Payment Processing

Intelligent reconciliation combines payment data, open items, and historical patterns to suggest the most likely allocation of a received payment.

visual-ai-home-4cee

From amount received to settled open items

Reconciliation is the process of linking a received payment to one or more outstanding accounts receivable entries. This not only records that funds have been deposited into the bank account, but also specifies which invoices, credit memos, or other entries are settled with that payment.

This is different from bank reconciliation. Bank reconciliation involves verifying whether bank transactions and the financial records match. In payment allocation, the accounts receivable system determines how an incoming payment should be distributed among outstanding items.

The bank transaction provides details such as the amount, currency, posting date, payer’s account, and any accompanying payment information. The accounting records include, among other things, customer numbers, invoice numbers, outstanding balances, due dates, credit memos, disputes, and payment terms. A structured payment reference can directly link both parties. ISO 20022 explicitly describes such a payee reference as a means to enable reconciliation by the recipient.

When the reference is unique and the amount received exactly matches a single outstanding invoice, a fixed rule can reliably reconcile the payment. The difficulty arises when the payment does not clearly indicate how the payer intended the amount to be allocated.

Where traditional automation falls short

Traditional reconciliation works with sequential rules. For example, the system first searches for an exact invoice number and amount, then for a customer with the same outstanding balance, and finally for the oldest invoices. Partial payments, payment discounts, and minor discrepancies can also be processed within predefined conditions. Such rules are transparent and remain valuable for unambiguous cases.

However, many payments do not have a one-to-one relationship with an invoice. A single amount can cover multiple invoices, while a single invoice can also be paid in installments. The amount paid may be reduced by a credit memo, payment discount, bank fees, a withholding, or a disputed invoice line item. The payment description may abbreviate, combine, or display invoice numbers differently than in the ERP system.

Furthermore, the payer does not necessarily have to be the same as the invoiced customer. A parent company may pay on behalf of multiple entities, a payment provider may appear as the sender, or a customer may pay from a new account number. Furthermore, the breakdown of a consolidated payment may arrive separately from the bank transaction, for example as a payment notification via email, PDF, or a portal.

Fixed rules can be expanded to cover every known pattern, but the number of possible combinations grows rapidly. The same amount can even correspond to more than one combination of outstanding invoices. A rule such as “oldest entries first” may result in an accounting allocation, but it does not prove that this was the payer’s intention.

Wider tolerances increase the degree of automated processing, but also the risk of incorrect matching. The bank balance may then be correct, while the accounts receivable system shows a paid invoice as open and incorrectly marks another invoice as paid. The receipt of the funds is a fact; their allocation may be a matter of interpretation.

How intelligent automation can improve the process

Intelligent automation can evaluate multiple clues simultaneously and rank possible allocations. In addition to the amount and the literal invoice number, factors such as the payer, historical relationships between bank accounts and customer accounts, payment date, currency, open items, credit memos, and previously confirmed payment patterns can be taken into account.

In addition, payment information can first be standardized so that differences in spaces, punctuation, leading zeros, and abbreviated references are less likely to cause data loss. When a payment advice is received separately, document processing can structure the mentioned invoices and amounts and help link them to the corresponding bank transaction. For a consolidated payment, the system can then search for the combination of items that matches both the total amount and the other indicators.

The result is not an arbitrary journal entry, but a probable allocation. If there is sufficient certainty, the workflow can process this automatically, provided all standard checks are passed. In other cases, it can display a ranked proposal with the criteria on which the link is based.

The final reconciliation remains constrained by accounting facts and policies. The allocated amount must match the receipt; an item may not be settled twice; and payment discounts, exchange rate differences, deductions, and remaining amounts may only be posted according to established rules. A probable model result must never itself create a missing credit memo, discount, or payment reference.

Intelligent automation helps determine which open items a payment is likely to settle; established financial controls determine whether that allocation may actually be posted.

Your Finance in flow

Office of the CFO solutions that bring E-transactions, Purchase to Pay, Order to Cash, and Financial Planning & Analysis into flow.