The Challenge
Staples, a giant in office retail, wanted to increase their average order value (AOV) for B2B customers. Their existing recommendation engine was rule-based and generic, failing to account for the specific buying patterns of different business verticals (e.g., schools vs. law firms).
Our Solution
We implemented a real-time recommendation engine using collaborative filtering and deep learning.
- Data Unification: Combined browse history, purchase history, and firmographic data into a single customer 360 view.
- Algorithm Selection: Deployed a hybrid model (Matrix Factorization + Neural Nets) to predict "next likely purchase".
- AB Testing: rigorously tested the new recommendations against the legacy control group to validate lift.
The Outcome
The new engine was a massive success. Customers found relevant products faster, leading to a significant bump in conversion and cart size.
Key Metrics
- 15% Increase in Revenue per Visit.
- 8% Lift in Average Order Value (AOV).
- 22% Improvement in "Add to Cart" rate.