Today, when an admin modifies the RPD or Semantic Model (renames a column, changes a join, adds a measure), DV workbooks break silently and only surface errors after the change is published and end users hit the workbook. This idea proposes a Staging Preview Mode where: before a semantic layer change is promoted to production, the system scans all workbooks and automatically identifies which visualizations, calculations, and filters will be impacted, displays a "Change Impact Report" listing affected workbooks and specific broken bindings, and allows admins to preview the workbook with the proposed changes applied in a staging sandbox - all before any end users are affected. This is architecturally distinct from the data lineage idea which focuses on lineage visualization, not pre-publish impact assessment.
Business Value: Eliminates production-breaking semantic layer changes; gives admins confidence to iterate the data model; dramatically reduces firefighting effort for analytics platform managers.
