Data Reconciliation Methodology for Enterprise Data Migration and Conversion Accuracy

Abstract

Life insurers periodically retire legacy policy administration systems in favor of modern platforms, moving an entire in-force block of business, its coverages, and every party attached to each policy from one data model to another. A single migration can involve tens of thousands of policies and hundreds of thousands of individually mapped attribute values, far more than manual spot-checking can verify. This paper describes a data validation reconciliation methodology by using Business Intelligence tools. The methodology compares source and target extracts record by record and attribute by attribute after applying canonical transformation rules, classifies every comparison as a match, a mismatch, or a missing value on either side, and aggregates the results into per-attribute and per-record validation scores refreshed daily.