Start with what the business must recognise
Imagine an order system moved to a new database. The number of orders matches, yet an invoice is linked to the wrong customer. The transfer may look complete while the business result is wrong. This hypothetical example suggests a practical acceptance question: can the team recognise the same customers, orders and obligations after migration? Define those checks while the source system is still available for comparison.
Use several levels of comparison
Google Cloud’s Spanner migration guidance describes checking schema mappings, table row counts, sample records, column aggregates and row-level hashes. It also recommends custom checks for more specific validation. Our suggested plan assigns each important business record a stable identifier and compares the fields that drive decisions. Totals help identify a discrepancy; they do not explain which record or relationship needs attention. Choose checks that fit the actual migration path. Google Cloud: Validate your data migration.
Read validation status before declaring completion
For supported migration tasks, AWS DMS can compare source and target records and report mismatches. Its statistics distinguish validated, pending, suspended and failed records. AWS also notes that validation queries consume additional source, target and network resources. In the migration plan, allow time and capacity for those checks. Review records that could not be compared as unresolved work, with an owner and a documented next step. AWS: DMS data validation.
Explain intended transformations explicitly
Our proposed review lists each intentional transformation, including date handling, rounding and status mappings. Write down what the target should contain and how the comparison accounts for that change. Do not simply exclude a difficult field to make the report pass. Use representative test cases, including missing values and unusual but valid records. Ask the process owner to review customer relationships and business totals alongside the technical comparison.
Keep a decision record for the handover
Before accepting the migration, collect the comparison scope, observation time, unresolved discrepancies and the business owner’s decision. If records are still changing, agree which consistent point or controlled window the final check represents. Retain the evidence and an approved recovery plan through handover. Our suggested success measure is completion of agreed business checks, with exceptions resolved or explicitly accepted; a completed copy job alone does not establish that outcome.