Claims Fraud Monitoring (CFM)

Improving precision and recall of finding suspicious transactions.

Deployed in

One of the leading Private General Insurance Business in India

Key Insights

The objective of this engagement was to explore’s AI Supervisor module to increase fraud monitoring coverage and efficiency and also improve ROI of the investigation process.

  • The insurer had their own fraud monitoring system, and they also leveraged TPAs' fraud monitoring systems. These systems were built using ‘triggers’ and rules set up a while back, and human experts were needed to update them periodically, to maintain relevance. Currently 1 in every 13 cases are sent for investigation by the rule engine system; the majority (about 72%) of these cases were found, post investigation, to be normal. Through being wrongly labelled as ‘suspicious’ the TAT for such cases went up by 3x to 10x, hurting customer satisfaction, causing wasted investigation costs, and lowering the ROI of the investigation process. The team also suspected that significant fraud slippage was occurring harming the bottom line.

  • Claims Fraud Monitoring (CFM) increased the efficiency of the fraud monitoring process, while also netting more new fraudulent cases for investigation. The module was able to generate fraud triggers in real-time, save dollar value fraud cases were prevented and also developed the agility to adapt to newer fraud trends. The module was able to learn the normal claims pattern, and raise alerts in case of any anomalies.

  • CFM module uses autonomous processes enabling self-learning from multi-dimensional feedback – to learn quickly, including absorbing from manual feedback. It is able to leverage a host of datasets (structured and unstructured) and deliver expert level accuracy in near real time.

Get Casestudy