Extract value from unstructured data to control exposure, augment capacity, and win more business

07-05-2021

Extract value from unstructured data to control exposure, augment capacity, and win more business

Some challenges can seem insurmountable—a complex product requiring a detailed and expert eye surely can’t be ripe for digital transformation. And yet, digital transformation is very much needed. Clients want to see more transparency and clarity over their cover. They want easier processes and fairer pricing. And, increasingly, there are new market entrants willing to provide it.

But insurers, brokers and agents do have opportunities to meet and even exceed customer expectations. At Intelligent Insurer’s Commercial Lines Innovation USA Virtual Event (May 18 to 20), more than 1,000 commercial insurers will gather to share their insights on how to meet these latest challenges and explore new strategies and solutions.

“We will learn how to interact with the machine, rather than the machine having to learn how to interact with us.”

One such solution is natural language understanding (NLU) and the opportunities it presents in the field of underwriting. Here, Pamela Negosanti, Head of Sales and Sector Strategy, Financial Services and Insurance (FSI) at expert.ai, outlines the difficulties commercial lines underwriters face and how automated solutions can help them deliver more accurate and appropriate solutions for clients.

Why has automation proved so difficult, yet is so needed, in commercial lines?
Unstructured data comes from documents that don’t follow a format. They’re not created in a template. Manuscripted policies are a perfect example: if you want to come to a conclusion about what they offer, you have to read them in their entirety. It’s not enough just to scan the table of contents.

If pandemic cover is excluded, you need to know. If it talks about a virus, you need to know if this is a medical or a computer virus.

Emails and tweets are unstructured documents, as are medical records. An invoice on the other hand is predefined: you can extract the total amount easily.

In re/insurance, the volume of unstructured documents is more than 80%, which is why there is such a need for NLU or natural language processing (NLP). You can’t apply pure automation or a keyword approach—it has to understand the cognitive process.

How can commercial lines insurers best manage the human/machine mix?
There is no one-size-fits-all in technology. Business knowledge is pretty important, it’s key when you want to come to some kind of conclusion. Claims, annual reports, risk—there are all sorts of elements in the policy process that are unstructured. You could extract the meaning from all of them but you have to go deep.

If you start automating that whole process without the context of human knowledge, you might get stuck. There is an actuarial element that needs to be applied to the start of the underwriting process because you’re at the beginning of the journey, and if you make a mistake it will ultimately correspond to a loss.

In international programmes with varying national and local policies, where customers may have a range of subsidiaries, the documents relating to a single policy can be completely different depending on location.

Why is this important? Because $40 million is lost every year to insurers because of small misalignments on policies. A single missing exclusion on one policy can mean millions in lost business.

How do you determine the success of implementing a degree of automation?
The main success metrics are risk reduction and loss avoidance, in addition to reducing unintended exposures. The commercial pressure is about being fast. That’s a metric related to winning more business. If you’re fast and deals are intermediated, you win the deal.

As a result, you will grow and business will flourish for your customer and your broker. Ultimately, you’re offering a better experience.

Will carriers ever be at home with the hybrid human/machine interaction?
We will learn how to interact with the machine, rather than the machine having to learn how to interact with us. We’ll get used to how we feed it information.

The perfect example is how we interact with our customers. We’re currently asking customers for a lot of information we could find out for ourselves and which would create a better experience if we simply asked them to validate the information we already hold on them.

Instead, we’re burdening them with lots of questions and that’s a poor experience for a customer at a time when they’re evaluating us as their potential supplier.

What do you hope viewers will take away from your session at Commercial Lines Innovation USA?
The main takeaway is this sense of urgency: that it’s important to use technology to be competitive. Insurers that use it will have a competitive advantage, so if you’re not yet thinking about how to internalise those technologies, make no bones about it—you will be left behind.

Pamela Negosanti will be speaking at 2:20pm EDT on Tuesday, May 18 in a session titled “Extract Value From Unstructured Data To Control Exposure, Augment Capacity, And Win More Business” at Intelligent Insurer’s Commercial Lines Innovation USA Virtual Event (May 18 to 20).

The event is free to attend for insurers and brokers/agents, but you must register in advance. Sign up to access the content live and on demand here.

expert.ai, Commercial lines, Innovation, Virtual event, Insurance, Reinsurance, Pamela Negosanti, North America

Intelligent Insurer