Data and analytics: the problem that insurers will have with identity

  • 29 January 2019

Insurance is changing. It’s moving in new directions, pursuing new ideas, new models, new technologies. All very exciting, but let’s not forget the relationship that underpins the market – that between the insured and insurer. What sort of reconfiguration of that relationship will emerge from all this change? Are we seeing a new dawn, or will trust still be a struggle? There are many ethical dimensions to this, yet underneath them lie some important fundamental issues.

In a series of occasional posts, I will be exploring those fundamental issues. It’s important to keep them in sight because over time, they will be determine how this changing market comes to be judged. They will frame trust in insurance. 

These fundamental issues will influence how we see, how we perceive, how we will end up experiencing the accumulation of all those changes taking place in the market. And understanding them helps us to see through the everyday detail and recognise common underlying ‘change progressions’.

What value is there in this for insurers? Well, by recognising the common underlying progressions of change, we are more likely to see the shape of the new insurance model that will serve the interests of the public and the industry together. In other words, the one that earns the trust that makes it truly sustainable.

And we’ve got to get this right, first time.  Insurance is too important for it to go wrong. The emerging insurance landscape can’t be allowed to develop cliffs for provision, stability and obligations to go shooting off over.  Not when foresight has mapped them out beforehand.

Something Fundamental - Identity

In the first of these posts, I’m going to look at something that is fundamental to all of us: identity. It is a longish post, so please bear with me as I set out an understanding of, and a meaning for, the role of identity in the emerging insurance landscape. 

We know that insurance is drawing on new sources of data, and more powerful analytics, to transform how it understands our lives, homes and jobs, and to transform the decisions that it wants to make about them. That’s the scale of the ambition that much of the sector is now focused upon.

Now, let's examine that 'how it understands’ statement for a minute. The data referred to comes from the many types of decisions we make as part of our everyday lives. This ‘data exhaust’ is vast and incredibly varied, with what at times seems to be no limits: for example, how you smile, or whether you drink bottled water.

Clearly, this is all drawn from a myriad of sources and collected together into what are now referred to as data lakes. These are great repositories of data of all kinds and qualities, pooled together, waiting for great insight to be drawn forth from it. That insight is invariably shaped around identities for each of us. Yet these identities don’t of course emerge on their own. They emerge through the analytics that insurers deploy to make sense of their data lakes.

Consider the Kelpie

I'm minded at this point of a ‘kelpie’, a water spirit of Scottish folklore, typically taking the form of a horse. A pair of them have been turned into a magnificent sculpture near Falkirk in central Scotland. Technology has allowed modern artists to give shape to giant steel representations of these mythical creatures.

Over 900 steel plates shape the kelpies, supported to over 30m in height by internal beams and wires. Sounds rather like those identities that insurers are creating for us – thousands of data points held together by the underlying analytics. One gives shape to the kelpie horse, while the other gives shape to an identify.

Yet if you think about it, those identities are no more real, in this world, than those kelpie horses. Both are constructs of our technological / mythological imaginings. We can build both of them, but that doesn’t mean that they bring what they represent into a present day reality. We may view both with an anticipation of loss, but the dangers they represent can be easily shaped and reshaped.

Indeed, those identities, and those kelpies, can be shaped and reshaped according to our current strengths and weaknesses, our own inner hopes, emotions and fears. And if at this point, you’re wondering whether this is becoming too detached from business, always remember that it’s hope and fear, and many emotions in between, that drive markets.  

I won't pursue this tale of data lakes and equestrian water spirits any more, but what I’ve tried to illustrate through it, is that data and analytics give shape to a digital identity for each of us within insurer systems. And that shape, and how it is perceived, is in fact no more real than those kelpies. Both are digital constructs seeking to represent a reality that can never actually be grasped.

Proxies haven't gone away

Why then can't data and analytics capture reality? Because that data given out as we live our lives, is in fact just proxies for those lives and how we live them. They're just numerical representations. They are not the life we actually lead.  

Consider those two earlier examples. I may buy bottled water because there’s no tap water available, or stop buying it because of environmental concerns. I may smile because of something the camera can’t see, or stop smiling because I don’t care for who is taking the photo. 

OK, so why does this matter, you may ask. It matters because as those proxies grow from a few dozen some twenty or more years ago, to many thousand micro-proxies now, there's a danger that insurers might think they are capturing more than they really are. What they are capturing and constructing is a digital identity for us. What they are not capturing is a digital identity that is us.

Each individual piece of data in that digital identity is shaped by how it is collected, by who collects it and by how it is bought and sold thereafter. And all those individual pieces of data are together shaped into a digital identity by who is choosing them, who is assembling them and the purpose for which all this is being carried out.

So, again, why does this matter? It matters because insurers will increasingly be relying on those digital identities to make marketing, underwriting and claims decisions about us. And those decisions will influence what is on offer to us, the price of the cover on offer, and how our claims are then settled. The insurable risk, the actions taken in relation to that risk, and the consequence of those actions will be judged in relation to that digital identity, and, we will increasingly find, little if anything more.

Surely though, as more data and better analytics are applied by insurers, the differences between the real us, and that digital identity, will narrow and even disappear? No they won't. More proxies do not deliver more certainty. They just break things down into smaller units.

The real you is not made up of data

And this then puts greater reliance on the re-assemblage task performed by the analytics. Yet at the heart of that analytics aren't one to one connections. It's a complex web of statistical correlations. And every correlation, by its very nature, in shaped by a deviation.

Yet won't more correlations iron out those deviations, bringing your digital identity closer to the real you? No it won't. The real you is not made up of data, yet that digital identity originates in nothing other than data. And the analytics used to assemble that data has been chosen because it makes best use of that data.

The data and analytics are welded together in a dance of certainty that is mutually self supporting. That the real you may be made up of non-data elements is something that the data and analytics cannot recognise, for it sees only what it is capable of looking for, that it is capable of utilising, that it is capable of welding together. The model's lens is at best partial, at worst distorting.

What this adds up to is a disconnect between 'insurer reality’ and 'consumer reality’. And that disconnect will I believe slowly and surely over time distort relationships between the two. If it becomes cemented into how the market works, that distortion will create a permanent resistance to trust in the sector.  

As insurer after insurer see their customers through the lens of bigger and ever more complex data analytics, the more that lens will put distance between the two. Insurers may refer to it as personalisation, when it is actually happening is individualisation. The person’s been lost.

I believe that those small distortions in how the market sees its customers will evolve into one of the defining challenges facing the digitalisation of insurance. Handled badly and the public will experience a market permanently out of kilter with its own reality. Done right and the public will experience a market that recognises them for who they really are, in ways that they understand. The decisions now being weighed up by digital insurance strategists will bequeath a significant legacy for the sector's future.