Our Achievement
300+ Users, 18 Countries, 1 Platform: Zeelandia Group’s Data Transformation and Direct Path to AI
Zeelandia is a global leader in bakery and confectionery ingredients, operating in more than 30 countries worldwide.
Facing a dynamic international market, the company recognized that sustaining growth and managing the business effectively required unifying its decentralized reporting environment.
Data fragmented across local subsidiaries made it impossible to get a complete view of the business and slowed down strategic decision-making at group level.
Zeelandia therefore turned to BizzTreat to jointly build a modern end-to-end analytics hub on Microsoft Fabric. The goal was to deliver, on an ambitious timeline, a solution that would set global standards while giving local subsidiaries room to address their own analytical needs.
Zeelandia
Zeelandia H.V. is a Dutch company specializing in bakery ingredients.
For the first time, a global view of performance across all markets in one place, in real time.
Technological readiness for AI adoption.
At the same time, the company is undergoing a major transformation, consolidating dozens of local ERP and PLM systems into a single group-wide solution. Zeelandia is building on the Microsoft ecosystem and aims to use modern technology to precisely manage purchasing, sales and financial planning at a global level.
Key challenges with the existing solution
Extreme decentralization of reporting: the use of many different reporting tools and data warehouses across local subsidiaries led to chaos.
No “single source of truth”: the lack of a central record caused data inconsistencies and left stakeholders unsure where to look for the right information.
Low data quality and adoption: fragmented data and the absence of Master Data Management undermined users’ trust in reporting.
Technological barrier: the need to connect users from different Microsoft tenants.
Challenges
Unifying a fragmented technology architecture: the need to integrate data from many different ERP and PLM systems into a single, coherent whole.
Working with new technology: the decision to build the solution on MS Fabric, a then-unexplored platform, brought both risks and technical challenges.
Ambitious deadlines: the architecture had to be designed and first results delivered before key subsidiaries migrated to the new group-wide ERP.
Goals
Build an end-to-end analytics hub as the single source of truth for the entire group
Analysis and prototyping: validating the proposed logical and technology architecture through a Proof of Concept (POC) and testing MS Fabric’s capabilities for data extraction and sharing.
Data management – MS Fabric: extracting data from source systems (Infor M3 and SFTP from partner subsidiaries) and cleansing it in a medallion architecture (layers L00 to L30).
Using custom Python libraries to automate incremental processing and snapshotting.
Establish global reporting standards while preserving local flexibility
Creating semantic models for key areas: Sales, Procurement and Finance.
Building robust data governance for effective access management at the level of individual areas, divisions and their combinations.
Monitor data quality and implement Master Data Management
Ensuring data reliability and users’ trust in reporting.
Set up an architecture ready for AI agents and LLMs, including a semantic layer, a data catalog and integration with MS Copilot Studio.
Deliverables
Central data warehouse in MS Fabric for all 18 countries: all company data (sales, purchasing, finance) available in one place in a clear medallion structure, integrating 18 countries from two data sources (Infor M3 and SFTP).
Data foundation for implementing AI agents: using MS Copilot Studio integrated into MS Teams for conversational analytics and querying metadata.
Automated documentation, data catalog and data lineage powered by LLMs: integrating Power BI, MS Fabric and Infor metadata to automatically document semantic models.
Automated data pipelines: data cleansing and transformation run without manual intervention.
Interactive Power BI dashboards: unified global reports with the option to drill down into local data. The solution is now actively used by more than 300 users across 124 security groups.
01/
Global overview with local detail
Management now has data from all subsidiaries in the same format, making it possible to compare market performance instantly.
02/
Fast error detection with DQA
Automatic alerts on data anomalies are sent out immediately, and an error caught at the source is usually fixed by the next day. The business no longer has to deal with discrepancies weeks after the fact.
03/
Self-service analytics
Three key power users at Zeelandia can build their own semantic models on top of the data layer, and the business divisions have added more than 20 dashboards of their own.
04/
Reduced technical debt
The move from many local systems to a single modern platform has simplified IT maintenance and improved data security.
05/
Release cycle efficiency
With DEV/Staging/PROD environments and Git version control in place, new analytics features can now be deployed dramatically faster and without the risk of downtime.
06/
Multi-tenant access
Úspěšné vyřešení přístupů pro uživatele napříč různými firemními tenanty umožnilo skutečnou mezinárodní spolupráci v reálném čase.
What Was It like Working with BizzTreat?
“We set out to build our central data platform on Microsoft Fabric (20 countries with different ERP systems) at a time when it was a completely new technology. We appreciate the courage and expertise with which the BizzTreat team took this on with us. Their professionalism and commitment turned our vision into a working solution, and today they continue to help us develop and scale our data environment.”
Michal Rada
Group IT Director, Zeelandia Group
Our collaboration with Zeelandia has shown that even in a complex international environment, the courage to innovate and the right choice of technology can bring order to data. The project not only resolved the immediate pain points of decentralized reporting but also laid a solid foundation for the future development of advanced analytics and machine learning across the entire group.
The results speak for themselves: more than 300 users across 124 security groups now work with data from 18 countries in a single, unified environment. Data errors are caught and fixed by the next day, business divisions build their own analytics on a shared data layer, and management has, for the first time, a truly global view in real time.
But perhaps the most important result isn’t visible in the dashboards. Zeelandia is no longer asking whether to adopt AI. The infrastructure, the semantic layer and the first AI agents in MS Teams are already in place. The only question now is what comes next.
