One of the great ironies of the insurance industry is that it was among the first to embrace hardcore analytics and yet is today known for its reluctance to adopt technological innovations, thanks in part to regulations that few tech innovators have been willing to tackle.
But that's all changing.
The startups transforming the insurance space right now are demonstrating that data science is useful far beyond its traditional industry application of enhancing risk analysis and predictive modeling. One of their most exciting targets is commercial underwriting.
Toward a more accurate risk analysis
As commercial underwriters and producers know all too well, analyzing risk and writing coverage for commercial clients is complicated. Even assuming a client understands every question on an application perfectly and answers 100 percent truthfully, it's important to remember that you can't get answers to questions you don’t ask — and you can’t ask every question.
Consider, too, that most commercial transactions are still carried out through traditional brokers, and you see where there's room for improvement. After all, brokers are made up of people, and people have biases. In addition to verifying an applicant's answers and tracking down missing data points, underwriters have to draw from their experience of risk factors in an industry. That means the premium one underwriter quotes may be different from the premium another underwriter quotes for the same business.
This isn’t to fault underwriters — they do difficult, high-stakes work better than it’s ever been done in the past.
But no one person can hold in their head perfect knowledge of a business and its operations, plus perfect knowledge of the risk exposure for a given industry in a given geography. And that’s what it would take to have truly accurate underwriting.
Enter data science.
Productivity, efficiency, accuracy
Data science is transforming commercial underwriting by replacing human-fueled parts of the underwriting process with more objective data and algorithms. This transformation manifests itself most obviously in three ways:
- Eliminating paperwork: The frustrating reality of traditional underwriting work is that a lot of it is data administration — moving data from one form or source to another. When underwriters have access to data-science-fueled underwriting workstations, they see 20 to 40 percent gains in productivity. The products and services in the works today will turbo-charge underwriting by automating information capture from forms and reports, providing access to data from thousands of sources and compiling it "intelligently," with help from machine learning.
- Enabling smarter risk analysis: Data science is also capable of "intelligently" compiling data from thousands of sources for underwriting analysis. The result of all that data being compiled by smart algorithms is better risk analysis. Increased data plus decreased human bias equals risk analysis that’s truly based on facts. Companies using workstations report initial loss ratio improvements of two percent on average.
- Increasing processing efficiency: Naturally, spending less time per account means underwriters can service more accounts. Insurance carriers are seeing a 50 percent decrease in submission turnaround times and a 40 percent increase in policy counts.
Insurance industry research group Strategy Meets Action (SMA) has found that underwriting is one of the most targeted areas for investment. That isn’t surprising, given that we’ve seen insurance companies that are unable to process as many as half of their inbound leads because of bandwidth.
Adopting tools and practices that leverage data science to transform the speed, efficiency, and accuracy of the underwriting process is one of the most direct ways underwriters can realize ROI. Perhaps just as important, these improved processes will also enable greater transparency in policy pricing, improving the user experience and removing some of the pressure to compete on price.
In fact, Said Taiym, CIO of AF Group, believes that improved customer experience will be one of the most noticeable effects of data science applications in commercial underwriting.
"Customers today expect the Amazon experience: they want what they want quicker. And that model — entering a few pieces of information and getting a quote — just doesn’t exist today in commercial insurance."
But, Taiym notes, it should. "All that data is out there," he says. "With data science applied appropriately, commercial underwriters will be able to take minimal information and come back to customers — quickly — with clear options of what’s out there."
Better profit margins and happier customers: these are the holy grail of commercial insurance. And they're exactly what data science applications have in store.