Early Fraud Detection and Improved NPS with Claims AI Cloud

, Claims AI Cloud Cuts Down Insurance Fraud Cost and Boosts Net Promoter Score


Claims AI Cloud

A leading P&C insurance carrier was suffering from fraud-related losses and damages while processing claims. They were looking to minimize this loss and serve its customers better approached InsurAnalytics.ai to help solve this problem.

Claims AI Cloud, an advanced AI-powered data analytics platform, has the capacity to identify and predict the propensity of a fraudulent claim quite early in the cycle. By making a close study of the insurer’s data and leveraging the existing model built on the platform, it was possible to develop and deploy a model that would be able to use pattern recognition and predictive analysis to identify potential claims fraud.


The insurer had a rule-based system for authenticating and flagging fraudulent claims; fraud detection was largely dependent on the experience of the adjuster. The problem with this approach was that it was completely dependent on the adjuster’s individual capability and drive and tends to be highly subjective. In certain cases the adjusters did not have access to, or were simply not privy to, all the data related to the incident or claimant; consequently leading to error-prone decision making. The limited Special Investigations Unit (SIU) bandwidth makes it harder to investigate every case referred to them. Another indirect effect was increased in end-to-end cycle time for claims settlement negatively impacting the Net Promoter Score (NPS).


Claims AI Cloud is a purpose-built platform for insurance claims, with pre-built connectors to ingest and synthesize a variety of claims related data sets, pre-trained models and API based integration; the platform can quickly identify relevant data patterns to help with detection of fraud. It looks at the points that lead to the referral, and adds these learnings to the pattern recognition algorithm. Our advanced deep learning neural network is able to pick up these patterns and apply the learning contextually to existing and upcoming data, while also adjusting for new information, such as new networks or fraud methods. The net result was:

  1. Reduced triaging effort by adjusters
  2. Improved referral rate and cycle time and better hit rate for Special Investigations Units (SIU)
  3. The insurer was able to realize significant ROI by identifying and preventing an additional 10% of fraudulent claims


Hit Rate for SIU
Reduction in SIU Referral Latency
Improvement in Loss Ratio
Reduction in Expense Ratio