In this essay
A bank statement shows a credit of 216. The sales dashboard for the same date shows 250. Before anyone calls the difference a payment loss, finance needs to establish which settlement batch the credit belongs to and which report lines belong in that batch.
This investigation starts with money that has arrived. Its artifact is a bank-credit-to-batch worksheet. A late payout investigation starts elsewhere, with an expected instruction whose arrival has not been established. A cash forecast asks a third question about future availability.

Match the identity before the amount
Record the receiving bank account, posted currency, bank reference, value or booking date as relevant, and amount. Find the provider payout or batch reference through the documented report and banking evidence. Use the appropriate merchant-account context. Two credits with similar amounts are not interchangeable candidates.
Adyen's settlement reconciliation guidance separates batch-level and transaction-level work. A batch can contain activity from different sales dates and payment methods. At transaction level, one payment can have multiple journal entries, so a payment reference alone is not necessarily a unique report row. Adyen batch-level reconciliation Adyen transaction-level reconciliation
If the bank reference is missing or the report maps ambiguously, leave the match provisional. Do not choose the nearest amount merely to clear the day's exception count. Ask the provider or bank for the supported trace and preserve what is still unknown.
Work from the batch's signed entries
The following worksheet is fictional and uses one currency. It assumes gross captured amounts, separately reported fees and the listed reserve movements. The labels and identifiers are teaching examples, not a copy of a provider export schema.
| Worksheet line | Source event | Amount included | Why it belongs |
|---|---|---|---|
| L1 | Earlier sales date, captured transactions | +180 | Assigned to the identified batch |
| L2 | Later sales date, captured transactions | +70 | Same batch despite a different sale date |
| L3 | Refund adjustment | −20 | Explicitly booked to this batch |
| L4 | Separately reported processing fee | −7 | Not already deducted in L1 or L2 |
| L5 | Reserve retained | −10 | Documented withholding in this example |
| L6 | Prior reserve released | +5 | Release joins this batch, not the original sales day |
| Expected batch amount | Sum of L1–L6 | 218 | 250 − 20 − 7 − 10 + 5 |
| Bank credit | Posted statement entry | 216 | Matched reference, amount still differs |
| Open difference | Expected minus observed | 2 | Cause not yet established |
A second currency needs its own worksheet unless an evidenced conversion connects the records. Do not add 100 euros and 100 dirhams into a 200-unit total. Where conversion applies, retain the conversion reference, source and destination amounts, rate convention and any separately identified charges.
Avoid two deductions that make a false match
First, establish whether a report amount is gross or net. If the fee is already included in net settlement, subtracting it again understates the expected credit. The worksheet deliberately uses gross values and a separate fee row so the arithmetic is visible.
Second, understand the payout balancing entry. Adyen identifies MerchantPayout as the debit that balances the paid-out batch. Treating it as another trading expense and subtracting it from an already calculated payout would count the movement twice. Use the report's documented journal semantics. Adyen transaction-level reconciliation
Keep the original export intact. Add classifications in a working layer that retains the source file, row identity and calculation version. An analyst should be able to explain why a row was included, excluded or allocated without editing the provider's evidence.
Match account, currency and supported references. The credit has arrived; its composition needs explanation.
Leave the match provisional and obtain the supported trace. Do not select a nearby amount.
Recheck the credit-to-batch identityRetain source rows; distinguish gross, net, fees, reserves and the payout balancing entry.
In the example, 218 expected versus 216 observed leaves 2 unexplained.
Keep an owned exception. A possible fee is a hypothesis, not a balancing entry.
Recalculate when new source evidence arrivesAnother analyst can reconstruct every inclusion, exclusion and correction from unchanged source evidence.
The posted credit is real within this fictional worksheet, but the two-unit difference remains unexplained. Keep the original calculation and its source rows.
Leave the two-unit difference open
The example has not reconciled. It has identified a batch, reconstructed 218 and observed 216. A possible bank charge is a hypothesis until the relevant bank or provider evidence establishes it.
Assign the discrepancy an owner and next action. The investigation might find an omitted adjustment, a mistaken batch association, a documented charge or a different posting. Record the correction with its source; preserve the first calculation rather than quietly overwriting it.
The exception-management guide covers queues and resolution evidence. For the larger architecture joining orders, provider events and cash, use three-way reconciliation. This worksheet sits inside that system as a specific explanation of one posted credit.
Before closing the case, ask another analyst to reconstruct the amount from the retained source lines. They should reach the same result and understand every unresolved entry. A matched reference with a known discrepancy is useful progress. It should remain visibly different from a fully explained credit.
The worksheet, dates implied by its labels and amounts are fictional. Provider report guidance was checked on 25 October 2026.
Closing thought and further reading
A credit on the bank statement needs an explanation tied to its batch. A signed worksheet separates a matched reference from an explained amount.
Need help applying this to your business?
Explore practical support for payment operations, product decisions, program delivery and AI automation, grounded in the challenges behind this essay.


