ANOMALY DETECTION IN INVOICE PAYMENTS

About the Client

The client is an automobile manufacturer. Client is one of the largest vehicle manufacturers by production in India and the largest manufacturer of tractors in the world  

Business Challenges

  • The client receives over 400,000 invoices from its vendors every year 
  • As some items are rejected, there are credit notes issued by the vendor 
  • The client had built a set of business rules to avoid any duplicate of wrong payment to the vendors
  • These rules were based on experience and there was an opportunity to build self-learning rules 

Solution

Our solution was an AI based anomaly detection model which would identify potential problematic payments. The solution included the following capabilities

  • Identification of anomalies based on past payment amount and timing 
  • Scoring of transaction based on three different algorithms 
  • Identification of transactions that had a high anomaly score 

Benefits

  • The new model along with the previous business rules improved the confidence in the financial controllership