The data is the deliverable.

Define the data tiers before designing the migration.

Not all data is equal. The migration plan sorts data into five tiers. Operational data, which is needed to run the business from day one and must be in Newco systems at cutover. Open transactional data, such as open orders, open invoices, and open service tickets, which must be migrated with referential integrity. Historical data, which is needed for trend reporting and operational analytics, typically the most recent 18 months to 24 months. Archive data, which is needed for legal, tax, and audit retention but is rarely accessed. And derived data, such as customer segments and analytical aggregates, which can be rebuilt in Newco rather than migrated.

Each tier has a different migration approach. Operational and open transactional data is migrated through structured extract, conversion, and load processes with full reconciliation. Historical data is migrated in bulk, often into a reporting data store rather than the operational system. Archive data may be kept at the seller under an archive only service, with read access to Newco on a service request basis. Derived data is rebuilt in Newco from the migrated source data once the source data is stable.

The tier decisions are commercial decisions, not technical decisions. The business owners decide what is needed in Newco and what is not. The temptation to migrate everything is expensive and usually wrong. The discipline of asking the business what they actually use cuts migration scope by 30 percent to 60 percent in most cases. The application portfolio rationalization pattern is covered in carve-out application portfolio rationalization.

Source ownership and access rights.

The seller controls the source. Without source access, no migration is possible. The TSA must include explicit data access rights for the buyer, including the right to extract, copy, and migrate the data that pertains to the divested business. Most TSAs include a general data access clause that is too weak to operate against. The disciplined approach is to negotiate a specific data migration schedule that lists the systems, the data scope, the extraction method, and the seller's obligation to provide the extracts on the agreed cadence.

Data scope is the second commercial battle. Sellers prefer to provide data that pertains only to the divested entity. Buyers need data that pertains to the divested business as it actually operates, which often includes shared records, shared masters, and shared history. The boundary cases sit in shared customers, shared vendors, shared employees, and shared products. Each boundary case is negotiated in writing and tracked in the migration plan.

Access rights are time bound. The TSA defines when the buyer can extract, in what format, and at what cost. Access rights typically expire at the end of the TSA, which means any data the buyer has not extracted by then is lost or accessible only under a separate archive service. The discipline is to extract early and verify completeness during the TSA, not at the end. The pre signing diligence pattern that protects the data access right is covered in TSA IT diligence.

Mock cutovers are not optional.

Three mock cutovers, in sequence, before the real one. Mock cutover one is a technical rehearsal. The extract runs, the transformation logic runs, the load runs, and the data lands in a Newco test environment. The technical team validates that the pipeline works. Functional users do not participate. The output is a list of technical defects to fix.

Mock cutover two is a functional rehearsal. The same pipeline runs, the data lands in a Newco test environment configured to mirror the target production environment, and functional users from finance, HR, operations, and IT validate that the data is usable. Open transactions can be processed. Reports tie back to the source. Master data matches business expectations. The output is a list of functional defects to fix.

Mock cutover three is a full dress rehearsal. The pipeline runs against the production source, the data lands in a production like environment, and the timing is measured. The cutover window is the same length as the planned production cutover. The business validates against the time pressure. The output is a go or no go decision for the real cutover. The application cutover pattern is covered in TSA exit application cutover planning.

Reconciliation is the proof.

A migration is not done until it is reconciled. Reconciliation is the discipline of proving that the data that left the source equals the data that arrived in the target. Reconciliation operates at three levels. Record counts, which prove that no records were dropped. Control totals, which prove that the financial values agree. And sample inspection, which proves that the data is correctly transformed at the record level. All three are mandatory before sign off.

The reconciliation pack is the artifact that the business signs against. It lists every data object in scope, the source count, the target count, the source total, the target total, the variance, and the explanation for any variance. Variances are explained, not waved away. A 0.1 percent variance in customer count may be acceptable if it is explained by inactive accounts excluded by the migration filter. The same variance unexplained is a defect that blocks sign off.

The audit lens matters. The external auditors will validate that the financial data in Newco at cutover ties back to the data in the seller's systems immediately before cutover. The reconciliation pack is the evidence. The discipline of producing the pack during mock cutovers, not at the production cutover, gives the auditors what they need without delaying the close.

After cutover comes hypercare.

The cutover is not the end. The first two close cycles after cutover are the real test. Data issues that did not surface during mock cutovers will surface during the first month end close. Open invoices that did not migrate cleanly will show as anomalies. Customer balances that drifted will show as reconciliation breaks. The hypercare period is the four to eight week window after cutover where the migration team remains active and resolves issues as they surface.

The seller's involvement during hypercare is in the TSA. The seller commits named individuals to support the buyer on data issues for a defined period at a defined rate. The buyer's migration team triages issues, identifies whether the root cause is in the source data, the migration logic, or the target configuration, and routes to the right party. Issues with the source data are the seller's responsibility. Issues with the migration logic and the target configuration are the buyer's responsibility.

After hypercare, the migration team disbands and the data ownership moves into the standard Newco governance. The migration artifacts, the reconciliation packs, and the issue logs are archived in the buyer's repository. The seller's data access rights are formally terminated. The discipline of formal closeout protects the buyer's audit position and ends the TSA's data obligations cleanly. The cost discipline that drives this is covered in TSA exit acceleration.

More on TSA exit strategy.