Write back planning data to semantic models

Operational and planning data have traditionally lived in separate silos. While planning data lands in a user-configured table, it often remains isolated inside the Planning artifact, unavailable for broader reporting or advanced analytics without manual intervention.

The Writeback to Semantic Model feature in Fabric Planning bridges this gap by seamlessly elevating written-back plan tables directly into the organization's existing Power BI semantic model. Ultimately, this feature ensures that planning data is correctly structured, properly related, and immediately useful across the entire Fabric analytics ecosystem.

Add writeback table to semantic model#

To integrate your planning data with the semantic model, you will first map and write back your plan destination table to the same SQL database hosting the model. Once created, you can establish star-schema relationships by defining matching join columns, appropriate cardinality, and cross-filtering behavior.

1. In the Writeback ribbon, select Add destination.
2. Select the SQL database connection. Select the database and schema where your semantic model tables reside. Enter the destination table name and select Add.

3. After you configure and add the SQL destination table, you will be able to write back to the source semantic model. Select Add table to semantic model.

4. Trigger an initial writeback to create the configured destination. Select Next.
5. Select + New relationship to create star schema relationships with dimension tables in the semantic model. Select the table to create the relationship with. This step is optional.

6. Under Writeback Table Column Name, select the common column, then select the matching column under Semantic Model Table Column Name.

7. Cardinality defines how many records in the writeback table can relate to records in the semantic model table. Select the cardinality based on your data.

8. Enable bi-directional filtering based on your data. In a standard data model (using a One-to-Many relationship):
   • Single filtering (Default): Filters applied to the One side automatically filter the Many side (for example, filtering a Customers table filters the Sales table). However, filtering the Sales table does not filter the Customers table.
   • Bidirectional filtering: Filters applied to the Many side will propagate back up to filter the One side.
9. Select Review. The Review section displays errors, if any, when creating the relationship. If cardinality and column configurations are correct, select Apply to model.

10. Refresh the catalog from the Data pane. Notice that the writeback table is added to the source semantic model. The configured destination table name is prefixed with "Plan".

11. The writeback table is automatically added to the semantic model ontology, allowing AI agents to immediately answer queries about your plan data.

Related content#

• What is Ontology