Business Analytics For Insurance

Harnessing Big Data in Insurance

In the context of an insurer’s three major functions – marketing, underwriting, and claims – Predictive Analytics is both revolutionary and evolutionary. Predictive Analytics is evolutionary to underwriting, and revolutionary to marketing and claims.


New customers

Target new customers with greatest likelihood to buy, and to produce the greatest profitability and relationship longevity. Create a model with a score to identify high priority and lower priority prospects. Focus Marketing and Sales efforts on the higher priority prospects, reducing wasted time on the lower priority prospects. This can greatly improve the “hit ratio” for the Agents.

Customer retention

Goal is increasing “retention ratio”.

  • Analyzing why customers are lost, and identify factors that can be improved to keep more customers longer.
  • Identifying which customers may be about to leave to a competitor and address their needs before they leave.

Increase sales to existing customers

Customer analysis and segmentation: Up-sell and cross-sell products through more targeted marketing. By performing “market basket” analysis, the optimal combinations of products can be understood, giving direction to marketing campaigns. Analyze data from internal customer history and industry data.

Product development

Develop new products and tailor existing products with greatest likelihood of profitability and adoption. Analyze data from internal customer history and industry data.


Screening new applicants

Analyze past customers and customer groups to establish a screening model to measure new applicants against. If they don’t pass this initial screening, then don’t waste further time researching and analyzing them.

  • Predict likelihood of claims based on individual and group characteristics such as demographics, property characteristics, past claim history, etc. Predict the policy’s ultimate cost. This determines appropriate pricing.



Analyze past trends for patterns in individuals and groups to identify (create a profile with scores) and predict future fraud activity by individuals and groups. An estimated $30B a year in fraudulent claims is paid. This can solve 2 problems:

  • Type I errors: Mistakenly identifying a legitimate claim as fraudulent (anger customers, possible litigation, etc.)
  • Type II errors: Not identifying fraudulent claims and paying them

Size of Claims

Score likely claims by size of settlements, allocating internal resources to higher priority (cost) claims. More individual attention by more highly qualified people can be applied, while leaving those scored as likely having lower settlement costs to be handled by more automated processes.

For more information on CGN's Business Analytics practice, contact Syamala Srinivasan at