Deep learning-based document automation can finally streamline and expedite analysis of loss runs across the insurance industry.
Information is crucial to calculating risk, both for sellers and for buyers of insurance. Some facts are in the public domain: What is the likelihood of a power outage halting production in this region? What are the odds that workers in this sector suffer injury on the job? But for all the questions general industry data alone cannot answer, there are loss runs. And until now, assembling and comparing the data they contain has relied on tedious manual work that spelled stagnation for the industry.
What Are Loss Runs and Why Are They Hard to Automate?
Loss runs can be likened to credit ratings, which mine the financial history of borrowers to expose and quantify the repayment risk. Importantly, there exists a robust carrier-agnostic system for rating the credit health of individuals with a credit score. In similar fashion, the insurance sector can evaluate risk for personal auto and property policies using the Comprehensive Loss Underwriting Exchange (CLUE), but unfortunately no such central repository exists for commercial claims. Instead, data must be requested, culled, and submitted on a manual basis.
Individually issued loss runs aggregate the insurance claims history of a business to reveal relevant patterns of damage and liability that may stem from malice or negligence. This helps brokers and underwriters assess—and put a dollar value on—acceptable risk. There is, however, no streamlined system designed for this: not only does every business have a different history of claims and payouts, but loss run reports themselves come in all shapes and sizes. What’s more, most businesses have multiple insurance policies (the average is 2.3) covering auto, property, liability, and/or workers’ comp insurance, typically using several carriers. And with brokers regularly receiving loss run reports for the five-million-plus businesses requiring commercial coverage, there are millions of loss runs in circulation annually, requiring millions of hours of manual analysis.
Loss run reports have typically been issued by carriers in unstructured PDF or Excel format, using in-house templates. No industry standards exist, so comparing different loss run reports has entailed the kind of apples-to-oranges work that is famously resistant to automation, prone to errors, and impossible to do quickly for a human.
Why Is Automating Loss Runs a Priority for Applica?
Our mission is to free companies from having to burden people with tedious, repetitive work if it can be performed with more speed and accuracy by machines that harness deep learning to do the job with the utmost accuracy. The less time underwriters spend manually processing loss run documents, the more time they can spend making decisions better suited to human intelligence. Faster risk assessments lead to more policies and happier customers. Imagine increasing your throughput by 50% or more without having to add staff!
Our company’s unique ability to automate documents that feature both structured and unstructured text—and to train the software using a minimum of examples—is ideally suited to the challenges posed by loss run reports. And as we inject speed and accuracy into a process long overdue for an update, we’ll be helping a whole industry shed old, antiquated ways.
What Are Some of the Problems We’re Solving?
First of all, we’ve mentioned the apples-to-oranges problem. No consistent terminology or methodology exists on how to report policy type, lines of business, claim categories, payment types, and scope of payout, which makes it extremely complex to analyze loss runs in a normalized format across carriers. What’s more, some carriers provide loss descriptions but no separate cause of loss info, which means that brokers must investigate loss causality patterns by manually reviewing as many as several hundred claims in a single loss run report. And even when all the information is there, it’s often buried in blocks of text that start with a name and end with a date and tend to confound legacy software-based solutions. Applica’s mission is to harness deep learning for this type of needle-in-a-haystack busywork, which is why we bring ease and speed to the loss run review process.
Second, insurance carriers aren’t motivated to invest in simplifying the problem. Because loss runs are essential for setting policy terms and premiums, carriers must comply with the regulatory requirement of providing brokers and insured clients with loss runs on request. This they do, but the data resists easy comparison against competitors. However, if carriers were to simplify the process of procuring and scrutinizing loss runs, brokers would more easily obtain better quotes from competing carriers for their clients. Keeping the status quo means brokers and clients are more likely to stay with the carrier on record out of simplicity. Shopping around is just too much of a hassle. Applica is here to change all that, because template-agnostic transparency in business is one of the values we champion.
Third, a staggering 80% of insurance information remains dark. The majority of the industry’s unstructured data is buried in the policies, loss runs, quotes, and emails stuck in carriers’ paper and digital repositories. Advances in cloud computing and deep learning show significant promise that this information can finally be put to work to create value for carriers and reduce risk all around. With Applica automating loss run reports, information capture can be instant, total, and far-reaching. After all, deep learning doesn’t just answer the questions a broker asks today—it extracts every piece of data available, to be ready to answer questions you might ask in the future. And that seems like where everyone’s headed, including the insurance industry.
To find out more about the ways Applica can solve the complex problem of loss runs, connect with an Applica expert today.