A pragmatic 90-day playbook for decommissioning Teradata

Teradata decommissioning rarely fails because the target is wrong. It fails because the migration scope is wrong — the team tries to lift everything at once, parity is unprovable, and the cutover slips by quarters.

Pick a wedge with a defensible boundary

Customer-360 is a good first wedge: bounded entities (Customer, Campaign, Touchpoint, Conversion), real consumers, measurable outcomes. The wedge gives you a reconciliation surface you can prove parity against.

Scope the ontology, not the migration

The wedge ontology is the contract. Anything not in the ontology is out of scope until phase two. This single decision is what keeps a 90-day playbook from becoming an 18-month program.

Reconciliation as a gating criterion

Cutover happens when reconciliation passes for N consecutive batches with zero drift. Not earlier. The gating criterion is concrete, not subjective.