How does automation make insurance less of a distress purchase?
Jun 17, 2022 by Jeremy Owenson
We’re living through the rarest of things; a period of rapid change in the industry with a glacial history.
Advances in technology, AI and machine learning have made it possible to set about modernizing outdated traditional insurance practices. A fresh approach like digital transformation and automation broadens our terms of reference for what an insurance product can look and sell like.
Revolutionary thinking such as bringing in automation for things like simple or common products, or claims, can deliver profit and a positive customer experience.
Take non-traditional parametric insurance, for instance. Parametric insurance makes a pre-determined payment for the occurrence of an agreed business or environmental trigger event like Caribbean Catastrophe Risk Insurance Facility (CCRIF SPC), the African Risk Capacity (ARC), or natural agricultural hazards in developing countries. In these instances (depending on the parametric policy) a trigger event such as abnormally excessive wind speed, rainfall or drought generates an agreed and automated payment almost immediately.
Without this kind of automation, the payout depends on the ability of the insured party to provide data-based evidence of their misfortune — a long, drawn out and far from certain process. With an automated product, it’s the other way around. The insurer accesses the appropriate data set, determines the claim and pays out.
Like with Lemonade Insurance in the United States — an organization built high on automation. Built on transparency, a near-enough immediate claims process and lower rates than the competition, Lemonade donates unused premiums to charity each year and delivers an all-round great customer experience.
One thing which might seem obvious but needs flagging-up straight away is that the technology insurers have in place, currently, must be capable of adapting to innovation. A lot of new products being brought to market don't necessarily fit a standard insurance platform.
We need to think about product development cycles; how do we get products to market, quickly? And where we have new propositions and products, how do we develop them on a proof-of-concept basis? How do we test them and get them to customers, knowing that we might have to dump products that don't catch on and industrialize those that do? “Develop fast, drop fast” — that’s our approach to building a diverse product suite.
It’s easy to forget that, once they log off for the day, prominent risk managers become customers again. They have the same expectations and anxieties as their fellow policy holders, and recognize the need for human interaction in certain complex cases. Technology-wise, the important point is to automate just the aspects that improve the customer service. It will also help to calm workforce employment fears of what will happen “when the machines take over”.
The terms AI and machine learning are often interchangeable. Machine learning is basically learning from a human who intervenes in a process that the machine doesn't understand. While the machines are understanding the process through determining a claim or the choice of a product, human intervention helps keep the machine on track. As the technology develops, machine learning is going to have a massive influence on the industry.
More and more tasks will be completed by machines rather than by humans. But that doesn't alter the fact that there are certain circumstances and classes of customer that warrant human intervention. Handling injury claims, for example. No one wants to be told by an unfeeling machine that “we're not paying your claim because…”. We want empathy and understanding, and that means speaking to a human being.
It's about getting the right balance between automating tasks that provide better service, reduce cost and deliver a better customer experience, and retaining staff for the tasks that people are best at dealing with.
Career-wise, one of the reasons I moved more into analytics was that insurance is analytics. Even if all we had were paper rate cards, somebody would still have to perform some analytics there. The truth is, claims departments are going to start recruiting more and more data analysts. Rating and pricing departments are going to recruit more data analysts. Fraud is going to be led by data scientists (not by fraud investigators necessarily).
Therefore, anyone about to make insurance their career ought to concentrate on coming to terms with data, because that’s the future. For sure.
So, we need to home in on the areas that AI and other technologies are not so good at, like customer care. Here’s a personal and practical example from the UK floods of 2007.
A claims manager said to me, “It's really interesting. We've got customers who were flooded out and have come to the end of their alternative accommodation. Alternative accommodation only lasts for 12 months and we haven't managed to dry out the properties yet, so they're kind of left hanging.”
My reply was, “Well, why didn't you just write the house off? If a car is beyond economic repair you write it off. Why encourage a customer to move back into a house that's uninhabitable?” We need to look at the claims process in a new light, and come up with better ways to handle those sorts of claims.
In that circumstance, what if the automatic tool said, “Actually, we’ve worked out that it’s going to take more than 12 months to dry out your house and reinstate you”? And then the insurer gave them a total loss settlement, paid the value of the house on day one and let them get on with the rest of their lives? How much better would that be for all concerned?
Insurers need to focus on giving customers an excellent, personalized, digital experience. We have enough information about segmentation to be able to personalize our offering to them. We just need to make sure we can deliver on that.
Luxoft is an analytics and engineering business. It's very big in banking and capital markets (BCM). Fortunately, we discovered a lot of the things we do in BCM, particularly in the analytics space, are directly transferable to the insurance sector. I joined the company to help build out that insurance practice, taking our BCM tools and showing insurers how we can deploy them in the insurance industry with great success.
If you’d like to find out just how much of a positive difference Luxoft could make to your organization, visit luxoft.com/insurance or contact financial services and we’ll chat through some of the remarkable solutions we have for the insurance industry in general, and for you in particular.