Marcus Delacroix

Alternatively, build a seasonal index measure in your semantic model (each month's % of annual total, based on 3-year average) and use it as the distribution weight when allocating annual forecasts to months.

Also check that your service account (or user identity) has Build permissions on the SQL database. Writeback needs write access, not just read. This is separate from the Fabric workspace role.

Direct Lake is recommended for planning workloads. The key to preventing DQ fallback is keeping your framing tables (dimension tables) small and warm. Load them into memory during off-peak hours.

This is exactly the reference architecture I've been looking for. Did you use RLS on the Planning sheet to prevent regional managers from seeing each other's inputs, or did you keep it open during the planning cycle?

We use Fabric Dataflow Gen2 with the SAP BAPI connector for delta loads into a Bronze layer, then transform in Spark to a Gold layer semantic model. Works well but the SAP authorisations for the BAPI calls need a dedicated…