New consolidated company, old history, and no database to query
The group had just merged three Business Central companies into one live company, keeping the old ones as read-only history. Business Central online gives no direct database access, so every table had to come through its APIs. Reporting still had to see across the live company and its history, and retail sales data was being downloaded by hand every month.
Prove it during discovery, then build it for production
We built a working extraction on a Fabric trial during the discovery, measured the real data volumes, and used them to size the platform and fix the price of the build. That pipeline then became the production lakehouse: Business Central data landing in Bronze, cleaned in Silver and modelled in Gold.
Three API surfaces, one pipeline
99 tables now come through three Business Central API surfaces: 95 per company and 4 shared across the tenant. No direct database access is needed.
About 35 minutes for a full pull
The live company's data, under 10 GB, extracts in around 35 minutes across about 80 endpoints.
A table drives everything
A metadata control table lists what to extract and from which company. Adding a company or a table is a new row, not new code.
History brought across
The three companies merged into the live one were loaded once, in around two hours, so reports can see across the group's full history, not just the period since consolidation.
Capacity sized from real numbers
Because the data volumes were measured, not guessed, the platform was sized to the smallest Fabric capacity that fits, F2, with room to grow.
Retail data on a schedule
The monthly manual download of retail sales files, about 20 files and 6 to 7 GB a month, was replaced with an automated load into the Silver layer.
Where the data lives and how it flows
One live company plus three historical companies kept read-only after the group consolidated them, read only through Business Central's APIs
What changed
Business Central data now lands in Fabric on a schedule, across the live company and its history, and the monthly retail sales files load automatically instead of by hand.
Adding a company or a table is a configuration change rather than new development, so the platform can grow with the group.