The vast majority of digital strategists working in insurance see embedded insurance as the future of the market. And some form of it will I believe be influential. For it to represent the future, it has to manage certain ethical risks particularly well. These become clearer when you break down what embedded insurance adds up to.
These people offer a pretty descriptive definition of embedded insurance…
“Embedded insurance works by seamlessly integrating insurance products… into a customer's buying experience, typically at the point of sale. This can be done through partnerships between insurance companies and other businesses, such as retailers, travel companies, or car-sharing services.”
The two aspects of this arrangement that set off ethical warning lights are…
- the seamless integration into the buying experience, typically at the point of sale;
- the partnerships with leading retail brands.
What you create through such arrangements are significant risks around conflict of interest, consent and fairness. And I call them significant because embedded insurance is nothing new. It is a digital form of add-on insurance, the most notorious version of which was payment protection insurance. There you also had seamless integration at the point of sale, with partnerships with leading retail brands. And the result was mis-selling on an industrial scale, generating economy shifting levels of compensation.
That sort of history should cause most insurers to take particular care around the governance of embedded insurance. In part, that’s down to the mis-selling history, but it’s also down to the current regulatory interest in product manufacturers and appointed representatives (AR). Insurers still struggle on the AR front (this for example), but they also know that regulators are monitoring the situation.
All fine then, you may say, but the risk in current governance arrangements could be that they are too orientated around past problems, and not paying enough attention to problems that could emerge in these now firmly digital times. Where I believe a lot more attention is needed is in relation to how the digital marketing systems have been designed and implemented.
These digital marketing systems have taken the place of the bank sales person who was taught to use all sorts of persuasion to make sure their customers bought add-on insurance. This was fuelled by the high levels of commission, something still very present in today’s embedded insurance. After all, why would a leading retail brand allow their point of sale to accommodate something as complicated as insurance, unless there was a lot of money involved?
Let’s go back to basics for a minute. We know that marketing combines knowledge of the consumer with the art of persuasion. Retail firms are designing their online presence to gather as much information about the consumer as possible. Every aspect of those interactions is tracked, stored and interpreted, in order to build personal profiles which are then used to tailor subsequent interactions for ever more profitable purchase opportunities.
While some of that tailoring can be orientated around the consumer's strengths, the really profitable opportunities often seek to exploit the consumer's weaknesses: those times when people are more likely to purchase impulsively. So for example, with studies pointing to Monday mornings as the time of the week when women often feel least confident about their appearance, that is becoming the time when they are most likely to find adverts for beauty products appearing on their device screens.
Context is King
Firms are gathering data not only about how we might sometimes act impulsively, but also about the context around which those impulsive moments occur. Referred to as persuasion profiling, this allows firms to then pinpoint what triggers those impulsive moments and to then have products optimally priced, ready and waiting.
So for example, one trigger might be how many decisions you make in a typical day. The retailer then tracks the decisions you make through your smartphone and presents its product just at the point when you're most likely to feel worn down by all those decisions. This is the point when you start impulse buying. It could be flowers for a friend who you've just sent 'get well' wishes to on social media.
Now, you may ask: what's wrong with all that? It's just the digital equivalent of all that knowledge and persuasion mentioned earlier. And there's something to that argument, but equally, there's a danger that it might disguise something more pernicious. What is different about digital is the extent to which that personalisation and persuasion can be systematised to a vast extent. There’s nothing wrong per se with scale, other than if the interests and incentives of the buyer and seller are not aligned. Exploiting consumer irrationality on a digital scale can generate exponential detriment and so constitute a form of market failure.
Let's bring this discussion round to embedded insurance. It would be quite possible for an insurer to not only use knowledge of the sophistication and resources of each consumer to optimise the price they could be willing to pay, but they could, by adopting digital marketing techniques already in use in other sectors, also use knowledge of when we could be feeling at our most risk averse, to time their personalised advert on our device screen. In other words, to exploit our feelings of vulnerability to increase click through and conversion rates.
Now OK, that is somewhat of a 'pot half empty' view, but it is simply saying that as these practices are now happening in other sectors, we should expect to see them being at least considered by insurance marketing people and in all probability, piloted.
So where does 'manufactured vulnerability' come in? Firms using the full range of today’s digital marketing tactics will know two things…
- what makes each consumer feel most vulnerable to persuasion;
- the context that most often triggers those feelings of vulnerability.
What they will then consider is how to purposefully manufacture that context in ways that increase those moments of most profitable vulnerability.
Just like in analogue times, some of you will be thinking. Yes, but only to a point. In digital market places, the scale and speed at which this can be done is quite different, and so will be the consumer detriment that could result.
Money and Reputation
Why should an insurer take that step from ‘considering’ how to purposefully manufacture the best buying situation, to actually designing their decision systems to bring it into play? Well, there is a lot of money and a lot of reputations being invested into embedded insurance. As I said earlier, it is seen as the future of insurance. Everyone wants it to succeed and for themselves to be top of the pile. Look at this another way – no one wants it to fail, for otherwise, where would the sector’s future lie?
Surely this is a matter of degree though, some of you will be thinking. After all, insurance has often been sold with a bit of an encouraging nudge. And I get that, except the sometimes more than just encouraging nudge is part of the reason why the sector has a conduct regulator hanging over it. We shouldn’t switch off to one person’s encouraging nudge being another person’s manipulating move.
Marketing people have been described as “mostly decent, likable people who want to control us just a little bit”. And by and large, most of us recognise this, playing along to that encouraging nudge from a likable sales person, yet on the lookout for signs of that manipulating move. However, the power of digital marketing raises questions like these:
- when does a nudge go too far and become manipulative?
- who decides what is manipulative and against what criteria?
- how do you control your decision systems to replicate this?
Insurers are not going to see the FCA spending any great amount of time addressing those questions. What they should look out for is the EU taking a regulatory initiative around online behavioural advertising. And the ripples from such a step would quickly spread to the marketing decision systems powering embedded insurance.
Time to Test
Is that someway off though? Not really. The foundations are already being laid, in the form of processing bans imposed EU wide by the European Data Protection Board in relation to behavioural advertising. And as the academic work I’ve seen in the US on manufactured vulnerability is already several years old now, this is not going to be landing out of the blue there either.
The many data governance people working in insurance now need to ensure that their firm’s models have been tested for any signs of manufactured vulnerability. They need to make sure that the levers and dials that could move nudge into manipulation have been set in ways that uphold the firm’s commitments to transparency, explainability and fairness. Difficult? For sure, but easier than the models themselves.
Marketing is often a function that sits to the side of most people’s ethical radars. That’s why the Chartered Insurance Institute asked me to write their guide to the ethics of insurance marketing back in 2016. The insurance world has moved on a lot since then, and insurers need to actively reposition the ethical risks from digital marketing more towards the centre of their risk radars.