Supported File Formats

What works

Format	Extensions	Notes
Comma or tab separated text	`.csv` `.tsv` `.txt`	The most common bank export. Semicolon- and pipe-separated files work too — the separator is detected automatically.
Excel	`.xlsx`	Read directly. No spreadsheet program needed.
Open Financial Exchange	`.ofx` `.qfx`	Cleanest option. The fields are already labelled, and it often carries your closing balance.
Quicken Interchange	`.qif`	Older format, also self-labelling.
JSON	`.json`	A list of transaction records. Useful for exports from other apps.

What to pick

If your bank offers OFX or QFX, take it. These formats state exactly which field is the date, the payee, and the amount, so the column-matching step arrives already filled in and is a single click. They frequently include the account’s closing balance, which fills in your Balance Sheet automatically. See Balances From Your Statement.

Otherwise take CSV. Well supported and easy to check by eye.

Excel is fine if that is all you are given. Aventurine reads .xlsx files directly, including ones with a title or metadata above the real table — it skips down to the first proper header row. Dates stored as Excel date values are understood.

What does not work

PDF statements. A PDF is a picture of a table, not a table. Aventurine rejects them with a message suggesting you download a CSV, OFX, or Excel export instead. If your bank truly offers nothing but PDF, look under “Transactions” or “Activity” rather than “Statements” — the export option is usually there, or contact them.

Old-style Excel files (.xls). The legacy binary format is not supported. Open it in a spreadsheet program and re-save as .xlsx or CSV.

Anything else. Other binary files are rejected with a message listing what is supported.

The extension does not have to be right

Aventurine identifies the format by looking inside the file. An OFX file your bank saved as .txt, or a CSV named .dat, still imports correctly.

Details that are handled for you

You do not need to prepare or clean the file. All of this is dealt with automatically:

Separators. Comma, tab, semicolon, or pipe, detected by looking at how consistently each one splits the lines. Quoted fields containing the separator are handled correctly, so "Acme, Inc." stays one field.

Quotes and line breaks inside fields. Standard quoting rules, including doubled quotes and multi-line quoted fields.

Text encoding. Modern files are read as-is; older “ANSI” exports are decoded correctly so accented characters and currency symbols survive.

Date formats. All of these are understood:

2026-03-15      15.03.2026     Mar 15, 2026
03/15/2026      20260315       15 Mar 2026
03/15/26        3-15-2026      15-Mar-2026

Excel date values (a plain number in a spreadsheet cell) are converted too.

Amount formats. Currency symbols, thousands separators, and accountant-style negatives are all handled:

$1,234.56    →  1234.56
1.234,56     →  1234.56    (European style)
(89.10)      →  -89.10     (parentheses mean negative)
-$45.00      →  -45.00

The European convention — dots for thousands, comma for the decimal — is detected across the whole file, so an ambiguous value like 1.234 is read the same way as the rows around it that make the convention clear.

Direction. Either from the sign on a single amount column, or from which of a Debit/Credit pair holds the value. See Matching Your Columns.

Blank rows and stray headers. Skipped.

What Aventurine needs from a file

Only three things, in some form:

A date it can read.
A description that is not empty.
An amount it can read.

A row missing any of these is skipped, with the reason shown on the review screen. Everything else — notes, balance, categories, reference numbers — is optional.