Lesson 6 of 6
Testing the close
7 min
Most finance test plans cover posting: create an invoice, post a payment, check the journal entry. Far fewer cover the close, because it is hard to test. It needs a full period of data, open and closed periods in the right order, exchange rates, purchase orders in every state, and people from logistics, controlling and finance working together. That is exactly why close defects are found in production at the first month-end. A dry-run close in the test client is the remedy.
What a dry run is
A dry run executes the close calendar, task by task, in the quality system (Contoso's client 100) on data that looks like a real month: invoices in several currencies, open purchase orders with partial deliveries, assets with acquisitions, accruals, intercompany balances. The same people who will close in production run their own tasks, using the same task list. At the end, finance reconciles the result against an expectation prepared independently.
- Week -4Seed a test monthPost a realistic month of logistics and finance transactions in client 100
- Week -3Prepare expectationsDan's team calculates expected valuation, accruals and reclassification outside the system
- Week -3, day 1-4Run the closeEach owner runs their tasks from the task list and keeps evidence
- Week -2Reconcile and fixDifferences analyzed; defects raised and retested
- Week -1Year-end stepsCarryforward and special periods tested on a copy
Evidence per task
Evidence for a close task is not "the job ran." It is the result compared to an expectation. Define it before the dry run so nobody argues about it afterwards.
| Close task | Evidence | Expected result |
|---|---|---|
| Open/close periods | Period table extract before and after | Only planned periods open |
| Accruals | Posting run log; journal entries; reversal entries next period | Amounts match the accrual schedule; reversals dated day 1 |
| Foreign currency valuation | Test and production logs per valuation area; journal entries per ledger | Differences match hand calculation for sample items |
| GR/IR analysis and reclassification | Reconciliation worklist; reclassification documents | Remaining balance reconciles; reclass sides correct |
| Regrouping | Program log; adjustment postings | Debit suppliers and credit customers regrouped |
| Balance carryforward | Old-year closing vs new-year opening per ledger | Balance sheet equal; P&L net equals retained earnings movement |
Evidence and expected results for the main close tasks.
Reconciliations that prove the close
- Trial balance by company code and ledger balances to zero, and agrees with the expectation within tolerance.
- Subledger to G/L: customer, supplier and asset balances agree with their reconciliation accounts.
- Inventory value by plant agrees with the inventory accounts.
- GR/IR balance in the reconciliation app agrees with the G/L balance.
- Each ledger shows the valuation and accrual postings designed for its accounting principle.
- Run taskOwner executes and records status
- CompareResult against independent expectation
- ExplainEvery difference has a cause
- Fix and retestDefects closed; affected tasks rerun
- SignClose lead and finance process owner sign each task group
What would you do?
In Contoso's dry run, the trial balance for ledger 0L reconciles, but the IFRS ledger shows a difference of USD 2,300 in exchange rate gains. Priya has two days before the cutover rehearsal. What should she do?
- Test the close with a dry run of the full calendar on a realistic month of data in the test client.
- Prepare expected results independently, before the dry run.
- Evidence is a reconciled result per task, not a job log alone.
- Include run times and year-end steps in the test, and sign off by task group.
Check your understanding · not graded
Lesson check
1. What makes a dry-run close different from posting tests?
2. Which is the strongest evidence for the valuation task?
3. Why record task run times in the dry run?