A Fabric platform, then a scorecard of fifty metrics with drill-through

A medallion lakehouse, a semantic model and Power BI reports, then a scorecard where every metric had its own logic, the list kept changing, and users wanted to click a number and see which rows made it move.

Every metric shared a date, a category and a name, so those became the dimensions and everything that varied became long, unpivoted rows. Metric logic written and validated with the client in a T-SQL notebook, then wrapped in a PySpark function so a new metric is one script.

In Power BI, one drill-through page shows the rows behind whichever metric is clicked.

One pattern covering every metric, including those not yet defined, with logic the client owns.