Lesson 5 of 6
Data validation
6 min
Comparison tells you whether totals match. Data validation goes further: it checks that the converted data is correct and usable for the business. A customer can exist with the right open items and still have the wrong payment terms on its business partner. A material can have the right stock value but a broken valuation class that will post future goods movements to the wrong account. Process owners plan and perform this validation, because they know what "correct" looks like.
Before conversion: clean and check
Finance data has its own pre-checks in a conversion. SAP provides a reconciliation program to run in ECC before converting (transaction FIN_CORR_RECONCILE, with related checks for asset accounting) that finds inconsistencies between ledgers and subledgers, so they can be corrected while still in ECC. SAP Readiness Check also includes a financial data quality analysis. During the finance data migration in the converted system, the migration steps are monitored (transaction FINS_MIG_STATUS is used for this), and unresolved inconsistencies show up there as errors. The technical team runs these tools; the process owner makes sure the findings are understood and resolved, because many fixes are business corrections.
- Months before mock 1Pre-checks and cleanupFinance reconciliation and business partner checks in ECC; owners fix findings.
- Mock 1First full validationAll four layers; expect findings in mappings and data quality.
- Mock 2Fixes verifiedRe-run comparisons; targeted samples on fixed areas plus new risky records.
- Final mockDress rehearsalSame people, reports and timings as cutover; validation should be routine.
- CutoverBusiness checkpointsSigned validation before users are let in.
A validation plan
For the field-level layer, sample with purpose. A random sample of 50 customers will miss the problems. A risk-based sample covers each account group or business partner role, customers with special terms, one-time and intercompany partners, and records the pre-checks flagged. For finance, include every reconciliation account, every bank account and a sample of open items in foreign currency.
| Object | What the business checks | Who |
|---|---|---|
| Business partners (customers) | Payment terms, dunning data, tax numbers, sales area data, credit segment and limit | Maria with finance AR |
| Business partners (suppliers) | Bank details, payment methods, withholding tax, purchasing data | AP lead with procurement |
| G/L balances and open items | Balances per account, open items by currency, reconciliation account totals | Dan |
| Materials and valuation | Valuation class, price control, standard price (for example 150 USD in Detroit, 156 USD in Chicago for CMC-ALT-100) | Plant controller |
| Fixed assets | Values by asset class and depreciation area, depreciation keys, useful lives | Asset accountant |
Field-level validation for key objects, with an owner for each.
Links to finance migration
For new implementations and selective transitions, finance data is migrated rather than converted, typically with the SAP S/4HANA migration cockpit for master data and open items, and balances loaded with a reconciled cutover approach. The same validation layers apply. The Academy's finance migration course covers balance loading and reconciliation in detail; here the point is that the process owner signs the business validation, whichever tool moved the data.
What would you do?
Two days before the final mock conversion, the IT lead suggests skipping the field-level samples this time because "they all passed last mock." What should Priya and the process owners decide?
- Data validation checks that converted data is correct and usable, not just that totals match.
- Run finance pre-checks in ECC early; fix inconsistencies there.
- Validate in four layers: technical checks, totals and counts, field-level samples, usability.
- Sample by risk, not at random, and repeat in every mock on fresh data.
Check your understanding · not graded
Lesson check
1. Which transaction runs SAP's finance reconciliation in ECC before conversion?
2. What can field-level samples catch that totals cannot?
3. How should samples be chosen?
4. Why repeat validation in every mock conversion?