Nine decisions the business cannot make reliably today, each sized from the governed model and traced to the systems that must feed it.
Grouped by the function accountable for it. Each opens on its numbers, then the ontology it resolves against, the models that predict on it, the agents that act, and what to ask.
4 acquired clusters still report Region-only at 55–62% coverage against 98% at mature plants — and ~7 deals a year keep arriving.
€600 M does not reach plant level at 62% coverage, against 98% at mature plants. Every period it stays here, group OTIF and stock exclude it.
Four clusters carrying €1.5 bn report region-only. At roughly seven deals a year this compounds — onboarding is a standing capability, not a project.
Mature plants report at 98% coverage. Until these four match, group OTIF, yield and stock all exclude €1.5 bn of revenue — the fastest-growing part of the book.
Link acquired plant records to group legal entities
Days until a cluster reaches 95% plant-level coverage
The same counterparty is coded differently in every system that touches it. Until those records resolve to one governed identity, every channel report, credit limit and DSO figure inherits the ambiguity.
Different keys for the same counterparty — no shared identifier. Spend, credit and terms are split, and the real relationship never appears on one screen.
Landed cost, service level and credit exposure are only trustworthy on a resolved entity graph. Skip it and the 360 is a spreadsheet that disagrees with itself — 10 of 15 records are unmastered variants and the weakest match is 86%.
The platform reads the systems Legrand already runs — as they are, no rebuild — and resolves them into one profile per entity that every use case reads from.
Most “360s” assume the hard problems are already solved. On a multi-brand group absorbing seven acquisitions a year, they are not.
| Capability | Governed 360 | Point tools | Warehouse + BI |
|---|---|---|---|
| Entity resolution & golden record | ✓Deterministic + survivorship | —Exact keys only | ✕Assumes clean keys |
| Acquired-company onboarding | ✓Plant level, day one | —Manual mapping | ✕Region-only for years |
| Landed cost & duty | ✓Per lane, per HS code | —Ex-works only | ✕Not modelled |
| Channel sell-through | ✓POS reconciled to sell-in | —Where EDI exists | ✕Sell-in only |
| Live lineage on every field | ✓Sourced & timestamped | —Partial, per tool | —Batch, stale |
| Agentic action & write-back | ✓Ask & act | ✕Read-only dashboards | ✕Reports only |
A resolved, predictive, governed platform changes what each function can decide.
Resolved model — plants, parts and counterparties deduplicated across every source system
Match confidence carried on every record, with the weakest surfaced rather than hidden
Coverage, duty, discount, channel stock and asset risk on every entity that matters
Use cases deployable on data already modelled — no new feed required to start
5 of 9 use cases compute from sources the cockpit already modelled. The other 4 — trade lanes, distributor sell-through, connected installed base and product footprint — required new tables, seeded here as modelled rows so the decision can be sized and designed; they are not live system feeds today. Model metrics, agent autonomy levels and next-best-action confidences describe the intended design, not a running system.