Reconciliation

What Is Reconciliation?

Reconciliation is the comparison of source data against archived data to prove the two agree. It answers the question every auditor eventually asks about a migration: how do you know nothing was lost?

The checks are deliberately unglamorous. Record counts per table confirm that every row arrived. Control totals, such as the sum of debit and credit fields or of payroll amounts, confirm that values were not altered in transit. Checksums or hash comparisons confirm that files and documents are byte-for-byte identical. Sampling confirms that business objects, such as an invoice with its line items, reassemble correctly.

Exceptions are the useful output. Every mismatch is logged, investigated, and either corrected or documented as an accepted difference with a reason. The signed reconciliation package becomes part of the decommissioning record, and it is what makes shutting down the source system a defensible decision.

Key Points

• Proof. Completion messages from a migration tool describe what the tool did. Reconciliation shows what actually arrived.
• Multiple techniques, layered. Counts catch missing rows, control totals catch altered values, and hashes catch corrupted files. None of them alone is sufficient.
• Run at the business level as well. Table-level agreement can hide broken relationships. Reconciling at the business-object level catches them.
• Exceptions must be resolved or documented. An unexplained variance is a finding. A documented, approved one is an accepted risk.
• It gates the shutdown. The source system stays available until reconciliation is signed off, since a shut-down system cannot be re-queried.

Common Use Cases

Finance sign-off

Controllers require that closing balances in the archive match the retired ledger to the cent before approving shutdown, with the reconciliation report filed as evidence.

Payroll history

Earnings and deduction totals per employee per period are compared between source and archive, since disputes years later depend on exact figures.

Clinical record migration

Patient counts, encounter counts, and attached document counts are matched between the source system and the archive, with exceptions reviewed by records management.

Reconciliation vs. Related Concepts

Several data-quality terms sound interchangeable. They check different things.

DATA VALIDATION

Confirms that data meets defined rules: correct formats, valid values, required fields. Reconciliation compares two data sets against each other.

DATA COMPLETENESS

The property being proven that every required record made it across. Reconciliation is one of the methods that establishes it.

CHECKSUM VERIFICATION

Proves that a file arrived unaltered. It does not prove that the right files, or all of them, arrived, which is what reconciliation adds.