Last week, I had the pleasure of co-hosting Novarica’s virtual Special Interest Group on the topic of analytics with Novarica SVP Mitch Wein. Novarica previously conducted these sessions once or twice a year before the pandemic, but can now deliver them much more frequently in a video conference format. In a time of significant change and disruption, the ability to deliver these sessions on an accelerated basis can allow for much more dynamic dialogue than we would have thought possible in the recent past.
The importance of having robust data and analytics capabilities has increased drastically during the pandemic. These circumstances require insurers to move quickly and operate more efficiently as they prepare to ride out the current recession and service their policyholders remotely.
Mitch began the conversation by speaking to the effects the pandemic has had on data and analytics within insurance: COVID-19 may have lessened the efficacy of certain predictive models, and they may now be counterproductive. Models decay over time because they are based on historical data. However, COVID-19 may have accelerated this rate of decay for certain models. Insurers should conduct an analytics inventory to combat possible model failures and assess whether they are still adding value with the insights they generate.
I then added that customer-centricity, churn, and “next best action models” are part of a group of analytics that may not be as effective as they once were. Additionally, models that rely on specific data points, e.g., credit score, may no longer make sense considering the impacts that COVID-19 has had, e.g., rising unemployment. The country has moved into a completely different economic situation; it has affected claims filing (especially in auto), workers’ comp, and GL. A participant chimed in to say that insurers may not fully understand the impact the pandemic has had or will have, but that is not something that should deter them. She added that her company is cataloging and monitoring potential COVID impacts and has plans to reassess these regularly.
The group went on to discuss the steps insurers can take when evaluating third-party providers. Determining the quality and suitability of external data requires labor-intensive and time-consuming testing. Solution providers should go through a vetting process that rules out vendors up-front based on easy-to-assess criteria and establishes success criteria at the outset. These criteria can include knockout items that are simply no-go or what use and business cases are behind the data initiative. Critical elements for insurer success are moving with purpose, understanding what the organization needs from the data, and finding a partner that aligns with the organization’s needs and culture.
The discussion then turned to talent and how organizations could find and retain strong data people. The importance of training and recruiting talent is something most insurers consider paramount. However, coming up with creative ways to do so can be a challenge. The conversation touched on WFH policies as a potential draw for younger workers. However, someone made the point that people desire a balance between working remotely and in the office; satellite offices still may have a place in the future.
Participants went on to talk about use cases and the importance of backing up analytics initiatives with business objectives. Predictive analytics exist across the insurance life cycle. Many models, however, focus on underwriting and claims, i.e., predictive scoring, fraud. Model management has recently become a particularly critical operation for many insurers. Much like with DevOps, insurers will need to be able to manage models and automate their deployment as they continue to grow their use when doing business.
The next virtual Special Interest Group with a focus on analytics will take place on Thursday, September 3, 2020, at 1 PM ET. Registration for that event can be found here.
Add new comment