CSV column mapping

Project Budget recognizes a handful of common CSV shapes automatically — Chase, Capital One, Discover, Mint, Actual Budget, and a generic fallback. When yours doesn't match any, the import preview shows a Map columns panel and asks you to point at each required field.

The required fields

Three fields are mandatory:

• Date — when the transaction posted. The mapper accepts any column whose values parse as ISO-8601, US MM/DD/YYYY, or European DD/MM/YYYY. If the column is ambiguous (numbers like 01/02/2026 could be Jan 2 or Feb 1), the picker asks which format your bank uses and remembers the choice for next time.
• Payee / description — the merchant or counterparty. Often the rawest field in a bank export; you'll clean it up later via bulk edit. See Filtering and bulk edits.
• Amount — either one signed column (-42.50 for outflows, 100.00 for inflows) or two columns (one for debits, one for credits). The picker auto-detects which shape your file uses.

Two fields are optional:

• Memo / notes — separate from payee. Useful for the long-form description some banks include.
• Category — the bank's category guess. Project Budget can drop it (most users do) or import it into a Bank category field that's visible in the register but separate from your envelope categories.

When the date column is ambiguous

A column with values like 03/04/2026 is ambiguous. The picker offers three resolutions:

1. Treat as US (MM/DD). March 4.
2. Treat as European (DD/MM). April 3.
3. Read another column. If your CSV also has an ISO-format date column, point at that one instead and ignore the ambiguous one.

The choice is per-bank, stored under the saved mapping (see below).

When amount is two columns

Some banks (Capital One is the textbook case) split amounts into Debit and Credit columns: outflow rows have a value in Debit and blank in Credit; inflow rows are the reverse. Pick both fields in the mapper; Project Budget combines them into a single signed amount internally.

Saving a mapping

Once you've mapped the columns and run a successful import, the preview footer offers Save as preset. Name it after your bank (Ally Checking, Schwab Brokerage) and it appears in the format dropdown next time. Saved presets are per-profile.

Reusing a saved mapping

On future imports, when you drop the same shape of CSV, the auto-detector matches your saved preset first. The preview opens with columns already mapped; just confirm and import.

What the mapper does not do

• Will not transform values. If your bank exports amounts as (42.50) for negatives (accountant style with parentheses), the importer doesn't parse those. Open the file in a spreadsheet first and convert to -42.50.
• Will not merge columns. If payee and memo are in separate columns and you want them concatenated, do that in a spreadsheet first or accept them as separate fields.
• Will not parse multi-line records. Each transaction must be one CSV row.
• Will not handle macro-flavored CSVs. Some Quicken exports include header rows that span multiple lines; strip those manually.

A common gotcha: BOMs and quoting

Files saved from Windows tools sometimes start with a UTF-8 byte-order-mark (the invisible  character). The importer strips it on read, but if you opened the file in a text editor and the first column name looks like Date, that's the BOM.

Quoted fields with internal commas (typical of memo fields containing "Last, first" names) parse correctly as long as the CSV is RFC-4180-compliant. Files that quote inconsistently — some rows quoted, some not — sometimes parse with extra columns; check the preview and reject the import if the column count looks off.