An end-to-end data pipeline simulating a vendor pricing analytics system for a global food delivery platform operating across 65+ countries.
This project addresses a core challenge in food delivery platforms: how to monitor and optimize vendor pricing strategies at scale to improve affordability for millions of customers while supporting revenue growth for 1M+ restaurant partners.
| Layer | Tool |
|---|---|
| Data Generation | Python |
| Data Warehouse | DuckDB |
| Data Transformation | dbt Core |
| Orchestration | Apache Airflow |
| Visualization | Looker Studio |
| Version Control | Git |
Raw Data (Python) --> DuckDB --> dbt (Staging + Marts) --> Looker Studio Dashboard ^ Airflow DAG (Daily Schedule)
- stg_vendors: cleaned vendor master data
- stg_orders: cleaned and filtered order transactions
- vendor_performance: aggregated vendor metrics with affordability segmentation and performance tiers
- 50,000 order transactions across 1,000 vendors
- 7 countries, 8 cuisine types
- Affordability scoring per vendor
- Performance tier classification (Top / Mid / Low Performer)
- Revenue is evenly distributed across countries — no single market dominates, suggesting balanced vendor acquisition across Germany, Turkey, UAE, Saudi Arabia, Taiwan, Bangladesh, and Bosnia
- Fast Food leads cuisine distribution (14.1%) — highest order share, followed by Turkish (13.1%) and Healthy (12.6%), indicating strong demand for quick and local options
- 95%+ of vendors fall into Mid Affordability segment — average order values are moderate, creating an opportunity to introduce premium pricing tiers for high-rated vendors
- Performance tier is skewed toward Low Performer — majority of vendors have fewer than 100 orders, highlighting the need for vendor activation and engagement programs
- Recommendation: Focus vendor support programs on the Low Performer segment (800+ vendors with <100 orders). Even a 10% uplift in order frequency per vendor would significantly impact platform GMV (Gross Merchandise Value — total transaction value flowing through the platform), directly supporting Delivery Hero's revenue growth targets.
Built in Looker Studio with 4 views:
- Revenue by Country
- Order Distribution by Cuisine Type
- Vendor Performance Tier
- Affordability Segment Analysis
- python -m venv venv && source venv/bin/activate
- pip install duckdb dbt-duckdb apache-airflow pandas
- python data/raw/generate_data.py
- python data/load_to_duckdb.py
- cd vendor_pricing_dbt && dbt run