Transaction filter and subtotal

Start with the question you want the rows to answer

Choose a CSV or paste its contents, then check the column mappings. Pick the date you actually mean: a card purchase date and its posted date can fall in different months. You also choose the number format and whether positive amounts mean money coming in or going out. If your file has no account or currency column, supply a label that applies to every row.

In the fictional example, the June transactions are TRAIN -40.00, HOTEL -200.00 and TRAIN REFUND +10.00 USD. There is another TRAIN -30.00 row on July 1. Set the dates to June 1 through June 30 and search for descriptions containing TRAIN: two rows remain, with 40.00 outgoing, 10.00 incoming and a net of -30.00. Add an absolute minimum of 20.00 and only the -40.00 fare remains.

Every filter must match, but you can add alternative text phrases. TRAIN or HOTEL will match either phrase within your chosen date and amount bounds. Search is literal: a star is a star, not a wildcard. Case matching is your choice; spaces and newlines stay part of the description.

Keep the rows and the reason you selected them

The count and subtotals cover the entire selection, even when the table shows only its first page. Incoming is the sum of positive amounts, outgoing is the size of the negative amounts, and net is incoming minus outgoing. Each account and currency has its own subtotal. No exchange rates are applied.

The row download retains all original columns, including ones you did not map. A separate assumptions and summary CSV records the date range, settings, totals and the selection decision for every original source line. That makes it possible to retrace the calculation, but the report also contains details about rows you filtered out. Read it before sending it to someone else.

Source order is the default. You can sort the selection by date instead, keeping same-day rows in their original order. Sorting never changes the physical source-line references in the report, and a quoted description that spans several lines remains one CSV record.

A match is a search result, not a verdict

Finding a charge does not establish that it is disputed, deductible or reimbursable. A TRAIN description might be the fare you want, a refund, or another payment with similar wording. Check the source statement and any receipts before drawing a conclusion.

The tool does not classify transactions, remove pending or posted duplicates, connect to an account, or establish complete history. It does not judge fraud or document authenticity, reconcile balances, or guarantee that another app or recipient will accept your download. Nothing here is tax, legal, lending or accounting advice.

If you are assembling evidence, keep the original statement and your working copy separate. The statement review log gives you somewhere to record what you checked; the tax-records guide explains why a filtered bank list does not replace supporting records.

Fix uncertain rows before trusting a total

This tool accepts one UTF-8 CSV or text file up to 5 MB and 10,000 data rows. Tables begin with a short preview and page through at most 200 rows at a time. A larger file must be split first; it is not silently truncated. The tool needs one signed amount column, so use the debit/credit converter first if your export separates money in and money out.

Invalid dates, blank amounts, repeated headers, subtotal-like rows and other unresolved row issues block the calculation, even if your filters would hide them. Replacement characters block exports too, including in unused columns or headers. The issues list points back to the original records and has its own download. Correct a working copy and leave your original alone. A valid file with no matching rows, on the other hand, is simply a zero-row result.

Date and amount bounds include both endpoints. Amount comparisons use the sign interpretation you selected, while the original-value row export keeps the bank's signs. Decimal arithmetic is exact at your chosen precision: extra decimal places are rejected rather than rounded. Two-digit years use the displayed cutoff, initially 69, and time or weekday removal is reported rather than hidden.

Account and currency labels are exact, including their case and spaces. USD and usd therefore remain separate labels until you deliberately correct your source. For a file-format check before filtering, see how to check a bank statement CSV.

What happens to your file

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.

Use Clear to remove the tool's active input and settings. Navigating away is not a clearing action: Back can restore uncleared input. Your browser manages its own memory and history, so Clear is not a secure wipe. Downloads create new files and never overwrite your original statement.

The spreadsheet-safe download prefixes formula-like text cells with an apostrophe and tells you how many cells need that change. Parsed negative amounts are left alone. A separately labeled original-value CSV preserves the cell text without that protection and should not be opened in a spreadsheet. CSV quoting and line endings may change in either copy; these are selected records, not byte-for-byte copies of the source file.

A spreadsheet job, without setting up a spreadsheet

The method is ordinary conditional filtering and addition, not a proprietary scoring system. We opened the worked conditional-sum question, the financial-csv-tool documentation and Store Map's filter and export description on October 17, 2026. They show other ways to do this job. This page uses the explicit rules above and does not claim compatibility with those tools.