Premium audits apply AI to lower underwriting risks

January 6, 2025 4:11 PM

An analyst uses a computer and dashboard for data business analysis

Premium audits are a key tool that commercial insurers use to check that the premiums they charge for liability, casualty and workers compensation coverage are in link with the actual risk. Davies, a specialist professional services and technology firm with a presence in the U.K., Ireland and the U.S., provides these insurers and other regulated markets with risk management and technology solutions, including premium audits. Digital Insurance spoke with Mark de Waal, head of premium audit at Davies, about commercial carriers' current needs, how AI is being used in the field, and the impact AI is having.

This article is from a longer interview and edited for clarity.

How does Davies manage premium audits?

Mark de Waal of Davies — Mark de Waal, head of premium audit at Davies.

Think of us as an information conduit. We're purely an independent group. We don't have any skin in the game in terms of the outcome. It truly is a way for the insurance industry to provide a premium that's based on actual results of a business. It keeps everything very fair for the business owner. If the economy is doing well and their business is doing well within that economy, and they underestimated what the exposures were going to be at the end of the term, there will be an adjustment. If they overestimated, the insurance company will then have overcharged, not intentionally, but just because that was what was known at the time.

There's four players in this transaction. There's the business that buys the insurance. There's the agent which sells the business the insurance. There's the insurance company, which is the carrier that did the underwriting of that policy, and then we are an independent premium audit group within Davies that will level set. We work between, typically, the business owner themselves, and the insurance company, even though the business owner's relationship is typically with the agent and not with the carrier specifically.

What do the insurers need in audits?

Our clients, the insurance companies, require several different things. One is accuracy. The second thing is timeliness, because there is a time component to this. The third thing is not just accuracy in developing the factors that they use to calculate the premiums, but also to make sure that they have a complete understanding of the operations of the business and whether or not the business was underwritten properly with the correct classifications.

Small businesses are always shrinking or growing. Very rarely are they steady. For example, a construction company that typically only builds decks and the next thing you know, they're building something bigger than that, and they need subcontractors where they didn't use subcontractors in the past. Subcontractors can create additional risk and exposure for an insurance company. That's something that a Davies premium audit will look at carefully. Is a business using additional labor that could create an exposure for the carrier that they were not able to collect a premium on?

How does AI come into all this?

In the 1990s, 1.0 was moving handwritten documents to computers. 2.0 was moving all the processes from computers to the internet. With AI, 3.0 is where we're at right now. There are a lot of processes that happen in premium audit, even at the auditor level. While 2.0 has made dramatic changes in the amount of people required to manage this information, for much of what the auditor does, there's still a lot of manual entry.

That's where the opportunity comes in for generative AI to basically take processes that are inefficient and make them more efficient. One of the biggest hold ups and problems with audit is that oftentimes you could be mired with just an incredible amount of data to be able to sift through, to distill and to report on, and so the auditor spends a lot of time in the trees. What they're not doing is looking at the forest. One of the most important things for premium audits to do is to look at the forest, understand the business and the operations of the policy holder, and truly understand them, to make sure that underwriting got it right, that the classifications were right.

Because, if they are wrong, it has a ripple effect through this industry. Specifically for workers compensation, rates are based on injuries within class codes and things like that. The more likely you are to suffer an injury as an employee of a company, rates will reflect that. If you have an employee classified in a position that is not dangerous, when they should be classified as being in a dangerous position, that creates a lot of problems.

AI gives us an opportunity, not to replace humans, but to make human beings more efficient and to allow them to do a better quality job, maybe a faster job, without sacrificing quality. That's why we're very excited about what I call 3.0.

Are there differences in premium audits depending on what is being covered – workers comp, liability or casualty?

Premium audit typically deals with the financial side. We deal not just with payroll and people and classifications of workers compensation, but also general liability, garage liability and other factors in the business. With anything that is a process related task, where the human in the loop is doing something that is purely a process, quality control is an aspect of premium audits and any insurance reporting.

AI has an opportunity to get involved with that, especially generative AI. Generative AI can learn from past errors, and help the human with an initial review of the quality of the work, and point out things that the human should pay attention to.

Have you identified other parts in the processes that can also be improved?

A lot is not replaceable with AI. For example, discussing with a business owner how their business operates and their future plans. Things along those lines are not really a great conversation to have with the computer.

We're focused purely on items where the human doesn't really actually add any value, where the process could actually detract from what the human can do. By focusing on those things, we're not looking at job elimination. We're looking at efficiency gains from any particular employee that does this kind of work. Not only will it help them produce more and better work, it'll create a better level of job satisfaction for the employee, which is, to me, very important. Anytime we can keep people using their more creative side and less of the process side, we're all better off as a society.

What impact is AI having on premium audits?

We can improve our services to our clients and the policyholders. We can improve the jobs of our auditing staff and everybody in the organization. We can become more efficient in the work that we do, and allow our teams to focus on activities that AI is not capable of. We can create a significantly improved product and improved services that are more accurate and driven toward what underwriting really, really needs.

Underwriting is all about accurate information, and information is purely the lifeblood of insurance. If we're gathering information that they need, but not creating groundbreaking new insights for insurance companies, we could do our job better. We could actually add more value by a lot.

What’s the future for AI in premium audits?

A lot of people are scared by AI. If it's used properly, the possibilities are so exciting. Some things I've been looking at for all these years, and never saw a good technological solution. With AI, now I have seen it, and how it can work if it's done responsibly. It opens up an incredible opportunity for people at Davies, for our clients and for everybody. All I see is the good side of it.