AI-Driven Underwriting by David Kuhn
Insurance: The New History
History, as we all know, is written by the victor. Winning means you get to tell your side of the story and have it live on through time as the facts as they accurately happened. How the simple passage of time elevates one person’s perspective to the absolute truth I’ll never know, but that’s how it is. In order to be history, you have to make history.
I’m going to venture into the world of insurance and tackle history with a more agile approach. In this new blog series, “Insurance: The New History” we’ll take a dive into the things happening in insurance and around insurance now, or things that I think will be a major player in the industry very soon. This is my perspective on the what, the why, and the how of history-making tech in insurance.
These articles should serve as an innovation jumpstarter or, better yet, get companies to think “let’s try something different” instead of “this is the way we’ve always done it; this is the way we’ll keep doing it.”
I’m hoping this helps people in insurance be the victor, the historian of record for their own stories.
Be the disruptors, not the disrupted.
Underwriting vs Underwriter
Underwriting is an essential part of how the financial world functions. It is the process any organization or individual goes through to assume financial risk for a fee. Underwriters are the individuals who evaluate and calculate that risk. The concept seems simple, “underwriters underwrite,” but the process is more complex than you’d think and even within the insurance industry often not very well understood.
In insurance, underwriters asses risk for their company. This helps determine premiums the policyholder should pay based on the risk. They usually has a checklist of underwriting rules that have been created to help the underwriter determine if the risk in question meets the risk appetite of the company. In most cases, the underwriter is given the authority to act on behalf of the underwriting company (traditionally a carrier) to assess the risk, apply a premium, and accept the risk. There is a component of this that I want to really make clear…they are given authority to act on behalf of the company.
You might be asking yourself, “why is this guy writing an insurance blog intended for the insurance industry telling us about simple insurance concepts?” This is a great question! First and foremost, it’s to provide some crucial backstory to what I want to talk about in this blog.
Secondly, insurance companies, as a whole, don’t understand this process well. The underwriting department knows this stuff in and out, but your IT department, the claims group, even agency teams or people that deal with distributors, would likely look at the traditional role of underwriter versus the process of underwriting and ask, “what’s the difference?”.
When you step back and look at the entire underwriting process instead of the just underwriter role, you should clearly see that there are many steps and people involved in the process of assessing risk. So, even if they don’t know the exact answers to those types of questions, they should at least understand why the questions are important.
Assessing risk: the never-ending to-do list
If you’ve looked at any insurance company’s financials, there is one place that has consistently cost a lot of money – the underwriting process. Because an underwriter is accepting risk on behalf of the company, they need to be sure that the risk they are looking at will be good business for the carrier. “Good business” is in the actuarial tables and assigned the proper premium for the risk profile that the customer represents.
The biggest time-waster in the pricey process of underwriting is the hours an underwriter puts in researching data to analyze the risk. In areas such as Personal Auto, this has been heavily automated using third-party data strategies because there are minimal complexities in a vehicle and driver. An underwriter can quickly determine if that is good business.
However, in commercial insurance, “minimal complexities” barely get you started.
Let’s look at a hypothetical example… Let’s imagine a manufacturing job shop, meaning that they don’t really do production runs with assembly lines and that each project they are doing is relatively unique. Not only do they have vehicles and a building that needs to be insured, they might have a fleet of vehicles and run multiple shops in several locations. This company might be buying new tools and technology to improve their value offering to their customer. Then there’s the employees – the people who run the equipment, program the tech, and drive the vehicles.
The risk is this example is practically infinite. The buildings, the equipment, the vehicles, the personnel – it all needs to be evaluated. An underwriter needs to not only look for how close the buildings are to a fire hydrant, but also at what fire suppression system or systems the company has in each of those structures. They need to understand how the employees work and look at payroll and previous OSHA violations to know how to deal with workers’ compensation policies. A lot of underwriters may also look at the board of directors and executives of the company. What if this isn’t a company that the carrier might normally accept, but the independent agent or broker that gave them this customer has an incredible track record when it comes to selecting good risk?
That’s just a cross-section of a nearly endless list of points to assess. Imagine how much time that all takes.
How do they research that and assess if it is good risk? How do they ensure that the company knows the risk appetite has changed? Underwriters have junior underwriters and as well as assistants – and all of them are kept overwhelmingly busy researching and assessing.
Over and over again.
Did I just say underwriting workbench (U/W Workbench) and not underwriter workbench? Yep, it’s not just about the underwriter. If you don’t understand why… read the backstory at the top.
A workbench is where you do a specific job. In the world of woodworking, you might have a cabinetmaking workbench or joinery workbench. In metalworking you might have a welding or forging workbench. Notice that I didn’t say for metalworking that I need a blacksmith’s workbench? So why do people refer to a U/W Workbench as an underwriter portal? This term has created a lot of confusion over the years as to what it should be or not be. For instance, I view an underwriter portal as a place that people with the underwriter job title can go to do all of their work as an employee. This might have a link out to the U/W Workbench, but probably also access to other HR systems, Payroll, internal job postings, etc. Do you see the difference?
Novarica has good definitions of the components of an underwriter workbench is. These components are:
- Application Submission
- Renewal Processing
- Third-Party Data Integration
- Quote Management
- Underwriting Rules
Novarica also defines an underwriter workbench as being “used by underwriters on a day-to-day basis to manage and evaluate information from the field, the enterprise, and increasingly from third-party sources. Also increasingly incorporate predictive models and scoring to help underwriters make the tough calls.”
Most people that I talk to in the industry about a workbench will only look at one or two components from this list and determine that is all they need. The most common request is “I need a place for our underwriters to view their tasks.” There is an outdated belief in the industry that underwriting is relatively simple and can be automated just by putting the underwriter’s task list in an app. Underwriters need all the attributes from Novarica’s list at their fingertips, but so does anyone who deals with the underwriting process. Underwriters really aren’t the only people who are involved in the process of assessing risk.
Laboring under the misapprehension that underwriting is the exclusive purview of underwriters is narrowminded. It’s also a big old roadblock to doing business better.
There is also another concept known as front-line underwriting. This enables your agents to check the risk profile of a customer before sending over the business. Some agents are great at this because they understand the owners and types of business, some aren’t as great. We all know that a distribution and sales team that don’t create a great experience will definitely not drive sales. These agents need to have their process simplified. Let’s create more time for them on the golf course, the shooting range, and social events to focus on relationships.
You must reimagine the U/W Workbench
Events, AI, automated research gathering, underwriter rules curated by underwriters, assessing good business and bad on claims…having all of these features in a single solution sounds incredible, right? But, you’re thinking, an underwriter workbench can’t do all those things.
You’re right – a glorified task list isn’t capable of more. But an U/W Workbench is. Dig into your brain, dig out all the preconceived notions you’re carrying around and slam dunk them into a mental trash can. To imagine what’s possible you have got to stop assuming something you haven’t tried yet isn’t going to work.
When people at insurance companies reimagine a process, they look at the existing process and try to identify what’s working and what isn’t. They’ll examine pain points. That’s all well and good.
But then they evaluate the systems that process runs on. And apply limitations to whatever new process they’re trying to come up with based on the current system, often with little to no input from IT. Right at the beginning of imagining a new a better way to do things, they limit themselves to thinking smaller.
You’ve heard all the excuses. “It’s our culture.” “It’s always been this way.” “We asked for that before and IT said they couldn’t do it.
Stop. You have to stop. While you’re stuck in a loop of old tech and bad ideas you’re competitors are thinking big and making major changes – both to the culture and their business.
You don’t need an underwriter workbench. You need to open up the process.
And you do that by asking a very simple question. Remember at the beginning of this blog? I defined the process of underwriting with some keywords “assume financial risk for a fee.” The risk has to be researched and assessed, sure, but the part that really matters is the company must promise to take on the financial risk by accepting a premium. The question that ALL financial institutes should be asking when dealing with the process of underwriting is:
“What are the steps it takes to assess and accept the risk at the right fee?”
Don’t bog yourself down with questions about what systems you need or what steps you need to take in that system. First, just look at the process.
That process should start at the interaction with a customer.
Now… we’re going golfing
Okay. So, if we’re going out with the old. It’s time to come in with the new.
How do you open the underwriting process? Starting with a page-one rewrite isn’t always easy, but when you’re trying to turn the focus back onto the customer, journey mapping can be helpful.
So we’re going to do that.
Amy is a commercial insurance agent. She’s out getting drinks after a rousing golf tournament. She ends up chatting with Sarah, a CEO of a large regional pizza company. What starts out as a discussion on good coaches for chronic backswing problems turns into a conversation about a recent outage that the company had due to a software upgrade on their digital pizza ovens.
How does the underwriting carrier win that business? What are the steps that must be taken for the agent to first determine if their carrier of choice has any issues around similar companies? Most importantly, the agent needs to know how to determine if Sarah would be good business.
By the way, this all has to be done in real-time. Once that agent has determined Sarah’s company is good business and sends it off to the underwriter, how can the underwriter have access to the same information the agent used to not force the agent or customer to give duplicate data? There is a whole lot more to this scenario, but you can see that underwriting needs to be done quickly and with as little information as possible.
Underwriting is a crucial part of the insurance industry. But improving that process ensures that people can focus on building a trusting relationship with the customer.
So with this scenario, what if Amy the agent had an app on her phone where she could enter the CEO’s name, relevant organizational details, and whatever additional information they could gather during that initial conversation? What if she could quickly get back how likely Sarah’s pizza company would be to qualify for a policy and maybe even create a range of what that type of business would typically pay for a policy?
Think about it. Amy could give Sarah all this relevant information in a follow-up email the next day before she’s even asked for a quote. How easy would that make it for Sarah, now as a prospective customer, to ask for more details on a possible policy?
Did you notice that there were only two names in this, Amy the agent and Sarah the business owner?
I didn’t mention the underwriter, George, who behind the scenes has already created a risk profile for what a good restaurant might look like and worked with his team to create all of the underwriting rules that need to be asked when assessing risk for that type of company. George and his team are vital to the insurance carrier, but Sarah doesn’t care much about either underwriting or underwriters at that point. She cares about speed, competence, and ease, she cares about the experience.
Amy and George might interact with Sarah throughout the rest of the process, but shifting an empathetic conversation at a bar to an information gathering discussion on a prospect could have very likely killed the deal before it even had a chance to start.
That app Amy entered Sarah’s information into is the underwriting workbench. It’s the same app George and his team accessed and used to access the risk of Sarah’s business. The relevant underwriting rules were there, the carrier’s history, decisions on similar companies, Amy’s notes on the prospect’s problem, and any helpful third-party data – it was all there. Amy used it. George used it. Kareem from Legal will check it for details when the final policy is drafted. Natasha from IT makes sure it always works.
A modern, comprehensive Underwriting Workbench changes the game.
How AI can help?
Artificial Intelligence is probably the most overused and misunderstood word in the dictionary. I won’t go into the details of why, but by generalizing everything into one bucket of “AI” people can’t understand what it can and can’t do. For the most part, and this is changing, AI doesn’t do emotional intelligence. It doesn’t have empathy…and if it seems like it does, it’s just following a prescriptive method for psychologists. In my previous blog about AI in insurance, I talk a lot about focusing on having humans do the value-add tasks. These are anything where empathy is essential and just having a human to talk to will create a level of trust that a computer can’t do.
After envisioning the new underwriting process, take a look and ask yourself and the team, what should a person do? Determine where the highest points for potential emotional engagement are going to happen.
When a policy is denied, a person probably wants to hear that from another person. But do you want your team assessing what rules triggered the most claims and how should you tweak those rules? A person does not need to do that manually. This is where AI comes in!
There are all kinds of ways to utilize machine learning to look at decisions that were made and tie them to claims. The question will then become, do you want the computers to decide which underwriting rules to create based on the data they find? What if that seems too far out there? Maybe you start with creating an underwriter rules component, but also create an application that can use AI to determine what impact the rules would do if they were in place before your current book of business. Other areas are around risk assessment. Maybe you only want your underwriters to look at the highest risk applications.
Utilize AI to enable the capturing of simple data fields, then utilize third-party data and internet sources to determine what level of risk the application fits into. If it’s 1-5, maybe automate that straight through, 6-8 have a junior underwriter review it, 9 and 10 it goes into a pile for the most senior underwriter to look at before the application is denied. Remember, there is no bad risk, just pricing that doesn’t fit the risk profiles.
How do you determine if it is a risk of 1-10? Use third-party data to get where the nearest fire hydrants are, look at the lawsuits against the company, see legal actions against the people running the company, use sentiment analysis to read reviews from customers, etc.
Another great place to use AI in underwriting is to determine the risk appetite. Most insurance companies don’t have this written down, it is in the heads of the underwriters and a lot of times misinterpreted or just wrong and outdated. Why not utilize AI to look at all of the policies and risks that are in place now and maybe over the past 10-15 years. By using machine learning algorithms, you can determine what risks are good and which ones you might have only done a few times but could be worthwhile to do more of in the future.
To build or buy – not a mutually exclusive decision
“I’m hearing that you’re telling me I need a completely custom solution?” The short answer is YES, but the long answer is much more complicated. When I look at a “solution,” I’m looking for a series of applications and processes that mature a business capability. The underwriting solution will be unique for your company, and it should be.
How you underwrite policies and accept risk is one thing that makes your company different. It is one of the major components that determine how successful you are as a company. So, don’t take this decision lightly. But I also don’t believe that everything in the full solution is unique to your company or even the industry. For instance, task management, case management, etc. – those are all common components in a U/W Workbench, but they also occur across most industries. Unless they are very simplistic, I would suggest buying a component that manages that very well. Underwriting rules is another area. Your rules are what your company unique, but the structure of a rule and applying that rule via web services is not. Maybe look at solutions that can be the source of record for rules and the underwriting rules are applied during the process of underwriting. AI is one thing that you shouldn’t buy. There are models and algorithms out there that will help make you very successful, but tools that offer “underwriting AI” are more than likely set up in a way that needs to be customized to fit your specific needs.
However, you decide to create a solution that works for your organization, your first step is to rethink the whole process of underwriting. Don’t break a new process that hasn’t even been built yet by tethering it to legacy systems or outdated assumptions. Make this process accessible to anyone who looks at the risk of a customer or prospect. And more importantly, use AI and automation to ensure that your writing process is modern and efficient. By doing these three things, you will find that your expenses will drop dramatically, claims will be settled easier, agents will love doing business with you, and the customers will be happier.
That’s a world of potential wins right there, but they come at the cost of significant change. Be bold enough to be different.