Bank CSV merge and split

Four rows in should mean four rows out

If two monthly downloads overlap, joining them is the easy part. Knowing what happened to each row matters more. Add each CSV separately, label its account and currency, then check the column mapping, date basis, signs and number format before confirming it.

The fictional example has 100.00 and -20.00 in Source A, then another -20.00 and -30.00 in Source B. The two -20.00 rows have the same January 2 date and description. Both stay, so the merged file has four rows totaling 30.00 USD, not three rows totaling 50.00.

Switch to a month split and January gets three rows totaling 60.00; February gets one row totaling -30.00. Their counts and totals still add back to the input. The manifest shows that comparison for each account and currency, along with the source settings and output filenames.

Pick the file boundaries you need

Merge makes one output for each account and currency. Account split uses that same boundary, which is useful when one file has a mapped account column. Month splits use calendar months. Custom ranges include both dates you enter, and any transaction outside those ranges goes into a remainder file instead of disappearing.

Record-count splits count CSV records, not lines on a screen. A description with a quoted line break is still one record. Each new file has its own header. You can leave rows in source order or sort by date; rows sharing a date keep their original order.

Every output includes normalized date and signed amount columns, source ID, source alias, original physical line, and the original cells as JSON text. Unmapped details, including balances, remain in those original cells. Keep the manifest with the files so you can recover the original headers and see which settings were used.

A combined CSV is not a new bank statement

File reading, format checks and the calculation have separate statuses. A successful count and total check means the accepted rows carried through this operation. It does not establish a continuous verified balance, complete history, or a newly issued statement.

Matching rows may be two real payments or two copies of one payment. This tool warns about identical date, amount and description combinations within an account and currency, but never deletes them. Use the duplicate finder to review possible overlap, and read the worked overlap example before removing anything.

These are generic review CSVs, not a promise that your bank or accounting app will accept them. There is no fraud or authenticity judgment, reconciliation of your books, or tax, legal, lending or accounting advice.

Resolve unclear rows before downloading

Use UTF-8 CSV or text files, up to 5 MB each, with no more than twelve sources and 20,000 records plus headers in a batch. The source tables show previews and then pages of at most 200 rows; output previews show 25 at a time. Downloads include all accepted rows, not just the page you can see.

Each source needs an account and currency label. If you map those columns, their cell values take precedence and blank cells block export. Labels are case-sensitive, so Checking and checking are different groups. Different currencies are never added together, and the tool does not convert exchange rates.

Choose posted date or transaction date consistently across the batch. Ambiguous dates and number formats need an explicit choice. Amounts use exact decimal arithmetic at your selected precision, with positive meaning money in and negative meaning money out. Card exports often need a different sign mapping, so compare the preview with a charge and a payment you recognize.

Blank records, repeated headers, subtotal-like rows and invalid amounts or dates block all transaction downloads. They remain visible with their source lines. If your file has introductory records, select its actual header record: the records before it are explicitly excluded from transaction outputs and preserved in the manifest. Trailing times or weekday labels removed from dates are reported for your approval, without timezone conversion.

Custom ranges cannot overlap, including on a shared endpoint. You can define up to 100 ranges, and days not covered by a range go to the remainder. Missing dates and quiet months are not evidence that a file is complete.

What happens to your files

Your file and the values you enter are processed in this browser tab. They are not uploaded or sent to analytics. We do not save them between visits. Downloads you choose to save remain on your device.

Clear all removes the current inputs from the tool. Leaving the page is not a clearing action: Back navigation can restore uncleared input. Clearing the tool does not delete your downloads or securely erase device memory. The site still makes ordinary requests to load the page, but the tool does not send your file data with them.

Downloads are new files with neutral numbered names; your originals are untouched. Spreadsheet-safe exports prefix formula-like text with an apostrophe and show how many cells change. Parsed negative amounts remain numbers. A separate machine CSV option leaves text unescaped, so do not open that version directly in a spreadsheet if its text might contain formulas. Spreadsheet software can remove protection when saving or reopening a file.

The original-cells column retains unmapped details, which may be sensitive. This is an organizing tool, not a redactor. Inspect every file before sending it to someone else.

If the missing months are PDFs

Already have a bank CSV? This tool is free. Only have PDF statements? First check your bank's native export or any included extraction option. AnyStatement can convert PDFs into reviewable transaction data. For a folder of monthly statements, this guide to combining monthly PDFs explains the checks to make before joining the extracted rows.

The file-assembly approach also has ordinary spreadsheet alternatives. PDFTables' consolidation walkthrough describes manual and macro-based assembly; financial-csv-tool documents local command-line date filtering. Neither alternative makes repeated transactions safe to delete automatically.