Ovum Publishes Report On Creating An Insurance Predictive Analytics Portfolio
If the past four years are a reliable guide, the insurance industry will face complexity and even chaos in the coming months and years. Insurers need to be concerned about a future which could promise "black swan" events (those that are unpredictable, infrequent, and have a severe effect), the quickening pace of customer-driven commerce, the continuing spread of the digital economy, and the annual occurrence of severe weather events historically forecast to happen only once a century.
Insurers already know that the future holds tightening regulation, demanding customers who expect a better-quality experience, aging populations, and economies still weakened by the financial crisis. Ovum's recently published report Creating an Insurance Predictive Analytics Portfolio discusses the importance of insurers using predictive analytics to prepare for future market challenges and opportunities. We also discuss the types of quantitative professionals that insurers need, data sources and data management issues, and the areas in which insurers should apply predictive analytics.
Many Opportunities Exist For Insurers To Improve Their Competitive Position With Predictive Analytics
Where to use predictive analytics is limited only by the creative imagination of the staff responsible for its application. Ovum suggests that insurers consider creating predictive analytics initiatives in the following areas, to achieve the example objective listed. More objectives and units of analysis for each initiative are detailed in the report.
- The insurance company itself as the focus of the initiative: To determine which markets to enter or leave.
- Marketing: To create a portfolio of customized marketing offers.
- Product development: To create the best product for each channel.
- Channel management: To determine which insurance agencies to appoint.
- Customer acquisition/retention: To estimate each customer's lifetime value to shape target market initiatives.
- Customer services: To estimate retirement income for each life insurance customer.
- Litigation management: To estimate litigation costs for each claim as it is reported.
- Claim management: To determine the best way to reduce loss expenses/combined ratio for each line of business and each selling agent or claims adjuster.
- Risk management: To estimate potential losses for the book of business as each new customer is added.
- Cost control: To determine how the cost levers might change for different company governance structures.
- Underwriting management: To estimate how many underwriters of what level of experience by line of business to have on staff.
The Insurance Industry Exists In A World Of Increasingly Rich Data
The insurance industry exists in a world of increasingly rich data. More and more data is available from existing sources (e.g. third-party providers offering information about weather events and forecasts, attributes of geographic locations, and consumer credit behavior), newer sources (e.g. social media), and those that are largely still conceptual (e.g. from machine-to-machine communications, also known as the "Internet of Things" — specifically from vehicle telematics).
In particular, insurers can access (although not necessarily free of charge) a never-ending torrent of (mostly) semi-unstructured and structured data from sources such as:
- insurance business systems
- social media
- embedded sensors (e.g. vehicle telematics)
- insurance company portals
- mobile apps
- location intelligence
- complementary insurance information (e.g. FICO scores, building repair cost, and business formation data).
The Data Scientist Role Is Emerging As Equally Important As The Data Miner Role In The Insurance Industry
The data miner role is no longer the only one to use predictive analytics in the insurance industry; a new role of data scientist is emerging. A growing number of insurance companies are creating new departments of these types of quantitatively skilled professionals.
A data scientist and a data miner could be the same person. But the two roles should have different perspectives regarding the scope of predictive analytics initiatives and the time horizon of predictive analytical models. Moreover, data scientists may need different skills to fulfill their responsibilities. An insurer should expect a data scientist to approach a predictive analytics initiative by first collecting data — although not necessarily all the data required to complete the initiative — and then investigating the data on an iterative basis until a coherent hypothesis emerges.
Furthermore, Ovum believes that data scientists should be responsible for models that support short, medium, and long-term corporate objectives. Data miners, however, should be primarily involved with predictive analytics initiatives that support short and medium-term corporate objectives.
