Product project · Data integration · Margin analysis
Marge Delhaize — Data Integration & Margin Analysis Platform
Development of an ETL and analytics platform designed to consolidate operational data from multiple systems in order to analyze real product margins over time.
Context
Independent Delhaize store owners rely on several internal systems to manage their operations.
However, these systems do not communicate directly with each other and provide only partial visibility into real product profitability.
The project aims to build a platform capable of collecting data from multiple operational systems, consolidating this information into a unified data model and providing a clear view of real product margins over time.
This allows store owners to better understand the gap between estimated margins communicated by Delhaize and actual margins observed in their operations.
The platform is designed to support decision-making at the level of individual stores or groups of stores.
Data sources
The system integrates data from three operational systems used by Delhaize stores.
Store Office
Database containing product reference information used across the Delhaize ecosystem.
Access to the data is limited and does not currently provide a straightforward export mechanism.
BabbleWay
System containing purchase information from product suppliers.
Data exports are available in structured formats such as CSV or Excel.
StoreLine
Point-of-sale system containing sales transactions recorded at store checkouts.
Exports are also available in CSV or Excel format.
Project objective
The platform acts as an ETL pipeline and analytical interface.
The long-term goal is to enable near real-time margin monitoring, cross-store margin comparison and improved purchasing negotiations for groups of store owners.
- Extract data from multiple external systems
- Normalize and structure the data in a central database
- Compute product-level margins
- Provide visualization tools for store managers
Architecture
The system follows a modular web architecture.
Backend
- Symfony 7 application
- REST API exposing data services
- PostgreSQL for centralized data storage
Frontend
- React application
- Communication with backend via REST API
Infrastructure
- Docker containers for each component
- Automated test suites for both frontend and backend
Testing
Several automated testing tools are used to maintain code quality and ensure reliability during development.
Backend
- PHPUnit
- PHPStan
Frontend
- Vitest
Use of AI-assisted development
Development of the project involves the use of AI-assisted programming tools, notably Claude Code.
The project workflow includes several specialized AI agents responsible for tasks such as task planning and backlog management, backend code implementation, code review, automated test generation, and frontend component development and review.
These agents assist the development workflow, but the overall architecture and implementation decisions remain under human supervision.
My contributions
The project is developed in collaboration with my brother, who focuses primarily on data acquisition and system architecture.
- Lead development of the backend platform using Symfony 7
- Design and implementation of the REST API
- Implementation of the data storage model using PostgreSQL
- Development of the React frontend interface
- Implementation of automated tests
- Integration of AI-assisted development workflows
Project status
The project started in September 2025.
Development of the application progressed significantly until December 2025, after which progress slowed due to difficulties accessing data from external systems.
Following recent discussions with the client, access to the required data sources may now be possible, allowing work on the extraction pipeline to resume.
Technical challenges
- Collecting data from multiple operational systems that do not communicate directly with each other
- Consolidating heterogeneous data formats into a centralized model
- Computing real product margins from purchase data, product reference data and sales data
- Handling data access constraints in external systems
- Designing a modular architecture capable of evolving toward near real-time monitoring
- Maintaining code quality with automated tests and static analysis tools
Expected outcomes / impact
The platform aims to give independent store owners more precise visibility into the margins actually observed in their operations.
It should make it possible to better understand gaps between estimated and real margins, identify problematic products or periods and support more informed decisions at store or store group level.
In the longer term, the system could also support comparisons between stores and improve purchasing negotiations for groups of owners.
Personal learnings
This project allows me to work on a strongly data-oriented problem, at the intersection of system integration, business modeling, analytical computation and visualization.
It strengthens my experience with Symfony, React, PostgreSQL, REST architectures, data pipelines and the integration of AI-assisted development tools into a supervised production workflow.