Transforming Commercial Insurance Underwriting through Data and AI

Business Objective

In the face of shrinking premium, commercial P&C insurers are constantly having to hold the line on underwriting expenses even as they strive for higher submission and policy counts. Additionally, with around 20% of data from agents and customers found to be inaccurate, commercial insurance underwriting companies need to leverage external data and AI to:

  • Validate information provided by agents and customers for pricing accuracy
  • Pre-fill application forms to deliver superior customer experience
  • Achieve substantial efficiency gains by automating information validation and pre-filling
  • Triangulation from multiple data sources is required to deliver acceptable (over 90%) hit/fill rates and information accuracy
  • Business names and addresses from application forms often are not an exact match to data found in external sources
  • External data is mostly significantly unstructured and one needs to extract the right signals

Solution Methodology
  • Prioritize underwriting questions for validation and pre-filling based on the predictive value of data to inform risk. For example, industry class, XMod, prior losses, safety violations, years in business, payroll, property characteristics, vehicles data, hours of operation, benefit plans
  • Identify external sources that enable underwriters to validate and pre-fill information:
    — Business description – Google Places, Yelp, Yellow Pages, Foursquare, etc.
    — Firmographics — D&B, Infogroup, Experian, Melissa, etc.
    — Property information – Build Fax, SMR Research, EASI, Pitney Bowes, etc.
    — Prior losses, injuries, and safety – BLS, OSHA, SOII, SAFER, etc.
  • This ML solution was deployed on the AWS cloud platform leveraging AWS S3 service for landing zone, Amazon ECR, AWS Far Gate services for CI/CD pipelines, etc.
  • Fuzzy matching, natural language processing, and machine learning to extract the right information and map it to the right business
  • User interfaces and APIs to orchestrate and integrate external information in client’s underwriting process workflows
Business Impact
  • More than half of the key questions can be pre-filled and validated with >90% accuracy

  • Up to 40% improvement in straight-through-processing (STP) rates

  • USD 100–150 million in impact identified with business teams of multiple carriers for General Liability, BOP and WC underwriting.

Copyright © 2023 Tiger Analytics | All Rights Reserved