The insurance industry is in big trouble. First of all, the trillion dollar business suffers from a very bad image. Consumers don’t trust it when they take out a policy and begin to hate it when they really need to rely on its services. We have all been there. In a recent global survey from accounting firm EY, consumers ranked insurance below banks, car manufacturers, online shopping sites and supermarkets for trustworthiness.
But that is about to change. The insurance industry we know will disappear in the coming decade.
Insurance is the perfect industry for AI disruption
Insurance is the business of uncertainty. It pools many people’s money in order to alleviate harm when the unforeseen happens. For centuries, insurance has suffered from terrible misalignment. For the consumers it is intensely personal, often to be relied on in situations of distress and misfortune. From the insurer’s side, it is a data-driven enterprise, impersonal, cold … a game of numbers and probabilities. Because of the former, people hate it; because of the latter, it can easily be replaced by artificial intelligence: better, faster, more efficient, and much cheaper. All you need is a bit of AI, behavioral science, and the guts to take on Goliath.
Think about it: what does an insurance agent do? He or she explains the different options and prices, then signs you up for a particular policy. By making the choices clearer and structure insurance in a way consumers easily understand, all this can be done on a website or an app. There is really no role for people here. One might argue that policies have become so opaque because insurers have an incentive to make them so. No one can read through 200 pages of detailed provisions–but everyone gets really pissed when it turns out the policy doesn’t cover what you thought it did.
The second part an insurance company does is evaluate your claim. This seems like a personalized and tedious process, but it is not. In fact, claims invariably fall into certain categories, exhibit certain common features, can often be verified through simple data checks, and so on. Payout calculation is inherently a process best done by a computer–it already is, and can easily be completely automated.
The reason insurers are resisting this transformation is that since they have to pay, they have an incentive to question claims, delay payout, and generally behave in a way consumers often detest. Opacity is in the insurer’s favor. If policies were really transparent and intelligible, they wouldn’t sell.
The second problem is insurance fraud. But contrary to what insurers claim, simple fraud algorithms now catch 95% of fraud cases before a human has to intervene. So there goes the last part of the employee pool.
The future of insurance is fruity
What then does the insurance of the future look like? Simple. The consumer uses an app or website to select options from a menu, calculate a premium, and sign off (perhaps by fingerprint) on the policy.
When bad luck strikes, the consumer uses the same app to file a claim. The claim is verified and checked against fraud algorithms. If a red flag comes up, a human intervenes. If it doesn’t, payment is authorized. The more cases are successfully completed, the smarter the AI behind it.
If you think this is far in the future, think again. It already exists. Go and check out https://www.lemonade.com/. The company is working with Dan Ariely, a professor of psychology and behavioral economics at Duke University, “to take antagonism out of its relationship with customers.” Which is, let’s face it, a big deal and a good thing. We are actually disrupting an industry that is hated by its consumers like no other.
Lemonade set out to create algorithms that make it easy and quick to sign up and approve claims – in minutes rather than days. By automating the service as much as possible, the company hopes to keep costs low.
According to their website, Lemonade is able to screen applicants or claims quickly because its software can pull data and cross-reference information about a particular home or neighborhood from a variety of sources. This reduces the need for the company to ask a lot of information from customers. Many countries already have the databases insurance companies consult, yet we are still paying an employee to cross-check information a computer can compare in milliseconds. Much the same is true for health or car insurance. The data is available. Human fact-checking is redundant in most cases.
To start with, Lemonade, in a stroke of genius, created two chatbots monitored and guided by actual employees. The algorithm will learn from observing the actual interactions with customers; service will improve over time. The two employees are available to speak with customers by phone and handle more complex cases. The longer the system runs, the more data it will collect, the better it will be able to handle even complex claims.
An estimated 80% of the cost of policy goes to paying for the operations of an insurance company. A startup in Germany I consult for and which is soon to go public with its health insurance app, claims that a 300 euro a month health policy shouldn’t cost more than 40 euros. Just imagine the impact on consumers when algorithmic insurance takes off. We will be able to afford so much more coverage, and we don’t have to deal with slimy claims adjusters who doubt every word we say. In a world of rising health care costs and discrimination on the basis of genetic information or health records, it is important to lower the cost where it is easiest: in the corrupt insurance industry. All other forms of insurance, from life, to travel, will follow in this path.
Unlike, say, the legal profession, I don’t see any part of the insurance business that isn’t better done by sophisticated computers. Insurance companies will have to adapt.
Just remember, if life gives you lemonade … change your business model.