Three ethical concerns about embedded insurance

People often think that hyper-personalisation is something that will happen in the future. It’s actually happening now, with a lot of upbeat attention being given to the form it’s taking. Yet it’s also happening in a part of the market that has historically generated the biggest mis-selling scandals associated with insurance. If the sector doesn’t handle this with the right level of critical thinking and challenge, then there’s a danger that history will repeat itself.

The form that this hyper-personalisation is taking is called embedded insurance. And it is being talked about as the big development in insurance at the moment. So what exactly is embedded insurance?

One favoured definition at the moment is that it involves “abstracting insurance functionality into technology in a way that enables any third part distributor (usually product or service providers in other sectors) to seamlessly integrate insurance products and solutions into their own customer propositions and journey.” Phew! A simpler definition might be ‘add-on insurance for the digital era’.

Add-on Insurance?

Is that an unfair comparison? After all, add-on insurance is associated with mis-selling on an industrial scale, payment protection insurance being the obvious example. Proponents of embedded insurance acknowledge that it is not a new business model. What they do emphasise though is that it is the technological integration that differentiates this digital development from its analogue history.

And they’re right – it is a key difference. What I’ve not seen though is how that technology integration is going to stop that grim analogue history being repeated. Sure, the danger is acknowledged, but little else. Given that it is the same banks, motor firms and retailers being positioned as partners for embedded insurance, that generated mis-selling scandals in the past, more than just an acknowledgement of the danger would seem in order. It should be firmly at the heart of how those partnerships are being designed.

Let’s move on from the past and look to the present, with embedded insurance as a form of hyper-personalisation. You’ll recall from my last post that hyper-personalisation takes personalisation and extends it into the ‘real to near future’ time. It is anticipatory, powered by predictive analytics feeding off huge data lochs, to deliver the right product to the right person at the right time. A simple click or two and that’s it.

Sure, that can happen, but it is not the whole picture. Let’s look at three issues: hyper-selection, value and marketing manipulation.

Hyper-Selection

Within those huge data lochs that embedded insurance must feed off will be a mix of signals. Some will say ‘sell, sell, sell to that person’ ; others ‘not with a barge pole’ for another person. Underwriting and sales levers will be set by the insurer and distributor to respond to people depending on their predicted profitability. Hyper-selection and hyper-personalisation go hand in hand.

So what, some of you will say. It’s a private market after all. And they’d be right, but at the same time, they’re forgetting that the issue here is scale. By definition, any form of personalisation works at the level of the individual, not the group. Those ‘in or out’ decisions are made specifically about you, and are being scored continually, both in the context of now and the near future. The question is not whether you need the insurance, but whether the insurer wants your business.

Do you want Insurance with that?

You could be offered insurance when you visit a running supplies shop, but not when you visit a caving supplies shop. When you book to fly with carrier A, you’re asked – “do you want insurance with that?” What about carrier B though, whose record has been more chequered these last few days? The reasons for this – irrelevant. And remember, the risk need not be different. The issue could be around commission. Partners and platforms compete too.

And you won’t be aware of which side of that ‘in or out’ decision you’ll be on at any one point in time. You won’t be aware of just what repercussions buying that bottle of mineral water, or visiting that craft beer shop, had for whether or when you’ll be offered insurance. Sure, that insurance was on offer to you a few weeks ago, but that was before you visited shop X or drove like you did last night.

The narrative of embedded insurance may focus on ‘what customers want’, but the reality is ‘what insurers want’. Embedded insurance is about the customer wants that the insurer wants, and the market should be more honest and acknowledge this.

Simplicity and Value

Embedded insurance is said to be most suitable to policies that are simple, easy to understand and with a simple claims process. That all sounds great, but at the same time, it does raise questions about value. Does the cover provided represent good value for the premium being charged?

Add-on insurance became famous for the huge levels of commission that were (and often still are) changed. Seventy percent was not uncommon. At that level of commission though, the behavioural economists working at regulators like the FCA will deem that to be automatically unfair.

Yet those platforms and distributors expect a fee for access to their customer base and for use of their brand (often more trusted than insurance ones). And in those markets where competition on the main product is intense, the commission from embedded insurance will make an important contribution to that partner’s profit. This will be an enormous challenge for compliance people. SMCR places accountability for appointed representatives firmly on the manufacturing insurer.

I’m not saying that embedded insurance will de facto be poor value. What I am saying is that embedded insurance will have to work hard to show that the customer does receive value. And that is value as judged by an informed customer, not the insurer or distributor.

Value and Context

One final point in relation to value is that it invariably sits in a context. I may need one type of travel policy if I haven’t travelled abroad before, but another type if I make numerous trips a year. Knowing your customer comes from more than just a lot of data points. It comes from having a relationship with them that opens up their needs and places them in a personal context. It’s difficult to see how a policy that asks no questions and which is bought on someone else’s platform with just a click could build any form of relationship. The platform or distributor will control that relationship, and it will be one built around the primary product, not the embedded insurance.

One of the big mistakes in the mis-selling of add-on insurances was that they ignored context. This resulted, for example, in insurance being sold to people who could never claim. And those people who couldn’t claim often only found this out when they tried to make a claim. Could that be resolved through technology? Perhaps, but I’d like to see it evidenced first.

Manufactured Vulnerability

A more subtle (but certainly no less serious) risk that the hyper-personalised nature of embedded insurance faces is manufactured vulnerability. I’ll explain it through the example of you being a nervous flyer arriving at the airport to catch a flight. The airline might pitch some form of life / health / accident insurance to you, perhaps in the form a leaflet in with your ticket. It’s very generic and despite being a nervous flyer, you decide to put it off until next time.

To be honest, that’s a rather old fashioned example, so let’s evolve it a little. Perhaps the insurer has been judging what type of character you are from the websites you visit or the posts you like. So after you’ve checked in, the insurer places an advert onto your social media newsfeed (which you read in a coffee shop while waiting to board) that, well, let’s just say in some way reminds you of your fear of flying and the consequences that have been keeping you awake at night. And so, in response, you click and buy. That’s manufactured vulnerability, which I wrote about in more detail here back in 2015.

All marketing is to some degree manipulative, and we can often see this. The problem here is that again, this is being done in hyper detail at a hyper scale. It is using knowledge of risk and character (both big features of digital insurance) to manipulate the desire to buy. It may be clever, but the public will hate you for it.

Summing Up

What I read about embedded insurance is heavily weighted towards technology and distribution. References to customers invariably start with something along the lines of ‘the customer wants this’. I worry that those have not been fully rounded conversations. Of course customers want the right product at the right time, but in saying that, they are taking it for granted that the cover will be good value and fairly priced.

Proponents of embedded insurance need to challenge themselves more. They need to put the technology aside, put the market opportunities aside, put the eco-systems aside and build a customer proposition based upon value and fairness. Without this, they’re in danger of replicating the painful and expensive history of add-on insurance.

Postscript

A subscriber to the blog has brought to my attention the deferred sales model regulations being introduced by the Australian Government. Also referred to as the anti-hawking regime, it’s being described as “one of the biggest reforms to impact the (Australian) insurance market for a long time”. They will come into force in October 2021 and introduces a four-day pause between the sale of a principal product or service and the sale of an add-on insurance product. Exemptions have been introduced for certain products. More on it here.