Bank Statement Import
Import bank statements to automatically match deposits to donors and create contribution records. Supports CSV, OFX, QFX and QBO formats, with a preview before anything is imported.
Bank Statement Import lets nonprofits upload their bank statements and match transactions to known donors, pledges and payouts, so a month of deposits becomes contribution records instead of a spreadsheet afternoon.
Where it lives
Muin for Finance → Reconciliation → All Transactions → Import.
Overview
| Step | What happens |
|---|---|
| 1. Choose a file | Drag and drop, or click Choose file |
| 2. Preview | We parse the file in memory and show what we read: nothing is imported yet |
| 3. Map columns | Only when we don’t recognize the bank: point us at the date, amount and description columns |
| 4. Import | You confirm, and the rows land in the register |
| 5. Match | Rows are matched to donor intents, pledges and payouts; confident matches are confirmed, the rest wait in the review queue |
Supported formats
| Format | Extensions | Notes |
|---|---|---|
| Bank CSV | .csv | Chase, Bank of America, Wells Fargo and U.S. Bank exports are read with no setup |
| Processor CSV | .csv | Venmo activity exports, including the payer’s note |
| Any other CSV | .csv | Works after you pick the date, amount and description columns once |
| OFX / QFX / QBO | .ofx, .qfx, .qbo | The “Quicken” or “QuickBooks” download most US banks offer |
| QuickBooks Online | via connector | Pulls transactions directly, no file needed |
File limits: 20 MB and 10,000 rows per file. Split a longer history into monthly files. PDF statements are not supported: export CSV or OFX from your bank instead.
Step-by-step
1. Choose your statement
The card is titled Upload bank statement. Drop the file on the area that reads “Drop CSV / OFX / QFX here, or click to choose a file”. The line underneath says “We’ll show you what we read before anything is imported.” If you would rather start from a file in the shape we read, use Download a sample CSV under “What files can I upload?”.
2. Check the preview
Nothing is imported yet. The preview tells you:
- which bank we recognized, or that we need your help with the columns
- which columns we will read for date, amount and description
- the first rows exactly as we parsed them
- how many transactions we found, their date range, and total money in and out
- any rows we could not read, with the raw values that broke them
If any of that is wrong, click Change columns or Cancel. The register is untouched either way.
3. Map columns (only if we ask)
When a file matches none of the formats above, we ask which columns hold the date, amount and description, and how the dates are written. Pick them from your own file’s headers and we preview again with your choices.
4. Import
Click Import N transactions. Parsing runs in the background; the card reports what happened: rows imported, auto-matched, needing review, and any that could not be read.
5. Matching
Every imported row is scored against open donor intents, pledges and payouts by:
- memo code from a giving page
- check number
- phone or email in the transaction description
- account last-4, amount and date proximity
- amount and name similarity
The register groups the scores into four bands: Auto (90% and above), Likely (70-89%), Uncertain (under 70%) and No match. Confident matches are confirmed and the rest wait in the review queue.
The point at which Muin confirms without asking you is yours to set, per surface, under Reconciliation → Settings → Auto-confirm thresholds (anywhere from 0.50 to 1.00). The bands above are what you see on a row; your threshold is what decides whether you are asked.
Matching is rule-based and deterministic, the same statement always produces the same result. AI is used elsewhere in Muin (categorization and free-text memo interpretation), and the Enable Smart Match setting adds AI-augmented matching if you want it; it is not on by default and it is not what decides who a deposit came from.
Re-uploading the same file
Safe. A re-uploaded statement is recognized by its contents and reuses the same import record, and individual rows already in the register are recognized by fingerprint and skipped rather than duplicated.
Troubleshooting
| Issue | Solution |
|---|---|
| “We couldn’t read this bank’s columns” | Your bank’s export isn’t one we recognize yet. Pick the date, amount and description columns when prompted: we remember them for that import. |
| Preview shows 0 transactions | The column mapping is probably pointing at the wrong columns, or the file is a summary rather than a transaction list. Check the preview’s column line. |
| Some rows “couldn’t be read” | The preview lists them with the raw values. The usual cause is a date written differently from the format you chose. |
| “File too large” | 20 MB is the limit: export a shorter date range. |
| “CSV row limit exceeded” | 10,000 rows per file. Split by month. |
| Low match rate | Matching leans on memo codes, check numbers, phone and email. Donor records with current contact details match far better. |
| The badge reads “Needs your input” and nothing imported | The background parse could not read the columns. Re-drop the file from the Import tab and answer Tell us which columns to read. |
| A handful of rows failed after the import | Open the file in the import list and use Retry parse on the failed rows. Rows that could not be read are never imported, so nothing silent happened. |
| An older import has vanished from the list | The import history covers a rolling 30 days. |
FAQ
Does anything get imported before I confirm? No. The preview parses the file in memory and writes nothing.
Can I import the same file twice? Yes. Rows already imported are recognized and skipped.
Are withdrawals imported too? Yes: the register holds both money in and money out, so the statement reconciles as a whole.
Can I fix a wrong column choice after importing? Upload the file again and choose different columns. The rows already imported are recognized, so you will not get duplicates.
Related
- Donation Reconciliation: working the register the imported rows land in
- DAF, Custodian Statements and Unidentified Gifts: what to do with a deposit you cannot trace
- Payments Reconciliation: matching processor charges rather than bank rows