gross claims by using IoT and real-time data is as follows:
Fire,
Escape of Water (inc Freeze),
Theft,
All Others.
Perils—Claims Frequency:
There is however a potential problem with the above view of claims when piloting IoT use. Although Fire is the largest peril by gross value of claims incurred, it is lower than Escape of Water and Theft in terms of frequency. If the pilot must yield a conclusive reduction in claims for the wider IoT project to proceed then Fire may not be the choice peril. It will take a higher number of properties in the pilot to show a reduction in claim numbers for Fire, given that there are fewer claim events to start with.
Theft frequency has a strong correlation to industry and will vary significantly by client. With a bit of luck and a lot of thinking, there may be a sensor to add or an existing data-feed to use that would make a difference to a client that has a particularly high Theft frequency profile. If this is not possible, then Escape of Water is generally the peril with the highest frequency, there are many relevant sensors available, some with a history of use cases.
Coverage:
By GWP, Buildings is the largest coverage, followed by Stock, then Business Interruption. It was noticeable in the recent SENSE use case seminar that Business Interruption got a significant mention. Perhaps this is because when the companies involved fitted sensors (which then got used in the insurance use case) they did so for continuity of production, and so the sensors involved are naturally aligned to Business Interruption. Perhaps it is because those companies fitting sensors have a higher proportion of Business Interruption risk, which is why they are more prone to fit such sensors. Or likely both reasons. The conclusion: don’t forget Business Interruption!
Client/Building:
A generally good starting point is: go large!
Devices measuring water flow do not work well for household insurance as the £200+ cost of supply and fit cannot be clawed back in the expected lifetime of a customer whose EoW saving to burn-cost is only likely to be circa £20 p.a. Commercial buildings however tend to be larger and have much higher premiums (higher EoW burn-cost and thus higher potential savings from IoT) than seen for residential property, yet the supply-and-fit cost of a device remains broadly the same. Therefore economies of scale make such devices viable for commercial properties.
There are going to be countless other considerations when choosing clients and buildings for the IoT project, including:
Large through having lots of small buildings, or a small number of large buildings. The former may fail on economies of scale if lots of sensors need fitting, the latter may involve buildings that are too unique to provide a proof-of-concept that is readily expandable to a significant proportion of the commercial property book.
Devices already fitted. This means a) Money is saved on supply-and -fit b) already having data allows a great deal of assessment as to how useful the data is for insurance purposes before even taking the client on in an IoT pilot.
How keen the client is to work with the insurer and to take action to mitigate risk when it is identified.
The risk profile of the client.
How expandable any pilot is with this client is to other clients.
Speed of implementation:
We will always need to consider how quickly we can set up a client on the IoT product alongside the likely usefulness of doing so.
Investment portfolio Vs User owned building:
User owned buildings are likely to be the best for an IoT pilot as there are fewer parties to engage with and there is a fuller exposure/ownership of risk by the client.
Conclusion
Justifying the use of IoT to the point of getting sign-off for a pilot is the easy part. There are many internal and external stakeholders that will be excited by the benefits and opportunities that IoT technology brings.
The hard parts are:
Getting all stakeholders to optimally implement their use of real-time data.
Gaining sufficient traction throughout the course of the pilot that it is successful in proving the insurance benefits of IoT.
Finding ways to make the project expandable across a wide range of commercial properties.
There is no one killer sensor and data-feed that will instantly move commercial property insurance to being IoT based. Sensor’s need to be considered building-by-building, client-by-client, and industry-by-industry.
There is reliance on the clients that will participate in a pilot, and on the sensors and data they already have. This needs to be considered alongside the urgency of proving results in the pilot. e.g. Is there a possible quick win on Theft? When should the longer game be played on Fire? In the absence of client-provided sensors, are there sufficient economies of scale to justify the supply-and-fit cost of flow devices for Escape of Water?
There clearly are significant reductions in insurance risk to be gained from good use of IoT technology. But, even more so than in the use of telematics in motor insurance (where the devices and technology available all give fairly similar data), there are many different potential paths to go down regarding how to structure a product, what sensors to use, and how to make use of the data. Therefore, a successful approach is likely to involve a lot of case-by-case decision making, human interaction, regular review of results, and an expandable structure which is broad enough to accommodate a wide variety of real-time data sources and application of that data.
For your chance to hear more expert insight on IoT and sensor technology, register for our upcoming event:
Comments