The logical layer that lets a half-migrated roll-up still answer one question consistently — model once, federate the data, generate insights anyway.
SBI Life can't wait for every office and channel to migrate onto the new Policy Admin System 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 product & channel owns its data as a product), with a semantic layer that federates them. Insights generate today; they just carry a confidence flag where a domain isn't on the new PAS yet.
Ten classes everything maps to. The Office cluster is the keystone: it's where segment, leader, product and zone reconcile.
60% of premium is already office-grain actual; the rest is read in place from legacy channel systems and reconciled — no big-bang migration required.
Every metric has one definition and a grain. The layer federates it across on-Policy-Admin and legacy domains, flagging where a value is allocated.
| Metric | Definition | Grain | How it federates across segments |
|---|---|---|---|
| Gross Written Premium | Σ premium written | office · policy | actuals where on Policy Admin; allocated from zone where not |
| Value of New Business | APE × VNB margin | segment · product | new-business value normalized to one basis |
| Renewal premium | in-force renewal premium | policy | from Policy Admin across all channels |
| Renewal mix | renewal ÷ GWP | segment | federated — same formula, many sources |
| Collection days | receivable ÷ premium × 365 | channel · office | legacy channels measured at zone grain, flagged |
| Persistency | premium retained at 13 / 61 months | cohort · policy | per IRDAI 14-Jun-2024 method, across channels |
| Cross-sell propensity | unmet need on the customer base | channel / customer | resolved across duplicate channel records |
Entity resolution matches legacy office / channel / partner codes to one canonical node — so a channel's data lines up with everything else.
Query reads each product & channel's data product in place; the semantic layer maps native Policy Admin / CRM fields to canonical metrics.
Where a domain reports at zone level, allocation disaggregates to office on learned drivers and marks it an estimate with a confidence band.
Allocated parts must tie back to the source total; anomalies and duplicate channel records & cost partners across systems 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 on-Policy-Admin and legacy channels alike; 60% of the numbers are office-grain actuals and the balance is Policy-Admin-allocated and labelled. As each domain migrates onto the new PAS, its data product's grain rises and estimates flip to actuals — the mesh closes itself.