Cashback initiatives are a powerful way for CPG brands to build direct relationships with consumers. However, when receipt reading is unreliable, these programs can quickly become a source of frustration.

This project was designed to make cashback management accurate, scalable, and efficient through the adoption of AI technologies.

The existing solution showed critical limitations:

  • Low receipt recognition rates
  • Difficulty in correctly identifying purchased products
  • High volume of customer care requests
  • Exposure to fraud, duplicate uploads, and manual errors

These issues negatively impacted both program effectiveness and user experience.

BitBang redesigned the entire receipt reading system by introducing an AI based solution orchestrated through a structured control flow.

The solution combines:

  • Advanced OCR for text extraction from receipts
  • Product recognition models to identify eligible items
  • Safeguards against fraud, errors, and multiple uploads
  • Intelligent orchestration of AI components to maximize accuracy and reliability
  • Strong improvement in receipt recognition rate (80% vs 50%)
  • Higher product identification accuracy (50% vs 30%)
  • Significant reduction in customer care requests
  • Improved customer satisfaction
  • Overall cost reduction compared to the previous solution