Multi-entity firms — holding companies, franchises, portfolio businesses, professional services networks — share a common data problem. Each entity runs its own systems, its own chart of accounts, and its own reporting cadence. Consolidation happens in spreadsheets, weeks after month end.
The unified reporting architecture we've seen work at scale has three components: a shared semantic layer, an entity normalization pipeline, and a consolidated data store with entity-level and rolled-up views.
The semantic layer is the hardest part. Revenue means different things in a professional services entity and a product business. Headcount is calculated differently across jurisdictions. Before you can unify reporting, you need explicit definitions for every metric, agreed to by the finance leads at each entity.
With that foundation in place, normalization becomes mechanical. Each entity's data feeds into a transformation pipeline that maps to the shared schema. Discrepancies surface as exceptions rather than errors — which entity is calculating this metric differently, and is the difference intentional?
The payoff is leadership reporting that reflects reality across all 10, 20 or 50 entities, produced automatically at month end, with drill-down to any entity or transaction. The model works at any scale once the semantic layer is settled.