Free CSV & Excel Transaction Reconciliation

Browser-local reconciliation

Use Local Reconciliation Suite

Files are read and compared inside this tab. Transactions are not uploaded, and every report is written here on your device.

1. FilesAdd CSV or XLSX exports (two or more)

2. What counts as a match

Start fromBase currency for FX (optional)

3. Reconcile

Saved setups

Profile nameLoad a saved profile

Import a profile

Saved profiles hold column mappings and rules only - never transactions - and stay in this browser on this device.

Add two or more exports. Every file is read inside this tab and is not uploaded.

Review

Matches marked as needing review are not confirmed. Reconciling files is not an accounting approval.

Nothing reconciled yet.

A card settlement file against a sales ledger: with a zero-day window, settlements posted two days after the invoice stay unmatched until the date window is widened. Where the processor reference does agree with the invoice number, the pair is made but the date gap is stated and the pair stays in Suggested until someone approves it. Repeated invoice numbers on one side are listed as duplicates rather than paired, because choosing either copy would not be defensible. A payout that is short by the processor's percentage only groups against its orders once a fee allowance is set, and the resulting group is labelled as fee-explained rather than as an exact match.

Explore developer and data tools

Find related browser-local tools for nearby tasks without starting another search.

Browse all developer and data tools

About Local Reconciliation Suite

Reconciliation is the routine of proving that two records of the same money agree. A bank statement against a general ledger, a payment processor payout against the orders it settles, an intercompany account against its mirror, last month's file against this month's - in each case you are asking which lines correspond, which are missing from one side, and where the totals diverge. Done by eye in a spreadsheet it is slow and easy to get wrong. Done by a tool that uploads your ledger it means handing customer names, amounts and bank references to somebody else's server, which is exactly the data most finance teams are least free to send anywhere. This suite does the whole job inside your browser tab. You bring in two or more exports, say which columns hold the reference, the date and the amount, then describe what counts as a match: which fields must agree, how far apart the dates may be, how much difference is tolerable, whether one side carries the opposite sign, and whether several rows on one side may answer a single row on the other. It runs off the main thread so the tab stays responsive, then hands you the exceptions - because the exceptions, not the matches, are the part a person is actually paid to look at.

Work through the input

Add two or more CSV or XLSX exports. Workbooks with several sheets let you pick the sheet, and the header row is adjustable for files that carry a title block or a few blank lines before the real column names.
Map the columns for each file separately, because two systems rarely name them the same way. Amount is required; reference, date, currency, description and account are optional and are used for matching, context and reporting. Date and number conventions are set per file, so a European export and an American one can be compared without editing either.
Choose a starting rule set - bank against ledger, processor payout, invoice against payment, card statement against expenses, marketplace payout, intercompany, or period against period - then adjust it. Every rule states in plain language what it will do before you run it, so a rule nobody can read is never a rule you have to trust.
Set the parts that matter for your data: an amount tolerance in currency or percent, a date window in days, opposite signs where a debit in one system is a credit in the other, a fee allowance where a payout arrives net of charges, and a group size where several rows legitimately answer one. Rules run in the order you give them, so the strictest can claim its rows before a looser one is allowed to look.
Reconcile, then work through the tabs: Summary, Exact matches, Suggested pairs, Needs review, Unmatched, Duplicates, Invalid rows and Decisions. Approve or reject a suggested pair, pair two rows by hand with a note, and undo the last decision - every one of those is recorded.
Export what you need: individual CSVs, the rules and column mapping as reusable JSON, draft journal templates for the differences, or a single ZIP audit package containing the lot alongside a plain-English report of what was compared and under which rules.

When local reconciliation suite is useful

Bank statement against a general ledger

Match on the unique reference first with a five-day window, then let a looser amount-and-date rule pick up what the reference missed. Where the bank posts debits negative and the ledger posts them positive, switch that rule to opposite signs rather than editing either file.

Processor payout against the orders it settles

One payout line rarely equals one order. Use a one-to-many rule with a group size large enough for a day's orders and a percentage fee allowance, and the group comes back labelled as fee-explained with the exact shortfall stated, so the fee can be journalled rather than argued about.

Invoices against part payments

Match on the invoice number with dates switched off, and allow several payments to answer one invoice. What is left unmatched is the genuinely outstanding balance rather than a list of rows the tool could not shape.

Check the output

Only one thing is called an exact match: a one-to-one pair, driven by a reference, with identical amounts, on the same day, in one currency, with the same sign. If the tool had to interpret anything - flip a sign, spend a tolerance, widen a date, convert a currency or assemble a group - the pair is a suggestion and waits for a person. That distinction is the difference between a report you can defend and one that merely looks finished.
Where two candidates are equally good the pair is not made. The row is listed as ambiguous with the competing rows named, because a tool that guesses in a reconciliation is worse than one that asks. The same applies to repeated references inside one file: both copies are held as duplicates rather than matched, since choosing either would be arbitrary.
Currencies are never added together. Totals are reported per currency, and a row with no currency is counted in its own group rather than being folded into another. Exchange rates are used only when you type them in; nothing is looked up, so no rate is ever assumed on your behalf.
Grouped matching is deliberately bounded. Several rows on one side can answer a single row on the other, up to a group size you set, but the search does not roam across both sides at once. Unbounded subset matching would eventually find some arithmetic that works, and 'reconciled' would stop meaning anything.
Dates and amounts are never coerced. A value that does not fit the convention you selected is marked invalid with a reason instead of being reinterpreted, and two-digit years are rejected because the century would be a guess. Amounts are compared as whole minor units, so no floating-point drift reaches a total. Leading zeros are preserved, so 0070 never matches 70.
The work is limited by what a browser tab can hold. Very large exports may exceed the CSV or workbook limits, in which case the tool says so plainly rather than failing silently; narrow the date range or split the file and run it again.
This produces a comparison, not an approval. It is not an audit opinion, an accounting sign-off or a filing, and the journal outputs are draft templates for the differences - shaped like the layouts common accounting imports expect, but not posted, not sent, and not connected to any accounting product. Check the columns against your own chart of accounts before importing anything.

Questions about Local Reconciliation Suite

Is reconciliation and report download free?

Yes. Import supported CSV or XLSX exports, configure matching rules, review exceptions, and download the reports without a subscription or account. Processing runs locally; large files remain subject to browser memory and import limits.

Does this connect to my bank or accounting software?

No. Export your transactions first, then import the files here. The suite does not fetch bank feeds, post journal entries, provide shared team approvals, or maintain a centralized audit record. It is a local comparison and review tool, not a full accounting system.

What does the suite actually decide for each row?

Every row ends up in exactly one state: part of an exact match, part of a suggested pair, approved by you, ambiguous, a held duplicate, invalid, or unmatched. Each carries the rule that placed it there and a reason in plain English, and the source file and original line number travel with it so any decision can be traced back to the export it came from.

Why are some rows left ambiguous instead of matched?

Because more than one row could reasonably pair with them, and the candidates were indistinguishable on the evidence the rule looks at. Picking one would produce a report that looks complete but is not defensible, so the row is listed with its competing candidates for you to resolve.

Can I match one row against several?

Yes. A rule can allow one row on the left to answer several on the right, or several on the left to answer one on the right, up to a group size you set. Groups are never called exact matches, however clean the arithmetic, because assembling a group is an interpretation.

Can I reuse the setup next month?

Yes. Column mappings and rules can be saved in this browser, or exported as JSON files you keep with your working papers. Profiles hold layout and rules only - never transactions - and a mapping is reapplied by matching the column headers, so next month's export does not have to arrive in the same order.

What is in the audit package?

A plain-English report naming the source files, the rules applied and the outcome; a complete row-by-row CSV; separate files for exact matches, items needing review, unmatched rows per dataset, invalid rows and every manual decision; the rules and mapping as JSON; and draft journal templates for the differences. Cells that came from your files are written so a spreadsheet cannot execute them as formulas.

Do the files leave my device?

No. Every file is read, compared and turned into reports inside this browser tab, and there is no endpoint on this site that accepts a file. Transaction values, file names and column mappings are never transmitted. The page itself is served over the network like any web page, but your data is not part of that.

Local Reconciliation Suite