October 26, 2021

INTRODUCTION

The explosion of data, recent advances in new technology, and increasingly competitive market conditions are driving interest by insurance companies in the application of data science. Insurance carriers understand the value of data and the insights it provides to underwriting. In addition to seeking better methods of harnessing their internal data, insurers are seeking unique new sources of third-party data to help gain a competitive advantage. The application of data science to the many parts of the insurance value chain has begun to prove worthwhile for some, but the full enterprise potential has yet to be realized as many projects are done in silos or as one-offs. Many data science projects are limited or fail because of issues with data quality and quantity, as well as challenges in defining the business goal.

Large amounts of data are needed to make the most of data science techniques. Large volumes of data and a wide variety of data sets help to produce more interesting insights and better results. Insurance carriers are interested in the potential insights gained from satellite imagery data, data from drones, telematics data, and other new types of available data sets. Such specific types of data can help underwriters better understand their customer base and the risks they are asked to assess. An increase in data volume leads to challenges in managing it all. Data is often trapped, becomes decentralized or is inaccessible. Historical data can lack consistency, be poorly defined and data quality often varies. Without an effective data management strategy, the success of machine learning and other data science techniques will be limited.

To make the most out of data science applications, insurers need to be thoughtful and develop a strategy for managing their data. Key elements of an effective strategy should include early alignment of data scientists and business user requirements; a commitment and plan to save all data; data provenance; centralized data access; and ensuring data quality. An effective data management strategy will lead to deeper and more accurate insights from machine learning and other applications and, in turn, better risk and overall business decision-making.