Lesson 2 of 5
Reconciliation
8 min
Reconciliation is the core of report testing. It proves that a report total equals the sum of the source documents behind it, or an independent balance, for a defined selection. If the report reconciles, its numbers can be trusted for that selection; if it does not, you have found a defect, a definition gap or a misunderstanding, and each is worth finding before go-live.
Three directions of reconciliation
| Direction | Question | Contoso example |
|---|---|---|
| Report to documents (bottom-up) | Do the documents add up to the report total? | June revenue for CMC-ALT-100 in the margin report equals the sum of the 37 billing documents in June |
| Report to ledger (top-down) | Does the report agree with the G/L balance for the same accounts? | Revenue in the margin report equals the revenue accounts in the Trial Balance for company code 1710, period 6 |
| Report to report | Do two reports that should agree actually agree? | Stock value in the inventory report equals the inventory G/L accounts |
Report to documents
Start small. With controlled test data, you know the documents. With a copy of production data, choose a narrow selection, such as one customer and one month, and list the documents with a source-level report: billing documents with VF05 or the billing document list apps, journal entries with Display Line Items in General Ledger or Manage Journal Entries, sales orders with VA05. Sum them and compare. Then use the report's drill-down to open a few documents and confirm the values trace back correctly.
Report to ledger
In SAP S/4HANA, the Universal Journal holds financial and management accounting line items in one place, so most finance reports can be reconciled to the general ledger. Use Trial Balance or Display G/L Account Balances for the same company code, ledger, period and accounts. If a profitability report shows revenue of 1,284,500.00 USD for June and the revenue accounts show 1,291,200.00 USD, the difference is a finding, even if it turns out to be an explainable definition.
Tolerances
Agree tolerances before testing, per report. For financial statements and ledger-based reports the tolerance is zero: they reconcile exactly or there is an explained difference. Small rounding differences can appear where a report converts currencies or calculates percentages and averages; accept them only when you can show the rounding rule. For operational reports, such as an open order backlog measured in pieces, the tolerance is also normally zero for a fixed selection at a fixed time.
| Difference | Likely explanation to check | Usually |
|---|---|---|
| A few cents on a converted total | Currency translation rounding per line vs on the total | Acceptable once the rule is shown |
| One document's amount | Document excluded by a filter, a document type, or a reversal not shown | Defect or definition gap |
| Exactly one period's activity | Posting date vs document date, or special period handling | Selection mismatch |
| Ratio between the two figures | Wrong currency or unit, or doubled join | Defect |
| Growing over time | Report reads a replicated copy that lags the source | Timing; confirm the refresh schedule |
Patterns of differences and where to look first.
Timing and real time
Embedded analytics reads live data, so a report run at 10:00 and a second report at 10:05 can legitimately differ if postings happened in between. Freeze the data for reconciliation by testing on a closed period, or run both sides back to back with no one posting. Where a report reads from a data warehouse or another copy of the data, check the load schedule: a dashboard refreshed nightly will not show this morning's billing.
- Define selectionCompany code 1710, ledger 0L, period 6, plant 1710, product CMC-ALT-100
- Freeze dataUse a closed period or a quiet test client
- Compute expectedSum the source documents and pull the G/L balance
- Run reportSame selection, saved variant
- Compare and explainZero difference, or each difference traced to documents
- Record evidenceSelection, figures on both sides, explanation, tester and date
What would you do?
Contoso's new margin analysis app shows June revenue for company code 1710 that is 6,700 USD lower than the Trial Balance revenue accounts. The report developer says "the app uses margin analysis data, so small differences are normal." What should Dan's team do?
- Reconcile in at least two directions: to documents and to the ledger or another trusted report.
- Most differences come from mismatched selections; write both selections side by side.
- Ledger-based reports have a zero tolerance; accept rounding only with a shown rule.
- Freeze data for reconciliation; check refresh timing for copies of the data.
- Every difference is traced to documents and either fixed or explained in writing.
Check your understanding · not graded
Lesson check
1. What does report-to-ledger reconciliation compare?
2. A report and the Trial Balance differ by exactly one period's activity. What should you check first?
3. What is the usual tolerance for a ledger-based financial report?
4. Why test reconciliation on a closed period?