Executive Summary: MarketPro had huge amounts of user data but no way to leverage it. We built a customer Data Lakehouse and a real-time recommendation engine. This personalization increased average order value and conversion rates, driving roughly $1.8M in incremental revenue.
The Challenge
A general merchandise retailer with both physical stores and a massive e-commerce presence. However, they faced significant hurdles:
- Data Silos: Online and offline data were never merged.
- Generic UX: Every user saw the same homepage products.
- Slow API: Existing search was too slow for real-time suggestions.
Our Solution
We built a custom recommendation engine:
- Data Lakehouse: Consolidated data from POS, website, and app into Databricks.
- Collaborative Filtering Models: To suggest "Users who bought X also bought Y".
- Real-Time API: Served recommendations in <50ms during page loads.
Conversion Rate by Segment
Generic
Personalized
Implementation Roadmap
2 Months
Ingestion
Building pipelines from 6 core data sources.
3 Months
Modeling
Training and back-testing ML models.
2 Months
A/B Testing
Testing personalization against generic.
Ongoing
Optimization
Retraining models weekly.
Key Results
Personalized product suggestions drove an 8% increase in conversion rate. The "Recommended for You" section became one of the highest-revenue placements on the site.
"We finally feel like we know our customers. The AI suggests products they want before they even know they want them."
CMO