Reconciliation discrepancies and how to resolve them
Reconciliation discrepancies are differences between two records that should agree, such as a bank statement and the ledger or a payout and its orders. Most fall into five types: unmatched items, partial payments, duplicates, amount differences and date differences. Classify each one first, then trace it to its source and fix the cause.
Why discrepancies are worth classifying
A reconciliation that does not balance is frustrating, but the difference itself is information. A total that is off by a round amount points somewhere different than a difference of a few cents, and a missing deposit is a different problem than a duplicate bill. Teams that label each discrepancy by type resolve them faster, because each type has its own usual causes and its own fix.
The five types below match the exceptions queue in Reconciles, and they work just as well in a spreadsheet. If you are new to the underlying task, start with our guide on how to do a bank reconciliation and come back here for the leftovers.
Unmatched items on one side only
An unmatched item exists in one record and has no counterpart in the other. On the bank side, think of a fee, an interest credit, an automatic debit or a deposit nobody recorded. On the ledger side, think of a check not yet cleared, a deposit in transit or a payment entered but never sent.
Start by asking whether the item is timing or an error. Items recorded near the period end on one side and early next period on the other are timing, and they clear by themselves. Items in the middle of the month with no counterpart are errors or missing entries. Bank-only items usually need a journal entry. Ledger-only items older than a few weeks need a person to find out what happened.
As an illustration, say the bank shows a debit of $85 on the 12th with a description you do not recognize, and nothing in the ledger matches. It is not timing, because it sits mid-month. It is likely a bank fee or a subscription paid by direct debit. Find the source, book it and decide whether it needs an approval rule so it does not appear unannounced again.
Partial payments and grouped amounts
A partial discrepancy happens when amounts relate but do not match one to one. A customer pays $2,000 of a $2,600 invoice. One wire covers three invoices. A processor payout combines a hundred charges minus refunds and fees. Matching by single amounts fails in all three cases.
The fix is to look for groups. Try one to many, where one payment covers several invoices, and many to one, where several payments settle one invoice. For processor payouts, rebuild the payout from its transaction report before comparing it to the bank. Our article on the payment reconciliation process covers grouped payments in detail, and the cash application software page shows how remittances are applied to several invoices at once.
Once grouped, record what is still open. If the customer paid $2,000 of $2,600, the remaining $600 stays on the invoice as a balance due, and the discrepancy is resolved even though the invoice is not fully paid. A partial payment is only a problem when nobody records it as one.
Duplicate entries and duplicate payments
Duplicates come in two kinds. A duplicate entry is the same transaction recorded twice in your books, often once from a bank feed and once by hand, or once from an imported file that was uploaded twice. A duplicate payment is money actually sent twice, for example a vendor paid by check and again by transfer.
Duplicate entries inflate balances and are fixed by deleting or reversing one copy. Duplicate payments cost real money and need recovery from the vendor or customer. Look for the same amount to the same payee within a short window, and for invoice numbers that differ only in formatting. The guide on how to reconcile accounts payable has a section on catching them in vendor ledgers.
Amount differences and how to read them
When two items clearly belong together but their amounts differ, the size of the difference is a clue.
- Divisible by nine: likely transposed digits, such as 3,450 entered as 3,540, a difference of 90.
- Exactly double an item: an item entered with the wrong sign, a debit recorded as a credit.
- Equal to a known fee or tax rate: a processor fee, a bank charge or sales tax handled differently on each side.
- Small and varying: currency conversion or rounding, common with payouts in another currency.
- Round and unexplained: a manual adjustment or a misposted journal entry worth a closer look.
Record the reason when you resolve the item. If the same reason appears every month, such as fees netted against revenue, change the process that creates it rather than correcting it each time.
Date differences and cutoff
A date difference is a pair with the same amount and reference but dates far enough apart to fall outside your matching window or into different periods. Checks that clear weeks later, card settlements that post a few days after the purchase and payouts that arrive after month end are the usual cases.
Agree on a date tolerance that fits each account. Three days is common for card and processor activity, while checks may need several weeks. At month end, list date differences that cross the period as timing items with an expected clearing date. If an item does not clear by that date, reclassify it as an error and investigate.
Document the tolerances you use in the reconciliation file. When an auditor sees that a match was accepted with a five day gap, the first question is whether five days is your rule or a one-off judgment, and a written tolerance answers it.
Preventing reconciliation discrepancies at the source
Resolving discrepancies is necessary, but preventing them is cheaper. Most recurring discrepancies trace back to a handful of process gaps.
- References missing from payments: ask customers to quote invoice numbers and pass order IDs into processor metadata.
- Manual entry alongside feeds: pick one source for each transaction type so nothing is entered twice.
- Fees and refunds netted: book gross revenue, fees and refunds to separate accounts.
- No clearing accounts: use one per processor or marketplace so payouts and their contents are reconciled separately.
- Old items rolled forward: set an age limit after which an open item is escalated.
Automated reconciliation software helps here too. Reconciles classifies each discrepancy into unmatched, partial, duplicate, amount difference or date difference, explains why it was not matched, and lets you save a rule when the same pattern repeats. Every match has a reason and a confidence score, and nothing posts to your ledger without review. Read how automated reconciliation works for the details.
Whatever tool you use, review the discrepancy log once a quarter. Count how many items of each type you had and where they came from. The source that produces the most discrepancies is where a small process change saves the most time, whether it is a bank, a processor, a vendor or an internal habit.
