The logical layer that lets a half-integrated roll-up still answer one question consistently — model once, federate the data, generate insights anyway.
Stark can't wait for every acquired house's ERP to migrate before it gets answers. The fix isn't one warehouse — it's a shared ontology (so everyone means the same thing) over a data mesh (each house owns its data as a product), with a semantic layer that federates them. Insights generate today; they just carry a confidence flag where a house isn't integrated.
Ten classes everything maps to. The Showroom is the keystone: it's where house, leader, entity and geography reconcile.
92% of revenue is already showroom-grain actual; the rest is read in place from legacy systems and reconciled — no big-bang migration required.
Every metric has one definition and a grain. The layer federates it across integrated and legacy domains, flagging where a value is allocated.
| Metric | Definition | Grain | How it federates across houses |
|---|---|---|---|
| Revenue | Σ recognized revenue | showroom · order | actuals where integrated; allocated from area where not |
| Adjusted EBITDA | revenue − COGS − SG&A (+ add-backs) | house · entity | entity P&L normalized to one chart of accounts |
| Repeat-designer revenue | spend from retained trade accounts | account | from CRM/order systems across all houses |
| Repeat-designer mix | repeat revenue ÷ revenue | house | federated — same formula, many sources |
| DSO | AR ÷ revenue × 365 | entity · showroom | legacy entities measured at area grain, flagged |
| Gross margin | (revenue − COGS) ÷ revenue | order · house | mapped via canonical cost categories |
| Designer-account retention | expansion − attrition on base | account | resolved across duplicate account records |
Entity resolution matches legacy entity / brand / showroom codes to one canonical node — so Scalamandré's data lines up with everything else.
Query reads each house's data product in place; the semantic layer maps native fields to canonical metrics.
Where a house reports at area level, the engine disaggregates to showroom on learned drivers and marks it an estimate with a confidence band.
Allocated parts must tie back to the source total; anomalies and duplicate accounts/suppliers across houses are surfaced.
This is not theoretical — it's how this cockpit already works. The Story, Briefing and 360 views read the same governed metrics over integrated and non-integrated houses alike; 92% of the numbers are showroom-grain actuals and the balance is area-allocated and labelled. As each house integrates, its data product's grain rises and estimates flip to actuals — the mesh closes itself.