A global operation shaped by decades of complexity
Over decades of growth, acquisitions, and regional expansion, each production site of the client had continued using its own ERP system. The resulting landscape included several local systems operating across Asia, the Middle East, Europe, and beyond. Each had its own data model, connection method, and operational constraints.
The business consequences were concrete. Procurement teams made decisions without consolidated visibility into global spend. Supplier data existed under different names and identifiers across dozens of systems. Reporting stayed partly local because the underlying information had not been brought together in consistent form.
The company had several data initiatives running in parallel, such as analytical development on the global data platform, regional ERP renewals, and a master data management programme. Norrin joined the customer's data platform function to help bring these different sources and programmes together.
Building the extract layer, one source system at a time
The work started by extracting data from individual source systems and landing it in a central platform. Norrin designed and built an extract layer on Microsoft Fabric OneLake, creating integrations for more than 30 ERP instances. Each integration had to account for the characteristics and limitations of its source environment.
The team established a modular pipeline architecture so that patterns developed for one site could be adapted and reused for subsequent ones. As the number of connected sites increased, engineering timelines shortened. Existing pipelines were migrated to Microsoft Fabric, reducing maintenance burden as the platform grew.
Making master data consistent and trustworthy for AI
Connecting source systems was only part of the challenge. The same supplier could appear under different names, tax identifiers, and internal codes across more than 30 systems. Without resolving those differences, consolidated reporting could not provide a reliable global view.
Norrin worked with the customer's team to build master data management capabilities using Semarchy. The work started with supplier data and expanded to customer master data. Data anomalies were investigated site by site, business rules were aligned across regions, and workflows were built to keep master data clean as new records arrived from source systems around the world.
This matters beyond reporting. As AI and automation take on more business processes, they draw on the same master data. When supplier or customer records are fragmented across 30+ source systems, AI operating on that data produces unreliable outputs at speed. Consistent, governed master data is what makes automation trustworthy.
Reusable data products and global insight
With a consistent data foundation in place, the team built toward reusability and reporting. Reusable data products driven by business use cases were introduced on the global data platform. Each product could serve multiple analytical and operational needs from the same underlying data, cutting duplicated development for every new reporting requirement.
The Global Spend Report gave procurement and finance leadership consolidated visibility into spending patterns across the entire global supplier base. Power BI reporting was developed and maintained alongside the platform. OneLake security was implemented to govern data access by user and geography, without requiring manual oversight for every change.
Read about data platforms
Read about AI solutions
