For underwriters, the term “informed decision” takes on new meaning when it comes to effective risk selection.
Some policyholders will go to many lengths to defraud their insurers… and these fraudulent actions are often tied to claims. But in the never-ending game of cat and mouse, insurers often learn that policyholders find other avenues for fraud. For example, what if, when signing a new customer, underwriting misses a critical element that would negatively impact onboarding? By the time the carrier connects the dots and identifies the error, the policy is already issued. If the carrier only sporadically runs additional loss reports, the error may not be revealed until renewal, and by then, the damage is done, reflected negatively in budget/loss ratio/etc.
What went wrong?
Such was the case with one insurer—a midsized P&C carrier that processed a new customer who had bundled their renters’ and auto policies for an advertised discount. The carrier’s underwriter conducted due diligence on all aspects of information gathering for both the auto and home, such as gathering all the necessary information from both third-party sources and the applicant himself, and, of course, running a C.L.U.E. report to reveal claims filed within the past 5-7 years. The C.L.U.E. report came back clean; unfortunately, the incoming customer falsely answered NO to the required claims question, in essence not revealing an extensive property loss claim that occurred just a week prior to application and therefore not included in the C.L.U.E. report.
Further discrepancies popped up after the fact, too. The applicant indicated that he lived alone, when, in fact, he had a roommate. He also said he had rented in his current location for a year; in fact, he had lived in his apartment for just one week.
In Search of Real-Time Insights
As a result, the insurer based its risk management assessment on false information and bound business with a policyholder that would ultimately be dropped as a bad risk.
Underwriters often shake their heads at the almost insurmountable challenge of structuring risk selection and fraud detection in such a way that allows them to optimize their portfolio. But what sounds like an unsolvable problem, isn’t.
What makes it solvable is being able to make data-driven decisions that are based on real-time insights into the risks being assessed. This means leveraging data from traditional internal and external sources, and applying an AI-powered, structured, uniform screening process that can be used to generate customer value models in order to score the applicant. By helping identify potential risks before accepting the risk, this gives “know your customer” a whole new meaning. And it does it in record time.
Fueling Informed Decisions
Let’s look at just a few of the elements that could be at work in support of this process:
- Data is at the heart of primary verification. Here internal and external sources are consulted to check whether the personal details of the applicant are correct, their income is correct, if there are other comparable insurance policies in play, if bank/credit card details are in order, etc.
- The use of text mining, in which large amounts of text are analyzed for behavior and event patterns such as sentiment analysis, automatic classification, entity recognition, detection of semantic relational patterns, and other methods that may reveal insights otherwise unknown to the underwriter.
- Anomaly detection. Here data from social network analysis helps identify items or events (often that don’t seem to be related to each other) can be analyzed to identify whether the information being provided by the applicant fits within an expected pattern.
- Financial risk analysis, which includes data from public sources (creditors, previous bankruptcies, previous losses, etc.), is used to determine the reliability of an insurance applicant.
- Predictive modeling, which can calculate the likelihood that an insurance applicant is reliable based on all of the insights gleaned from above.
In the case of the “bad risk” renter noted above, several of these elements could have raised a red flag, causing the underwriter to investigate further. And using this digitized, automated underwriting process, the underwriter in all likelihood would have used the insights to either reject outright or accept the applicant with conditions, saving his company time and revenue.
If the underwriter applied AI-enabled straight-through underwriting to a legitimate applicant, the onboarding process would be seamless, the customer experience would be a positive and long-lasting one, and the insurer’s portfolio would benefit as a result.
Informed underwriting decisions are powerful ones, making what would otherwise feel like an insurmountable challenge a manageable risk assessment process that guarantees a win-win for the insurer and the policyholder.
Is the idea of automated underwriting too good to be true? Check out our e-book, The Computer Says Yes.
