New Data For Insurtech Has Lots Of Promise But Will Require Much Iteration

New Data For Insurtech Has Lots Of Promise But Will Require Much Iteration
Life insurance is a unique product — even when compared to other consumer financial services. For starters, a life insurance product, at its core, is all about managing mortality risk, which has a wide range of contributing factors and can take decades to emerge.

Issuing a policy requires assessing complicated data about an individual, including detailed records such as prescription history, prior lab tests, motor vehicle data and sometimes an invasive medical examination that includes a blood and urine sample. This is a financial product that requires a deep level of personalized information, and as a result, you can’t just buy it off a shelf or expect it to be delivered with two-day shipping.

After years of investment into data collection, a lot of this information can be obtained electronically and in nearly real time, transforming the customer sales experience. However, this space continues to evolve, and getting fast access to more comprehensive medical records remains difficult. Transforming this data into actionable insights is another challenge. Many carriers are looking to leverage algorithmic and automated methods for analysis and review, ideally providing a decision back to the applicant within seconds. Those techniques can be challenging to interpret or explain.

As you can imagine, this is all much easier said than done. It took years to get to this stage of maturity, and some carriers still lack the digital infrastructure to take advantage of new data and processes. Especially in light of the transformations rushed in during the Covid-19 pandemic, the insurance technology (insurtech) industry is facing a big inflection point.

Electronic medical data is a double-edged sword

Electronic medical data provides a wealth of information for insurers looking to evaluate risk, including diagnoses, lab results, prescription history and physician notes, as well as important considerations like weight, blood pressure and tobacco use. However, due to a lack of standardization within the medical industry, it can be extremely challenging to consolidate this data to evaluate it systematically. Providers use different coding systems for diagnoses and procedures, and some doctors prefer unstructured notes altogether.

Furthermore, accessing the data remains challenging. Most medical data exchanges operate regionally, though some vendors are working to provide national coverage. And, while applicants authorize access to these types of data sources when applying for an insurance policy, not all medical providers accept a standardized authorization, which can lead to additional paperwork and delays. 

Despite these challenges, the promise of standardizing and incorporating this data into existing streams remains strong. Consider today that the most equivalent comparison is requesting a full attending physician statement, which usually requires weeks of processing and manual review, slowing eligibility decisions and increasing costs.

Iterate, and then iterate some more

Many insurance companies have undertaken new data implementation projects by starting with historical retrospective studies that can be time-consuming and expensive. One data acquisition project I worked on spanned three calendar years just to acquire a dataset to start designing a program! Suffice to say: These types of data projects require an incredible amount of patience.

More and more, and accelerated by the pandemic, I’m finding that carriers are hoping instead to learn as they go by piloting data in small ways and building up programs around it. This can require a heavy investment of human review at the beginning, and it’s not always possible to learn as you go if the ingestion of new data influences outcomes you are trying to measure against. Ultimately, the decision to perform retrospective analysis or to try to learn as you go is a careful balance.  

All of this iteration, painful as it may sound, is to be expected. The property and casualty insurance space has been through its own data-driven evolution over the last several decades. In the auto insurance industry, telematics has emerged as a sophisticated, data-driven approach to accident risk. The earliest insurance companies to use the technology started collecting very simple data points like speed from devices installed in cars. Over time, data collection expanded to include acceleration and braking, GPS positioning and road condition data. Analytics techniques evolved alongside the data, and today, telematics is able to provide a powerful and customized view of driver risk. 

New data provides new opportunities, even though it may take time to learn and adapt. Electronic medical data has the potential to dramatically change the purchasing experience for both the healthiest consumers, who can expect more real-time offers and improved pricing, and those with more complicated medical histories, who have historically had to wait for months for more comprehensive reviews. With continual learning and progress, this data can be truly transformative. Small datasets can be incorporated and tested in simpler reviews to make the easiest applications even easier and then continue to build up from there, slowly allowing life insurance pricing — and the algorithms — to get even better for consumers. Slowly but surely, we will get there.

Via Forbes

Share this article: