A Data Forward Approach to Winning in the Marketplace

The data and analytics arms race is on. Carriers are working to gain insights about existing and potential customers to create winning strategies. The insurance industry is no exception to this trend. Insurance always has been a data driven risk business, even before the term Data Science was coined. What has changed is the explosion of non-traditional data sets from third-party sources (e.g. IoT, Social Media). Coupled with cheap computing power and predictive algorithms, this data driven approach is truly ushering in an era of the insight-driven digital enterprise. Insurance sectors are at different stages of maturity in this pursuit. P&C carriers are investing heavily in this area, while the pace of change in Life and Annuity (L&A) and Group Benefits sectors have been more measured. This paper examines machine learning use cases that can propel L&A and Group Benefit carriers forward, to reflect on some of the limitations of these third-party data and analytics fueled advances.

Product Management

A chronic challenge for L&A industry has been low insurance ownership rates with stagnant premiums. The perception of insurance as expensive and something which only older or sick people need has been hard to overcome. With additional data which allows for more granular risk pricing and rewarding a client’s healthy life style, L&A carriers have an opportunity to increase sales through effective product management. Historical and current attending physician statements, electronic health records and prescriptions records provide an excellent source of data to build machine learning models to price risk at appropriate levels. The advances in treatment for certain types of tumors, cancers have changed the risk profile of certain pools. Leveraging research published by leading medical institutions and re-insurance companies can open these otherwise non-addressable risk pools to products like CI. Building products with the risk-reward component based on wellness data captured via wearable devices is a widely accepted approach. Carriers can model data from wearables and historically favorable claim experience from these heathy populations.

In contrast to individual risk assessment, Group Benefits has long relied on aggregate SIC data, experience rating, and census level data for underwriting. The opportunities for a data analytics driven product approach in Group business include: Evaluate a group’s corporate behavior in health, safety management and employee well-being programs from public sources such as corporate recruitment material, litigation or access to workers comp data (if available). Identify geographic risk to group’s workforce based on vehicle accident data, crime data, water quality, etc. Offer unique employee coverage to align with the emerging social trends such as accident coverage in rideshare, supplementary fertility, etc.

Sales and Distribution

Two trends in insurance distribution are:

  • Micro-segmentation to reach the right potential customer at the right time with the right products.
  • Migration of simple products to direct to consumer model.

Today, data and analytics is making these once directional notions real:

  • Producer and compensation management for L&A and Group carriers is a norm. Additional analytics based on specific products, market campaign effectiveness by producer, depth and breadth of demographic access, and affinity relationships in underserved communities can pinpoint key relationships that carriers should strengthen, develop or weed out.
  • The customer ease with buying online has extended the online distribution beyond term life and identity theft products to include what would be traditionally “benefit products” like Dental, CI, and Vision. Analytics for campaigns, predicting the next best action and web analytics to follow up on drop outs are key to success of this direct to consumer channel.
  • Mining of social data for analytics to reach individuals at the moment that is ripe for an emotional and need-driven buy such as chronic illness, major accident, child birth, marriage, or divorce in immediate friend and family circles. This is perhaps the most controversial of all third-party data sources and carriers should continue to experiment within the bounds of existing and emerging legislation like GDPR (General Data Protection Regulation EU).

New Business and Enrollment

Adverse risk selection and churn have always been key concerns for carriers in the retail life and annuities sector, and with just in time coverage, it becomes an even greater consideration to have insight into potential customers. In the Group context, carriers are focused on enrollment take rate to make the case profitable.

  • ML models based on demographic, social, motor vehicle report, attending physician statements and credit data are increasingly becoming sophisticated to predict risk scores and life time value of customers.
  • ML models that identify characteristics of eligible customers and predict their likelihood to enroll in voluntary products are of great interest. Carriers are mining their historical enrollment data to draw insights.
  • Pre-filling of application forms with data from third party sources to be verified by applicant is almost a standard operating procedure with carriers.

Claims and Servicing

Risk management, the DNA of insurance companies, is evident nowhere more than claims related processes in the enterprise. The traditional use case of data analytics for fraud is becoming sophisticated and mature. Emerging technologies such voice-based analysis is driving the next generation of service models while data analytics proliferate all aspects of claims payment, especially ‘managed case’ kind of products which have a long tail like Disability and Workers Comp.

  • Fraud model extensions to leverage third party social media, credit data, in addition to a carrier’s own data in a continuous machine learning paradigm.
  • Service model to validate the insured based on voice recognition and voice stress analysis for fraud detection.
  • ML models for predicting STD to LTD bridging, Return to Work, and identifying claims that are candidates for offset and settlement.

A Cautionary Note

Extensive data and analytics driven approach is not a panacea. In addition to being subject to legislative scrutiny; the robustness of some of the alternative underwriting and service models based on emerging technologies is yet to be proven out. The reconciliation between data driven models which highlight correlation, but at times do not align with the causality as understood by business experience is still unfolding.

The Life and Annuities sector has evolved since the basic tobacco/non-tobacco classification of the 1980s to include to more lifestyle-based underwriting for preferred and super-preferred risk. The additional data sources available today extend that risk selection process, in combination with medical underwriting to prevent anti-selection. Some third-party data elements like credit scores, which have become standard in rating, are being scrutinized by states and their proxy as a person’s risk is being questioned due to credit score fluctuations. States like California, Hawaii and Massachusetts have already banned credit scores for rating. The passage of GDPR in European Union in 2018 is likely to push the envelope on personal data privacy further in the US, starting with states like California.

The recent SelfieQuote launched by Legal and General insurance which utilizes Lapetus’ facial analytics platform to determine, age, gender, and BMI for term life policies is yet to be proven at scale across the spectrum of carriers.

In Conclusion

L&A and Group Benefits sectors have long been a distribution-driven business. Changing end customer behavior and expectations are now bringing significant changes to every part of the business: product development, distribution and service. Carrier are now driving a digital transformation where data and analytics have gone from being an architectural consideration to the lynchpin of a differentiated business strategy. Targeted execution of specific machine learning use cases are moving the needle on business outcomes. An enterprise data platform is no longer a pre-cursor to these solutions. The time to act is now.