If your bank isn't supported by automatic sync, or you'd rather not connect it, import the files your bank offers. Designed for repeat use — this month's statement, then next month's — with overlapping rows matched rather than duplicated.
Importing
- Open Transactions and click the account.
- Choose Import file.
- Upload the file.
- Review the preview.
- Confirm.
Import is per-account: you're always importing into the account you opened.
Supported formats
| Format | Notes |
|---|---|
| OFX | The standard bank export. Works without configuration |
| QFX | Quicken's OFX variant. Works without configuration |
| CSV | Needs a column mapping the first time |
Prefer OFX or QFX where your bank offers them — they're self-describing, so nothing can be mapped wrongly. Look for "Download transactions" and pick Quicken, Money, or OFX format.
Mapping a CSV
CSV has no standard, so you tell Earmarkr what the columns mean:
- Date — the column, plus its format. You must state whether it's MM/DD/YYYY or DD/MM/YYYY. Earmarkr never guesses, because guessing wrong silently corrupts dates.
- Payee — who you paid. Optionally a separate memo column.
- Amount — one of three shapes:
- a single signed column (negative = money out)
- separate debit and credit columns
- an amount plus a type indicator (
DEBIT/CREDIT)
Earmarkr proposes a mapping from the headers; correct it if needed.
Save it as a preset and next month's import from the same bank is prefilled. Presets can be tied to one account or shared across all of them.
The preview
Before anything is written you'll see:
- Rows found, and the date range covered
- N new · N matched · N duplicate
- A sample of parsed rows so you can confirm dates and amounts read correctly
Check the sample rows. A wrong date format is obvious here and invisible later.
How duplicates are avoided
Every imported row gets a fingerprint. Import the same file twice and the second run inserts nothing.
For OFX and QFX the fingerprint comes from the bank's own transaction id, so it's stable across statements — overlapping periods are matched exactly.
For CSV, which usually has no id, the fingerprint is built from the account, date, amount, payee, and description. Overlaps are also matched against transactions you already have, including ones you entered by hand.
Two genuinely identical transactions on the same day — two $5.00 coffees — both import correctly.
One important limitation
Use file import or automatic sync for a given account, not both.
A transaction that arrived through sync carries a provider id and is matched only by that id, so file import can't recognise it and will insert a duplicate.
Pick one method per account. Mixing is the main way to end up with a mess.
After importing
- Uncategorized rows appear under Needs review. Existing rules are applied automatically.
- Card reserves and reports refresh.
- Reconcile the account to confirm the balance matches your statement.
If something looks wrong
Dates are wrong — the date format was mapped incorrectly. Fix the mapping and re-import; the originals are matched, not duplicated.
Amounts have the wrong sign — the amount convention is wrong. Check whether your bank uses a single signed column or separate debit/credit columns.
Nothing imported — every row was recognised as already present. That's usually correct. Check the preview counts to confirm.