Since 2013, I’ve been urging the insurance sector to think more critically about its enthusiasm for personalisation. Those questions are now being raised more broadly, yet as is common when change happens, the sector has moved on. The emerging view is that hyper-personalisation is now what insurers need to be engaging on. It’s a view that comes with lots of ethical questions attached.
Let’s be clear about what hyper-personalisation means, by understanding it in relation to personalisation. Personalisation involves the delivery of insurance products and services orientated less around policyholders as part of a pool of risks, and more around each individual risk. It’s about seeing that individual in terms of their own risk profile, their own claims experience and their own characteristics, rather than as part of a pool of similar risks. The logic can be summed up as ‘why pay for the claims of your accident prone neighbour’ when you can have your policy and premium tailored specifically around you.
It sounds great, until of course you need to make your own claim, even if you’re not at fault. There are significant ethical questions associated with personalisation, which I covered in this 2018 blog post.
Into the Near Future
So how does hyper-personalisation fit into this? As the name implies, it takes personalisation further. It is about being highly customised and dynamic in dealing with customers. This is not just in terms of the individual as they present themselves now, but in terms of what they are becoming over time. It moves insurance from dealing with customers in near past to real time, to dealing with them in real to near future time.
Hyper-personalisation is often spoken of in terms of anticipating the needs of a customer and then adapting their insurance to reflect their changing preferences and life circumstances. It sounds very customer centric, and that’s often at the heart of the hyper-personalisation narrative. And of course that’s great, and an area in which insurers recognise they need to improve. Yet the narrative has other dimensions.
Take the hyper-personalisation of flood insurance. The narrative focusses on how customers can be forewarned of impending flood events, so that they can take action to reduce or eliminate any potential loss. It sounds a brilliant way of delivering real value to those customers and communities.
Yet the market also uses that narrative to talk about delivering flood risk mitigation through policy conditions, no claims discounts, increased deductibles, reduced limits and increased co-insurance. This is the language of containing exposures.
And there is a narrative of expectation too. The insurer expects customers to take those mitigating steps, and for those mitigating steps to be effective, in order for cover to be maintained. The actuarial logic is to question why those policyholders who do take mitigating steps should fund the losses of those policyholders who don’t take mitigating steps. Note though that this is about ‘do take’, not about ‘can take’.
You can see how the anticipatory and predictive nature of hyper-personalisation fits into something like flood insurance. The time horizon is a few days, often a few hours. Surely enough time for policyholders to respond, says the market.
Consider other markets for hyper-personalisation, such as life, health and protection. The time horizons are much longer – months, often years. So as your eating and exercise habits are streaming data to insurers, and as insurers gather more information about how effectively your health is being maintained, one side of the narrative will be about helping consumers maintain a healthy lifestyle, but the other side will be about re-calibrating your cover in response to how well you’re doing. For the insurers, this is loss prevention through policy (let's be clear) enforced changes in consumer behaviour.
The logic again is about ‘actual response’, rather than ‘ability to respond’. And while this may cast the insurance sector in a harsh light, the narrative then points out that it is a private market, and why should that market pick up the costs of those who can’t change their behaviours. Yet bear this in mind. That last question is being raised not because the market has become more brutal and unforgiving, but because hyper-personalisation has changed the narrative to make it reasonable to do so. The norms of the market are being changed, yet public expectations of what the market is there to provide haven’t. A chasm of expectation is opening up.
At the heart of the engine room for hyper-personalisation is artificial intelligence. AI is enabling the sector to move beyond static data (the stuff that rarely changes) and into dynamic data, based upon where we are, what we’re doing, what we’re feeling and how we’re behaving. Dynamic data opens the door for insurers to see patterns, sense where they may lead to, and respond accordingly.
And the sector is becoming increasingly open about monetising these opportunities. After all, collecting all this data and building sophisticated AI is expensive. Customer service and loyalty alone does not pay those bills. Underwriting and claims decisions do.
An Unbalanced Power
Where ultimately does this lead the insurance sector? If the future is not personalisation, but the emerging hyper-personalisation, then where is the advantage for the insurer going to lie? The answer for some will be to build even more powerful AI and look that little bit further into the future. After all, this gives you that little bit more time to respond to what you find. Yet like it or not, as you do so, the predictability of events will diminish. The residual that is then left is the exercise of power, of the insurer to collect what data it likes, design AI how it likes, and pursue decision frameworks as it likes. This is however an unbalanced power that creates an existential risk to the sector.
I believe the move from personalisation to hyper-personalisation puts the sector into an arms-race of AI utilisation, yet one in which the advantages become progressive so low as to cast huge doubt upon their pursuit. As investors find the improved loss ratios more ephemeral than they were promised, so pressure builds on insurers to rely more and more on that power, not so much through data and AI, but through how decision frameworks are calibrated around them. As a result, what consumers experience will look little like what they were promised. What consumers receive will be little like what they need. At which point, does that power begin to look rather like that once wielded by Ozymandias?
To Sum Up
Hyper-personalisation will reduce access to insurance and undermine the market’s engagement with vulnerable people. It will seriously reduce the value provided to those that are left. The market is being disingenuous when it talks about consumers wanting anticipatory service, for it fails to recognise what the public actually needs from the product itself. That is a question the market must confront, in order to build a digital future for insurance that is healthy and robust.
How can an Insurer approach this?
A lot of the narrative around hyper-personalisation is being driven by data brokers. They unsurprisingly will tend to take an uncritical view of the ramifications that hyper-personalisation is going to have. Insurers are afforded no such luxury, given their obligations under SMCR and regulatory interest in fairness, access and vulnerability. It is incumbent upon insurers to weight up a data broker’s proposals with a more critical eye, challenging assumptions, calling out simplifications and assembling a less one-sided narrative. After all, what insurer would change their business model with one eye closed?