We’re in the business of making workers’ comp work better. That’s why we’ve always paid close attention to the buzz about the potential for artificial intelligence to transform people’s lives in positive ways. Everyone in the insurance industry has taken notice, as well. In fact,
Its recent report notes that AI will transform “every aspect of the industry in the process. The pace of change will also accelerate as brokers, consumers, financial intermediaries, insurers, and suppliers become more adept at using advanced technologies to enhance decision making and productivity, lower costs, and optimize the customer experience.”
With so many touchpoints across the insurance industry, and so many potential use cases, there is a lot of room for innovation — and there is a lot of room for failure. As an early adopter of AI technology, I will share some of my key insights.
Optimizing for success
In order to get what you want out of a project, you should know exactly the types of business opportunities you want AI to address and why. For example, at MEMIC, we started talking about what we hoped to accomplish with our mountains of data five years ago; we wanted to improve outcomes for all. Breaking this down, it translated to multiple areas: reducing the time to close cases, improving medical care for injured workers, making caseloads more manageable for employees, and achieving stronger financial performance for stakeholders. To accomplish these goals, it was clear we needed to take a holistic, data-driven approach to understand and manage claims.
One of the biggest issues in workers’ compensation is determining if a claim can be resolved quickly or where it might take a wrong turn. Predicting when claims could escalate to the point of taking preventative action could save us millions of dollars each year. We recognized that this kind of output could also be invaluable to our company from a resource perspective. If done well, we could let data drive the process of matching a claim to the best representative to handle it based on complexity as well as use that information to guide a claim rep’s workflow. That kind of insight goes a long way toward advancing our goals.
Leveraging existing investments
There are a lot of manual processes in insurance, and automated AI systems can relieve much of that burden. The challenge is that most organizations are unwilling to abandon existing investments — and for good reason. Aside from not wanting to scrap an application that cost a good deal of money and time to implement and swapping it for something the company is not sure will even work, it’s important that systems are integrated to get the most out of data in the easiest possible way.
Leveraging existing investments also means leveraging your investment in your people. Any new technology should make their jobs easier, not act as one more thing that they have to learn or something that they might be threatened by. In our case, claims handlers were responsible for perpetually digesting new details about each claim in their workflow, knowing when it was appropriate to act, and making sure that follow-ups happened when they were supposed to. With so many aspects going on at once, things inevitably get missed. We needed a system that could alert claims representatives whenever new information became available or when action was required — as well as a system that could recommend what steps they should take.
Finding the right solution
Insurance companies are in the business of accepting and mitigating risks while paying claims when they happen. We’re not technology companies. We may be able to adopt new technologies — and get excited about them — but, in general, we do not have the team or the background to build a custom offering in-house. As such, that means looking for a vendor who can build what we need.
In our case, many AI vendors were focusing primarily on insurance fraud detection. With the operational objectives highlighted above, we also needed a vendor that understood the intricacies of workers’ compensation. This whittled the pool of potential providers down even further. For a while, it was pretty discouraging, until reinsurance broker Aon Benfield introduced us to CLARA Analytics.
From our early conversations, it felt like we had found a partner who understood our business and goals. The vendor team also provided some key insights for additional capabilities that would help us beyond what we had initially imagined. Because of the high stakes involved, we asked for a proof-of-concept project to test things out before signing a contract. I highly recommend this step to any organization contemplating an AI project in order to ensure that what you wind up with does, in fact, do what you want it to do.
The proof-of-concept exceeded our expectations, so we broadly deployed our solution approximately one year ago. It has already achieved significant positive ROI, saving us over $5 million and helping us to become more efficient all the way around. We were right to conduct our due diligence; it has paid off substantially, and I believe we are just scratching the surface of what is possible.
I know AI has the potential to dramatically improve or even save people’s lives with proper application. And, as our experience with CLARA has demonstrated, everyone in the value chain is now using data to drive decisions, increase efficiency, lower costs, and keep policyholders’ and workers’ best interests top of mind. That’s just another way how we’re making workers’ comp work better.
