What format should the CSV be in?

A header row with columns such as date, amount, reference, and description, then one transaction per row. Columns can be in any order and extra columns are ignored. A file with no header is read as date, amount, ref, description.

Is my data uploaded anywhere?

No. The file is read and matched in your browser. Nothing is sent to a server.

How are near-date transactions matched?

After exact-reference matches, remaining rows with the same amount are paired when their dates fall within the tolerance you set, choosing the closest date. Raise the tolerance for statements that clear a few days apart.

Why is a transaction shown as unmatched?

Either no row on the other side shares its reference, or no row has the same amount within the date tolerance. A different amount, a typo in the reference, or a date gap wider than the tolerance will all leave a row unmatched.

All tools

Free tool

Bank vs ledger reconciliation

Paste or upload two CSV lists, one from your bank and one from your ledger, and see what matches and what does not. Everything runs in your browser. No file leaves your device.

Your two lists

Bank statementUpload a .csv fileor paste rows below

LedgerUpload a .csv fileor paste rows below

Date tolerance (days)

How many days apart two dates may be and still match on amount.

Result

Add both lists and select Reconcile to see the match.

How matching works

Reference first, then amount and date

Two passes. First, rows that share an exact, non-empty reference are paired. Then, of what remains, rows with the same amount and a date within your tolerance are paired, closest date first. Each row is used at most once. Anything left is reported as bank-only or ledger-only.

Tolerant CSV reading

Common column names are recognised in any order: date, amount, reference, description. Currency symbols, thousands separators, and parentheses-for-negative are handled. A file with no header falls back to the column order date, amount, ref, description.

Nothing leaves your browser

Parsing and matching happen on your device. No statement or ledger is uploaded to a server, so you can reconcile real data without sending it anywhere.

A first pass, not a sign-off

This lines up rows that agree. It does not judge whether an entry was booked to the right account or explain a genuine difference. Treat the output as a starting point for a manual review.

Sources

Bank reconciliation is a standard bookkeeping process rather than regulated data. For a neutral reference on the process itself:

Bank reconciliation guidance: AICPA & CIMA

This is a process tool that matches rows you provide. It is an example first pass, not an audited reconciliation or accounting advice. Review every match and every exception against your source records.

Reconcile continuously, not once a month

Rivane matches bank feeds against the ledger as transactions arrive, so month-end is a review, not a hunt for differences.

Request demo