Content authored by practitioners with experience at Amazon India, Intuit QuickBooks, and the Tata Group. Meet the team →
A junior accountant matching a bank statement to the ledger finds receipts that cover several invoices, settlements that combine many sales and amounts that differ slightly, and does not know how to pair them.
Transaction matching pairs lines across two records on keys such as reference, amount and date. Matches can be one-to-one, one-to-many, many-to-one or many-to-many, and a partial match records the amount difference separately.
A match order (reference first, then amount, then date window), a written tolerance for rounding, and a column recording the match type and any difference for each pair.
Every line on both sides is either paired with its counterpart, with any difference explained, or listed as an open reconciling item for follow-up.
Transaction matching is the part of reconciliation where you pair each line in one record with its counterpart in the other, using keys such as the reference number, the amount and the date. A match can be one-to-one (one bank line, one invoice), one-to-many (one payment for several invoices), many-to-one (many sales, one settlement) or many-to-many. Lines left unpaired become reconciling items.
What are the four types of match?
| Match type | In words | Indian example |
|---|---|---|
| One-to-one | One line on each side | A customer’s NEFT with UTR quoted against a single invoice |
| One-to-many | One line on one side, several on the other | One NEFT from a distributor paying five invoices |
| Many-to-one | Several lines on one side, one on the other | A day’s card sales on a POS machine settled as one bank credit the next day |
| Many-to-many | Several lines on each side | A marketplace that nets sales, returns and fees across days and pays in two instalments |
One-to-one is the easiest and the most common in small businesses. As volumes grow, one-to-many and many-to-one become the norm, because customers bundle payments and banks and gateways bundle settlements.
A worked example: one receipt, three invoices
A customer sends ₹2,36,000 by NEFT. The narration says only the customer’s name. Their open invoices are:
| Invoice | Amount |
|---|---|
| INV-201 | ₹1,18,000 |
| INV-207 | ₹59,000 |
| INV-212 | ₹59,000 |
| INV-215 | ₹88,500 |
No single invoice equals ₹2,36,000. Trying combinations, INV-201 plus INV-207 plus INV-212 gives ₹1,18,000 plus ₹59,000 plus ₹59,000, which is exactly ₹2,36,000. This is a one-to-many match, and INV-215 stays open.
Because a different combination could add up to the same figure, it is good practice to confirm with the customer’s remittance advice or an email before closing the invoices.
Which fields do you match on first, and why?
Match on the most specific field first, then widen.
- Unique reference. The UTR on NEFT and RTGS transfers, the UPI transaction reference, or the cheque number. If both records carry it, the match is almost certain.
- Invoice or document number. Many customers type the invoice number in the narration. It is reliable when present but often abbreviated.
- Amount. Exact amounts are strong evidence, but two invoices of ₹59,000 show why amount alone can mislead.
- Date. Use a window rather than an exact date, because a cheque deposited on the 28th may be credited on the 30th. A window of a few working days is typical.
- Name or narration text. The weakest key, because banks shorten names and narrations differ by bank.
Matching in this order means you clear the easy, certain pairs first and spend your time on the doubtful ones.
What is a partial match, and what difference is acceptable?
A partial match is a pair that clearly belongs together but where the amounts are not equal. Common Indian reasons:
- The customer deducted TDS: a ₹1,00,000 professional fee paid as ₹90,000, with ₹10,000 shown as TDS.
- A bank charge was deducted: ₹50,000 sent, ₹49,975 received.
- A short payment for a dispute or discount.
There is no rule that says what difference is acceptable. Each team sets a written tolerance, often a few rupees for rounding, and treats anything larger as an item to explain. The danger is a tolerance that is too wide, because it can quietly absorb real errors.
Where does this lead next?
A many-to-one match works the same way in reverse. If a shop’s card machine records 42 sales on 14 October totalling ₹86,400 and the bank credits ₹85,000 the next day, the 42 lines match the one credit, and the ₹1,400 gap should equal the card fees and the GST on them shown in the acquiring bank’s settlement report.
Once you match by hand for a few months, you will notice that the same rules repeat: reference first, then amount within a tolerance, then a date window. Writing those rules down is how teams move from ad hoc matching to a consistent method.
What to read next
- Next in this series: Tolerance matching in reconciliation
- What is a reconciliation engine?
- For tools that support this work, see reconciliation software for Indian finance teams
Content authored by practitioners with experience at Amazon India, Intuit QuickBooks, and the Tata Group. Meet the team →