Motivation
Why CRISP exists
Most estates already have landing zones, cleaned staging tables, and curated marts. That is necessary plumbing. It is not enough when the same party appears in three systems, finance and actuarial must share one GWP definition, and regulators ask what you believed on close day.
Strengths of each — held in one method
CRISP keeps what medallion, Data Vault, and dimensional modelling already do well, then adds the trust boundaries those approaches leave open.
Familiar strengths · one method
Medallion
Layered hygiene
Capture → clean → curate still happen; CRISP does not replace landing.
Data Vault
Audit history
Insert-oriented, history-preserving assertions — raw trail stays reconstructable.
Dimensional / BI
Consumer grain
Business-ready shapes and measures — but owned in Publish, not per mart.
Only CRISP treats these as trust boundaries
Beyond the union of the three
Cross-source conform
One programme entity contract across every feed — not per-lane silver.
Release-controlled Rules
Inspectable catalogues for maps, sign, and lifecycle — not buried CASE logic.
Governed Identity
Determine cross-source matches — sparse Identity clusters as a stage, with exceptions and rebuild discipline.
Span contracts
Represent many as one; AS_IS / AS_WAS / AS_WAS_AT as named projection APIs.
Publish-owned KPIs
Consumer facade of KPIs at any point in time — BI and APIs consume; they do not invent.
What other methods leave open
| Familiar approach | Gap | CRISP answer |
|---|---|---|
| Medallion bronze / silver / gold | Organises storage quality. No trust boundary for programme rules, cross-system identity, or replay contracts. | Capture + adaptor input stay outside CORE. CORE = CRISP stages with named ownership. |
| Data Vault hubs / links / sats | Excellent audit raw history. Does not by itself define enterprise canon columns, Identity clusters, or governed KPI surfaces. | Conform holds assertions; Identity determines matches; Publish names KPI contracts. |
| Dimensional marts + semantic layers | Each mart or BI model can redefine grain and measures. Drift is inevitable under multi-consumer pressure. | Span represents many as one across time; Publish freezes KPI names. Consumers are peers on one schema. |
| In-place “current row” masters | History is overwrite. You cannot answer “what did we believe on close day?” without snapshots or luck. | Tritemporal Conform + Span AS_IS / AS_WAS / AS_WAS_AT separate restatement from audit. |
Five capabilities storage layers do not provide
- 01
Cross-source conform
Medallion silver cleans within a lane. CRISP Conform + Rules make programme meaning hold across every feed before Identity runs.
- 02
Tritemporal facts
Valid time, system (assertion) time, and warehouse control time on every Conform fact — so close-day defence is a contract, not a rebuild.
- 03
Governed catalogues
Rules live as inspectable, release-controlled rows — maps, sign, lifecycle — not scattered transforms that silently redefine truth.
- 04
Identity as a stage
Determine cross-source matches — same party or account across PAS, claims, and finance is a stewardship boundary with a four-step rebuild.
- 05
Publish owns KPIs
Consumer facade of KPIs at any point in time. BI tools consume; they do not invent the measure.
CRISP is not a lakehouse replacement
Capture and staging still happen — often as bronze and silver in practice. CRISP sits above that as the conformity layer: what crosses into CORE, how rules and identity are owned, and what every consumer reads from Publish. Most estates use both.