ACCOUNTS RECEIVABLE · CASH APPLICATION · RUNS IN YOUR BROWSER

Which invoices did this payment pay? Match bank receipts to open invoices.

A customer sends one wire for three invoices, takes 2% off one of them, and their bank keeps $15. The memo says "INV 1001/1002". This tool reads your open-invoice export and your bank's incoming payments and proposes, for every receipt, which invoices it settles, with a confidence and the reason in plain words. Built for AR clerks, bookkeepers and small finance teams who do cash application by hand today.

1. Open invoices

An A/R aging or open-invoice CSV: invoice number, customer, invoice date, terms and open balance. QuickBooks, Xero and most ERP exports work as they are.

2. Bank receipts

Your bank's CSV of incoming payments: date, amount, payer name and the memo or remittance text. Outgoing (negative) lines are ignored.

as a bank charge

No files to hand? : 11 made-up invoices and 8 made-up bank lines with one of each result.

Who it is for

The person who opens the bank portal every morning, sees "ACH CREDIT HARBOR FREIGHT LOGIST 5,160.25" and has to work out that it is invoices 1006 and 1007, less a $15 wire fee, before marking them paid. If your customers pay by ACH, wire or cheque and you apply the cash by hand in a spreadsheet or in your accounting system's receive-payment screen, this does the first pass for you. If you already run an AR automation platform with lockbox and remittance capture, you do not need it.

How it matches, in order

  1. Invoice numbers in the remittance. "INV-1023", "INV 1023/1024", "1023, 1031" and "001023" are all read. If the payment equals the named invoices (after any discount the terms allow), that is a high-confidence match.
  2. The payer name. Banks truncate and shout ("HARBOR FREIGHT LOGIST"), drop "LLC", and write "MFG" or "ST". The payer is compared with your customer list token by token, allowing those changes.
  3. The amounts add up. Among the recognised customer's open invoices, the tool searches for the combination whose total equals the payment, allowing an early-payment discount where the terms give one and a bank charge up to your limit.
  4. Ties go to the oldest invoice. When two combinations fit, the oldest invoices are proposed and the row says so.
  5. Amount only, last. A cheque with no name that equals exactly one open invoice is proposed for review, never as a sure match.

Read the worked example of one payment across several invoices, with a discount and a short-pay.

Common questions

What files do I need?

Two CSV exports: your open invoices (an A/R aging or open-invoice report with an invoice number, customer, date, terms and open balance) and your bank's incoming payments (date, amount, payer name and the remittance or memo text). Common header names are recognised automatically.

How does it decide which invoices a payment pays?

In the order a clerk would: invoice numbers in the remittance text first, then the customer recognised from the payer name and a combination of their open invoices that adds up to the payment, then an exact amount with no name. Every proposal shows the method, a confidence and a sentence explaining it.

Does it handle early-payment discounts like 2/10 net 30?

Yes. If an invoice's terms read 2/10 net 30 and the payment arrived within 10 days of the invoice date (plus a 2-day posting grace), the discounted amount settles the invoice and the discount is shown. A payment short by exactly the discount after the window closed is flagged as a late discount for you to allow or chase, not written off.

What happens when two combinations of invoices add up to the same payment?

The oldest invoices are proposed, which is the usual first-in first-out rule, and the row is marked Review with a note that another combination fits. Check the remittance advice before posting.

Is my data uploaded?

No. Both files are read in your browser and never leave your device. The optional usage counter stores only how many times features were used, in your own browser.

Does it post to QuickBooks, Xero or my ERP?

No. It produces a review list and a CSV with one line per invoice application (applied amount, discount, write-off, balance left open) that you post yourself or import.

How accurate is it?

On a generated test set (not real statements: 1,606 made-up receipts across five seeds, with truncated names, retainer invoices of equal amounts, discounts, fees and short-pays), 99.8% of High proposals and 97.3% of Medium proposals named exactly the right invoices, against 63.5% of Review rows, which is why those are marked for review. The benchmark is in the source; it has not been measured on a real company's statement yet.