Reports were slowing down the systems the business runs on
Reporting queried the group's operational systems directly. As reporting grew, those heavy queries competed with day-to-day work and slowed the live systems down. The group also needed to choose between building a data platform from separate Azure services or adopting Microsoft Fabric.
Compare the options on real numbers, then build on Fabric
We first compared Azure (SQL Database, Data Factory, Databricks and Data Lake) with Microsoft Fabric for a representative daily workload, on both monthly cost and build effort. Fabric came out cheaper and faster to build, so we delivered a Fabric lakehouse and Power BI model that takes reporting off the live systems.
A side-by-side decision
For a daily load of around 500,000 rows, Fabric came to about $262 a month against about $470 for separate Azure services, and 17 build days against 24.
Bronze lakehouse
Existing reporting views were landed in a Bronze lakehouse and profiled for structure, completeness and quality.
Gold lakehouse
Curated fact and dimension tables in a Gold lakehouse, shaped around how the business actually reports.
Automated movement
Pipelines and Dataflows move data from Bronze to Gold on a schedule, with no manual steps.
Performance tested
Refresh times and query response were measured on an F2 capacity, and results checked against the existing systems for accuracy.
Ready for Copilot
A Power BI semantic model and dashboards sit on the Gold layer, built so Microsoft Copilot can work with it.
Where the data lives and how it flows
Existing reporting views from the live systems, landed once instead of queried all day
What changed
Reporting now runs from a Fabric lakehouse rather than the operational systems, so the live systems are free to do their job and reports stay fast.
Because Fabric includes pipelines, notebooks, storage and the warehouse in one capacity, the group pays one predictable bill instead of tracking four separate services.