Importing Bank Statements

Import transactions directly from your bank's exported files instead of entering them manually. The Budget app supports CSV, Excel and LibreOffice spreadsheets, OFX, QIF and ISO 20022 camt.053 XML formats with automatic duplicate detection. An [import folder](#import-folder) imports the statements you drop into it without you opening Budget.

Importing a bank statement

Supported Formats#

FormatBest ForNotes
CSVMost banks, custom exportsMost flexible; requires column mapping
XLS / XLSX / ODSBanks that only export a spreadsheetRead like a CSV file; see Spreadsheet files
OFXUS/Canadian banks, direct downloadsFields are detected automatically; the text fields can be adjusted
QIFQuicken exports, older softwareLegacy format with basic field support
camt.053 XMLEuropean banks (ISO 20022 statement export)Accounts, balances and references detected automatically; see camt.053 Import

Tip: If your bank offers multiple export formats, OFX is usually the easiest since its fields are detected for you. Use CSV when you need full control over how fields are interpreted.

Moving from Firefly III, YNAB, Actual Budget, Mint, Monarch Money or Toshl Finance? Their exports import with a preset that needs no column mapping. See Switching from Another App.

CSV Import Step-by-Step#

CSV is the most flexible import format. The import process walks you through uploading, mapping, previewing, and executing the import.

1. Upload Your File#

Navigate to Import > Upload File and select your CSV file. If the file has no header row, untick Skip first row (headers) on the mapping screen; the preview redraws straight away with the first row as data and the columns named by position.

A field that runs over more than one line, such as a description or note the bank has written inside quotes with line breaks in it, stays part of its own row.

Statements are imported into your own accounts only. An import runs as you rather than as an account's owner, so accounts someone has shared with you are not offered as a destination, here or for every format below.

Note: The app automatically detects and converts common non-UTF-8 encodings (ISO-8859-1, Windows-1252, and ISO-8859-15) to UTF-8. In most cases no manual conversion is needed. If special characters still appear incorrectly, re-save your file as UTF-8 from your spreadsheet application.

2. Delimiter and Character Encoding#

The app automatically detects whether your file uses commas, semicolons, or tabs as delimiters. The detected delimiter is shown in the preview. If the detection is wrong, pick the right one and the columns and preview redraw immediately, so you can check the split before mapping anything.

Tip: European bank exports commonly use semicolons as delimiters since commas are used as decimal separators in those regions.

Character encoding#

Next to the delimiter is a Character Encoding picker, set to Detect automatically by default. It names the encoding it settled on in the hint beneath it.

Detection is reliable for Unicode (UTF-8) files, and for OFX and camt.053 files, which state their encoding in the file itself. It cannot be reliable for everything else: older single-byte encodings accept every possible byte, so a Cyrillic file and a Western European one are genuinely indistinguishable from their contents alone. When the app has to guess, it assumes Western European.

You will know the guess was wrong because the preview below shows mojibake — Äàòà where you expected Дата, or Café where you expected Café. Pick the right encoding from the list and the columns and preview redraw immediately. The options cover Unicode, Western and Central European, Cyrillic, Greek, Turkish, Hebrew, Japanese, Chinese and Korean.

The picker applies to every format, not just CSV, and your choice is used for the actual import as well as the preview — so what you see is what you get.

Tip: If only some characters look wrong — curly quotes or a euro sign showing as gibberish while the rest reads fine — the file is most likely Western European and no change is needed. Whole words of nonsense mean the encoding is genuinely wrong.

3. Column Mapping#

Map each column in your CSV to the corresponding transaction field:

FieldRequiredDescription
DateYesTransaction date
AmountYes (unless using dual columns)Transaction amount
Income AmountNoSeparate column for credits/deposits
Expense AmountNoSeparate column for debits/withdrawals
DescriptionNoTransaction description or memo
NotesNoColumn stored in the transaction's notes field
TypeNoWhether the row is income or an expense (see below)
VendorNoPayee or merchant name
ReferenceNoCheck number or reference ID
CategoryNoCategory name; categories are auto-created if they do not exist
AccountNoAccount name; accounts are auto-created with inferred types
CurrencyNoCurrency code for the transaction

Select the appropriate column header for each field. Columns you do not map are ignored.

Description, Notes, Vendor and Reference accept more than one column. Tick every column that belongs in the field and the values are joined with a comma, in the order the columns appear in the file; blank cells are skipped. A bank that exports a merchant column and a separate details column can have both in the description, for example Tesco, Weekly groceries. The same applies to those four fields for OFX, QIF and camt files, and a saved import template keeps the whole selection.

If your bank export carries extra transaction information that does not belong in the description, map that column to Notes and it is kept on the transaction rather than discarded. Notes appear as their own column in the import preview whenever something is mapped into them, so you can check the right column was picked before anything is written.

Very long values are truncated to what the field can hold: 1,000 characters for the description, 2,000 for notes, 255 for the vendor and 100 for the reference.

How a negative amount is written#

Exports disagree about how to write a negative, so Budget reads all the usual notations. Every one of these is an expense of 1,234.56:

NotationExample
Leading minus-1234.56
Trailing minus1234.56-
Brackets (the accounting convention)(1234.56)
A typographic minus or dash instead of a hyphen−1234.56
A DR (debit) marker after or before the number1234.56 DR or DR 1234.56

Currency symbols, spaces and thousands separators around the number are ignored, so -€1.234,56 and ($1,234.56) are read the same way. A CR marker is read as money in, as an amount with no sign would be. A lone - in an amount column means "nothing here" and is read as zero, not as a negative.

How decimals are read#

When an amount has both a . and a ,, whichever comes last is the decimal point. When it has only one of them, once, it is read as the decimal point unless it can only be a thousands separator, which means exactly three digits after it and one to three before it, not starting with a zero, as in 1.234 or 12,345. So 117.50000000, 0.00012345, 12.3456 and 0.123 all keep their decimals, which matters for crypto and investment exports, while 1.234 is still read as one thousand two hundred and thirty-four. The same applies to the app presets under Switching from Another App; OFX, QIF and camt.053 files have their amounts read in their own way and are not affected.

Type Column#

Budget needs to know which way each row goes. By default it reads the sign of the amount: negative is an expense, positive is income.

Plenty of exports do not sign their amounts — they write every value as a positive number and put the direction in a separate column instead. Nextcloud Tables does this, and so do several banks and budgeting apps. Map that column to Type and it decides the direction, overriding the sign.

Recognized values (case and surrounding punctuation are ignored):

Means incomeMeans expense
Income, Credit, CR, C, Deposit, Refund, InExpense, Debit, DR, D, Withdrawal, Payment, Purchase, Out

Rows where the Type column is blank or holds something not on this list fall back to the sign of the amount, and the preview tells you how many rows that applies to.

Tip: If your amounts are all positive and you do not map a Type column, every row is imported as income. The preview warns you when a batch is about to go in almost entirely one way while the account it is going into leans the other way — check that warning before confirming.

Category Column#

When you map a column to Category, the app automatically creates any categories that do not already exist. Imported transactions are assigned to the matching category by name. This is useful when your bank export or finance app includes category information.

A name is matched against a top-level category first. Failing that, a subcategory with that name is used when it is the only one, so a column saying "Groceries" files rows under your existing Food › Groceries rather than creating a second Groceries at the top level. When several subcategories share the name, a new top-level category is created rather than guessing between them.

The column can also hold the whole path, with > between the levels, which is how Skrooge writes its categories. Food > Groceries > Fruit files the row under Fruit in Food › Groceries, and any level that does not exist yet is created under the one above it. The first level is looked up in the same way as a single name, and a new subcategory takes the type of the category it sits under. Paths are written the same way as in a category file, so you can import your category tree first and then the transactions. A category whose own name is the whole path, such as one made by an earlier import, still gets the rows for that path, so delete it first if you want them to go into the tree instead.

A row looks for a category of its own type first, income for money in and expense for money out. When the only category with that name is of the other type, that one is used rather than creating a second, so a refund goes back into the expense category it names.

An import rule that sets a category comes first: the column only names the category for rows that no rule has categorized. The preview shows the category each row will end up with, whichever of the two it comes from, and counts both under Auto-categorized.

Account and Currency Columns#

Mapping an Account column lets you import transactions across multiple accounts from a single file. Accounts that do not already exist are auto-created with types inferred from the account name (e.g., names containing "Cash" default to the cash type, "Investment" to the investment type, and so on) and in your default currency, unless a mapped Currency column says otherwise. The Currency column sets the currency for each transaction and is especially useful alongside the Account column for multi-currency imports.

Two things to know about the Account column:

  • A row whose account cell is blank goes to the account chosen under Account for rows without one on the preview step (the same select that is otherwise the import's destination). It is not dropped, and it does not create an account. Leave that select empty and such rows are skipped. The preview tells you before you import, above the statistics: how many of the file's rows will not be imported, why, and which rows they are, along with a reminder that choosing an account in that select is what brings them back. Pick one and the warning clears and those rows rejoin the count.
  • Only a row that would import as a transaction can name an account. If the file's header line is read as data (see skip first row), its account cell holds the column title — "Account:" or similar — and that used to become a real account. It no longer does; the row simply fails with a clear reason.

A row that names an account which cannot be resolved is reported rather than guessed at, since filing it under the wrong account would be worse than skipping it.

Because one account is created for every distinct value in that column, mapping the wrong column here is expensive: map the date column and you get one account per date. The preview warns you before that happens. Above five new accounts it shows how many the import would create, above the statistics where you cannot miss it, and if most of the new names look like dates it says so and points you back at the column mapping. It is a warning, not a block -- a file that genuinely spans a dozen accounts imports normally.

If an import does go in with the wrong column mapped, the accounts it created can be removed together from the Accounts page rather than one at a time -- see Deleting several at once.

4. Dual-Column Amount Mapping#

Some banks, particularly European ones, export income and expenses in two separate columns rather than using positive and negative values in a single column.

If your file uses this format, map the Income Amount and Expense Amount columns individually instead of mapping a single Amount column.

Warning: You must map either a single Amount column or the Income Amount and Expense Amount pair. Mapping both at the same time is not allowed and will display a validation error.

5. European Number Format#

There is nothing to set for European number formatting, as the app works out which separator is the decimal one from each amount:

Amount in the fileRead asWhy
1.234,56 or 1,234.561234.56When both appear, the last one is the decimal separator
12,5 or 12.512.5A single separator is a decimal one
1.234 or 1,2341234A single separator followed by exactly three digits, with one to three digits before it, is read as grouping thousands
1.234.5671234567The same separator more than once groups thousands

The one amount that cannot be told apart is the third row: 1,234 could be one thousand two hundred and thirty-four, or one and a bit with three decimals. Bank statements use two decimals, so it is read as thousands. Check the preview if your file holds amounts with three decimals, such as fuel prices or exchange rates.

6. Preview#

Under the column mapping, Apply import rules for categorization decides whether your import rules run on this file. For each new file it starts ticked or unticked from the Auto-apply Import Rules setting in Settings. Untick it and the rules are left out of both the preview and the import, so the preview shows exactly what will be written. Choosing a saved template sets the box to what the template stored, and you can still change it before importing.

After mapping your columns, click Preview to see a table of parsed transactions before anything is written to the database. Review the preview carefully:

  • Verify dates are parsed correctly
  • Confirm amounts have the right sign — expenses are shown in red with a minus, income in green without one
  • Check that descriptions and vendors look right (a mapped payee has a column of its own)

Rows that match a transaction already in the account are flagged with a Duplicate badge. They are shown for review but skipped when you import — see Duplicate Detection. The Show duplicates and Show uncategorized checkboxes only filter what the preview displays; they do not change what is imported.

A row the import cannot take at all is not in that table, so the preview counts those separately, above the summary: how many of the file's rows will not be imported, the reason for each one and the rows it applies to. The summary counts what will be imported, not what will not, so this is the only place a dropped row shows up before it is too late to do anything about it. The commonest reason is a blank account cell with no fallback account chosen, and the warning says so and points at the select that fixes it (see Account and Currency Columns).

The preview also warns, above the summary, when the direction looks doubtful:

  • Rows with no usable transaction type — you mapped a Type column but some rows are blank or hold a value Budget does not recognize. Those rows fall back to the sign of the amount.
  • A batch heading the wrong way — nearly every row would be added as income into an account whose history is nearly all expenses (or the other way round). This usually means the Type column is unmapped, so check the mapping before importing.

Matched to bills counts the rows that will pay one of your bills or take the place of a payment one already booked, and each of those rows carries a Bill badge with the bill's name. A row that could be the payment of more than one bill is marked Check bill instead, and the preview says how many there are. Those rows are imported without a bill, for you to match from Find Duplicates on the Transactions page or with Link to bill.

It also says when rows have no description: how many, and which rows they are. Description has to be mapped, but a file can still leave the cell empty on some rows. Those rows are imported all the same, however an import rule that matches on the description has nothing to match, so they come through uncategorized. If the cells should not be empty, the file is the place to look: an export that shows a value on screen does not always write it to the CSV. A row whose description is filled in by a Set Description rule is not counted.

None of these warnings block the import; they are there to catch a mapping mistake before it reaches your balances.

Tip: If something looks wrong in the preview, go back and adjust your column mapping or delimiter settings. No data is saved until you execute the import.

7. Execute Import#

When the preview looks correct, click Execute Import to save the transactions to your selected account. The app reports how many transactions were imported and how many were skipped as duplicates.

If some rows could not be imported, a window opens listing why, one line per reason with the row numbers it applies to — so a file where fourteen rows have the same problem reads as one problem, not fourteen. The usual causes are a date the app could not read, an amount that is not a number, or an account name that does not exist. Fix the file (or the mapping) and import again; the rows that did import are not duplicated, because they are recognised as already present.

camt.053 Import#

camt.053 is the ISO 20022 bank-to-customer statement — the XML export offered by many European banks, and the standard one in Switzerland. Intraday camt.052 account reports import the same way. Upload the .xml file on the Import page.

  • Accounts are recognised by IBAN. Each statement in the file lists its account; if one of your Budget accounts carries the same IBAN it is matched automatically, otherwise pick the destination (the routing is remembered for next time).
  • Descriptions come from the richest field available: the remittance text the payer wrote, then the counterparty name, then the bank's own entry text. The counterparty is also filled into the Vendor field, and the bank's entry text into Notes when it adds anything.
  • Batch entries are expanded. A collective booking that carries several transaction details (a salary run, a card settlement) becomes one transaction per detail, each with its own counterparty and amount.
  • Duplicates are detected by the bank's entry reference, which survives re-downloads and overlapping statement periods. If a bank stamps the same reference on every entry, the importer falls back to matching on date, amount and description instead.

Note: As with OFX, the date, amount and type are always taken from the file's own structure; only the text fields (description, vendor, notes, reference) can be remapped.

Spreadsheet files#

Some banks only offer a statement as an Excel file. Pick the .xls or .xlsx file (or a LibreOffice .ods) on the import screen as it is; there is no need to save it as CSV first.

The app reads the first sheet whose table has at least three columns and a column of dates, so a cover sheet with the account details or an account overview is passed over, and turns it into a CSV. Hidden sheets are skipped. If your workbook has one sheet per account, the first one is imported. It then carries on exactly as for a CSV file: the same column mapping, saved templates, rules and duplicate checks. Lines the bank puts above the table, such as the account holder or the statement period, are skipped, and the table starts at its row of column headings.

  • Dates are taken from the cell's value rather than how the spreadsheet displays it, so they cannot come out with the day and month swapped. A date that is typed in as text, rather than being a real date cell, is read like a date in a CSV file.
  • Amounts are taken as the stored number, not the rounded or grouped figure on screen.
  • Some banks' "XLS" files are really a web page with a table in it. Those work too, and their cells are read exactly as written.
  • A CSV or tab-separated file that was saved with an .xls name is read as the text file it is.
  • A password-protected spreadsheet cannot be read. Remove the password, or save the sheet as CSV and import that.

Import folder#

An import folder imports statements for you. You drop the file your bank gave you into a folder in your Nextcloud Files, and Budget imports it in the background, into the account and with the import template you chose for that folder. There is no column mapping to do each time, and the folder can be shared like any other, so the people you share it with can drop their statements in without opening Budget.

Setting it up#

  1. Import one statement from the bank the usual way, and on the mapping screen choose Save mapping as template…. This is what tells Budget how to read that bank's files. OFX, QIF and camt.053 files need no template.
  2. Go to Settings > Import folder. The folder defaults to Budget/Import; choose another with Browse… if you prefer.
  3. Add a subfolder: give it a name, pick the account its statements belong to, and pick the template. Budget creates the subfolder in your Files.

Add one subfolder for each account. An import folder can only post into your own open accounts, not into an account someone has shared with you. The folder itself has to be in your own Files as well: a folder somebody shared with you, or one on external storage, cannot be used as an import folder.

You can share a subfolder with other people the usual Nextcloud way. Anyone you give write access can add statements to it, and so transactions to that account, so share it only with people you would let do that.

What happens to a file#

Budget looks in the subfolders every 15 minutes, and Check now in Settings looks straight away. A file that was changed in the last two minutes is left for the next look, in case it is still uploading.

Only files added after a subfolder is set up are imported. Files that are already in it when you add it in Settings, or that are already in the new place when you change the main folder, are left where they are, so pointing Budget at a folder you already keep statements in does not import or move your old files. Settings says how many it found. A file left alone stays that way until it changes: if you save a new statement over it with the same name, through Replace in Files or the desktop client, it is imported like any new file. Restoring a backup leaves the files already in the restored subfolders alone in the same way; if they cannot be read at that moment, the subfolders are switched off and the restore says so.

The fileWhere it goesWhat you are told
ImportsImported, inside the subfolderHow many transactions were imported, how many were already there, and how many rows could not be read
Was imported beforeImported0 imported; every row was already there
Cannot be importedFailed, inside the subfolderThe reason

A file ends up in Failed when:

  • the template expects columns the file does not have (the notification names them);
  • it is a CSV or spreadsheet and the subfolder has no CSV template;
  • nothing in it could be read as a transaction;
  • it is not a statement at all, is empty, or is larger than 10 MB;
  • Budget stopped while reading it, which a damaged or unusual file can cause. It is set aside the next time Budget looks rather than read again, so it can't hold up anyone else's import folders. Try it on the Import page, or download the statement from your bank again;
  • it has transactions for more than one account, as an OFX or QFX file of all your accounts at a bank can, or it is in a different currency from the subfolder's account. Import it from the Import page, where you choose the account for each. An account the file has no transactions for does not count, and neither does the same account named more than once, such as a camt file with a statement for each day;
  • nearly every row would be imported the wrong way round, income as spending or the reverse. This almost always means the wrong template, and an import cannot be undone, so the file is left for you to import from the Import page, where the preview shows what will happen.

Rows that cannot be read do not stop the others: the rest of the file is imported and the notification says how many were left out.

Review first#

A subfolder can hold its files instead of importing them. Choose Hold for review for it under Settings > Import folder. A statement dropped into it is moved into a Review folder inside the subfolder and you get a notification; nothing is imported yet.

Open Import and the file is listed under Waiting for review:

  • Review opens it in the usual import steps, with the subfolder's template and account already chosen. Check the preview and import it as you would a file you uploaded; it is then moved to Imported. If you close the steps without importing, the file stays waiting.
  • Skip moves it to a Skipped folder inside the subfolder. Nothing is imported and nothing is deleted.

Unlike a subfolder that imports automatically, a reviewed file uses the whole template, including an Account or Currency column, because you can see in the preview where each row will go. A statement that holds several accounts is matched account by account, the same as a file you uploaded, and the subfolder's account is only chosen for you when the file holds that one account.

Switching a subfolder back to importing automatically does not import what is already waiting: review or skip those.

Good to know#

  • Every row goes into the subfolder's account. If the template maps an Account or Currency column, those two are not used for an import folder.
  • A spreadsheet is read as comma-separated whatever delimiter the template has, so a template saved from the bank's CSV works for its Excel file too.
  • Removing a subfolder in Settings stops the importing. The folder in your Files, and everything in it, stays.
  • An account cannot be closed while an import folder still posts into it. Switch the subfolder off or remove it first.
  • Deleting an account or an import template removes the subfolders that used it from Settings.
  • Changing the main folder does not move subfolders you already have; they are created again under the new folder.

OFX Import#

OFX (Open Financial Exchange) files are structured financial data files that many banks offer as a download option, sometimes labeled as "Microsoft Money" or "Quicken" format.

  1. Navigate to Import > Upload File and select your .ofx file
  2. The app parses the file automatically and fills in the field mapping for you
  3. If the file contains an account identifier, the app attempts to match it to one of your existing accounts
  4. Review the parsed transactions in the Preview step
  5. Click Execute Import to save

Note: OFX files contain standardized field names, so the date, amount, and type are always taken from the file's own structure and cannot be remapped.

Choosing which OFX field becomes the description#

An OFX transaction carries two pieces of text: a name and a memo. Most banks put the payee in the name, so that is what the app uses by default, and the memo is kept in the transaction's notes.

Some banks do the opposite and put the useful text in the memo. In the mapping step, set Description to memo and the app will use that instead. The available fields for OFX are:

FieldDefault sourceNotes
Descriptiondescription (the OFX name)Falls back to the memo when the name is empty
NotesmemoLeft empty when the memo is already the description
Vendordescription (the OFX name)
ReferencereferenceFalls back to the bank's transaction ID

Note: Changing the mapping only affects transactions you import from then on; it does not rewrite transactions you have already imported. Re-importing the same statement will not update them either, because duplicate detection recognises them as already imported. Delete them first if you want them re-created with the new mapping.

QIF Import#

QIF (Quicken Interchange Format) is a legacy format still exported by some banks and financial software.

  1. Navigate to Import > Upload File and select your .qif file
  2. Adjust the field mapping if you want different text in the description or notes
  3. Route each account in the file to one of your accounts
  4. Review the preview and click Execute Import

A QIF file often holds several accounts, and can also contain your category list and other non-transaction sections. Only the accounts are offered for import; everything else is ignored.

QIF fields#

Accounts are identified by the name in the file. A file that does not name its accounts shows them as Account 1, Account 2 and so on, in the order they appear.

FieldDefault sourceNotes
DescriptiondescriptionThe payee line
NotesmemoLeft empty when the memo is already the description
Vendordescription
ReferencereferenceThe check number

As with OFX, the date, amount and type come from the file's structure and cannot be remapped.

Tip: QIF has limited field support compared to OFX and CSV. If your bank offers OFX as an alternative, prefer that format for more complete data.

Switching from Another App#

If you are moving to Budget from another budgeting app, you do not have to start again. Export your data from the app you are leaving and import the file here. Budget has a preset for each of the apps below, which knows the columns of that app's export, so there is no column mapping to do and every account in the file is imported in one go.

AppWhere to find the export
Firefly IIIOptions > Export data, then export all transactions.
YNABThe plan menu, Export Plan Data. Unzip the download and pick the file ending in "Register". A YNAB 4 register export works too.
Actual BudgetOpen All accounts and choose Export from the menu above the transactions.
MintThe transactions.csv file from Mint's Export all transactions.
Monarch MoneySettings > Data > Download transactions.
Toshl FinanceExport your entries as a CSV file, covering all accounts. See Toshl Finance below.

Choosing the app#

On the Import page, open Switching from another app? under the upload area. It lists the same apps with a one-line reminder of where each keeps its export. Pick yours, then choose the file. You can also upload the file first and pick the app from the Import Format dropdown on the next step.

You often do not need to pick at all. When a CSV file's header row matches one of these exports, Budget selects that app's preset for you and says so under the dropdown. Choose Custom CSV (manual mapping) instead if you would rather map the columns yourself.

The column mapping controls are hidden, as the preset supplies the mapping, and the preview lists the accounts and categories the import will create before anything is written. Tab-separated .tsv files are accepted as well as .csv, since that is what YNAB exports for a plan whose currency uses a comma for decimals.

What is created#

  • Accounts. Each account named in the file is matched to one of yours by name. If you have no account by that name, it is created, with its type guessed from the name ("Savings", "Credit Card", "Mortgage" and so on). A Firefly III loan, mortgage or credit card takes its type from Firefly III instead.
  • Categories. Categories are created as they are needed, and one you already have is reused, even when it is a subcategory, such as Groceries under Food (see Category Column). YNAB and Actual Budget file categories in groups, and each group becomes a parent category with its categories beneath it. Income that YNAB assigned to "Ready to Assign" is left uncategorised, as that is not a real category in YNAB.
  • Tags. Firefly III tags, Mint labels, Monarch Money tags and Toshl tags come across as tags, filed in a tag set called "Tags" under the transaction's category.
  • Payees and notes. The payee or merchant goes into the vendor field, and the app's notes or memo go into the transaction's notes. A Firefly III transaction with an amount in a second currency is imported in the account's own currency, with the foreign amount kept in its notes.

Transfers#

Money moved between two of your own accounts is imported into both of them, and the two sides are then linked as a transfer, so it counts as neither income nor spending. Each app marks a transfer differently: Firefly III has one of your accounts on each side, YNAB names the other account in the payee ("Transfer : Savings"), Actual Budget leaves a transfer without a category, and Mint and Monarch Money file it under "Transfer" or "Credit Card Payment".

Two sides are only linked when both are in the file: different accounts, opposite directions, the same amount, and dated within three days of each other. If you exported only one account, or the other side is missing for some other reason, the row is imported as an ordinary transaction and you can match it by hand later.

Split transactions#

Split transactions are not rebuilt from these exports. Where an export writes each part of a split as a row of its own, each part is imported as an ordinary transaction with its own category. Actual Budget also writes a line for the split's total, and that line is dropped, or the money would be counted twice.

Importing the same export again#

Each row is recognised by what the export itself contains, never by anything you choose on the import screen, so importing the same export a second time adds nothing. This makes it safe to import a later export that overlaps an earlier one.

That also holds when you imported the export before as a plain CSV with your own column mapping. A row is then counted as a duplicate when the account already has a transaction on the same day, for the same amount in the same direction, whose description or payee matches the row's: the same text, or one holding the other as whole words. Notes are not compared, so two different purchases that only share a memo such as "Subscription" are both kept, and the order the file lists its rows in makes no difference. Each transaction already there stands for one row of the file, so two genuinely identical purchases are still both kept. These rows are flagged and skipped like any other duplicate, and Import flagged duplicates too brings them in if they really are new.

Things to know#

  • Only Firefly III says which currency an account is in. Accounts created from any of the other exports are created in your default currency, so set that in Settings before importing if it is not the one you use. An account you already have keeps its own currency.
  • Tags need a category. A tag set belongs to a category, so a row with no category is imported without its tags.
  • YNAB dates follow your YNAB date setting. Budget works out whether the day or the month comes first from the dates in the whole file. If none of them has a day after the 12th, every date can be read either way and the file is read month first, so check a few dates in the preview.
  • Transfers between accounts in different currencies are not linked. Both sides are imported as ordinary transactions.

Toshl Finance#

To import from Toshl Finance, export your data as CSV from Toshl and upload the file in the Budget app. Then select Toshl Finance from the Import Format dropdown.

Key features of the Toshl preset:

  • Language-independent — Works regardless of the language Toshl used for the export. Column detection is based on position, not header names.
  • Date and number handling — Automatically parses European-style dates (DD.MM.YY) and comma-decimal amounts (1.234,56).
  • Multi-currency support — When a transaction is in a foreign currency, the preset uses the converted amount from Toshl's "In Main Currency" column, so all imported amounts are in your main currency.
  • Category auto-creation — Categories from Toshl's Category column are created automatically if they do not already exist in Budget.
  • Tag set integration — Tags from Toshl are imported as tag sets attached to the corresponding category. This preserves your Toshl tagging structure.
  • Account auto-creation — Accounts from Toshl's Account column are created automatically. Account types are inferred from the name (e.g., "Cash" becomes a cash account, "Investment" becomes an investment account).
  • Transfer handling — Rows where Toshl's Category is "transaction" (inter-account transfers) are skipped automatically since transfers are not regular transactions.
  • Full preview — Before executing, the preview shows accounts to create, categories to create, tags to import, and transfer rows that will be skipped.

Saved Import Templates#

If you import from the same bank regularly, you can save your import configuration as a reusable template instead of repeating the setup on every import. Templates are private to your account, and what a template stores depends on the file format:

  • CSV — the column mapping, the CSV delimiter, the "skip first row" option, and the character encoding if you picked one by hand (plus an optional default destination account).
  • OFX / QIF — the account routing, i.e. which destination account each source account in the file maps to, together with the Description / Notes / Vendor / Reference mapping if you changed it from the defaults.

All templates also remember the import options (skip duplicates, apply rules). When you upload a file, only templates matching that file's format are offered.

CSV: saving a column mapping#

  1. Upload a CSV file and map the columns as usual (see Column Mapping).
  2. Click Save mapping as template… in the column mapping step.
  3. Give the template a name (for example, My Bank Checking) and save it.

To reuse it, upload a CSV file and pick your template from the My Templates group in the Import Format dropdown. The column mapping is filled in automatically, and the preview redraws under the template's delimiter, header setting and encoding. A template saved without an encoding leaves the encoding picker as it is. You can still tweak any column before previewing — adjusting a mapping switches the import back to a custom mapping for that run, leaving the saved template unchanged.

Note: A CSV template stores column names, so it works on any future export from the same bank as long as the column headers stay the same. If your bank changes its export format, save a new template.

OFX / QIF: saving account routing#

OFX and QIF fields are standardized, so their mapping is filled in for you and rarely needs saving. What is repetitive is routing: a single file can contain several accounts, and each import you re-pick which of your accounts each one maps to.

  1. Upload an OFX or QIF file and reach the Review & Import step.
  2. In Map Source Accounts to Destination Accounts, set each source account's destination.
  3. Click Save routing as template… above the list and name it.

To reuse it, upload a file of the same format and pick your template from the Saved Account Routing dropdown; the destinations are filled in automatically. You can still change any destination before importing.

If you changed the Description / Notes / Vendor / Reference mapping before saving, that is stored with the routing and restored alongside it, so a bank that puts the payee in the memo does not have to be re-mapped on every import.

Note: OFX already auto-matches source accounts to your accounts by account number; a saved routing template is most useful for QIF (which has no auto-match) or when the auto-match is wrong or incomplete.

Automatic routing memory#

You don't have to save a named template to benefit from routing memory. After any OFX/QIF import, the app quietly remembers which destination account each source account was routed to. The next time you import a file of the same format, those destinations are pre-filled automatically — including for QIF, which has no other auto-matching. If you route a source account somewhere different, the new choice is remembered instead. This works alongside named templates: selecting a template still takes precedence, and you can always change any destination before importing.

Managing templates#

Click Manage templates (in the column-mapping step for CSV, or above the account-routing list for OFX/QIF) to rename or delete your saved templates. Each is labelled with its format.

Duplicate Detection#

The app automatically checks for duplicate transactions during import. A transaction is considered a duplicate when it matches an existing transaction in the same account on all of the following:

  • Date
  • Amount
  • Description
  • Reference (if mapped)

Duplicates are skipped during import and reported in the results summary. This makes it safe to import overlapping date ranges without creating duplicate entries.

Detection is occurrence-aware: if one file legitimately contains several identical rows (e.g. two same-priced coffees on the same day), they all import as distinct transactions — and re-importing the same statement still skips every one of them. Re-importing an older statement also recovers any rows that earlier versions mis-flagged as duplicates.

A CSV file imported again with its columns mapped differently, for example with the category added to the description, is recognised as well. A row counts as already imported when the account holds a transaction that an earlier import of a file put there, on the same day, for the same amount in the same direction, whose description, payee and reference all come from that row's own columns, alone or joined the way a mapping joins them. A purchase that only shares the shop or a few words with one already there is not taken for it, so the lines of a later statement still come in. This also works when one of your import rules set the payee or rewrote the description or reference, because the app remembers what it read from the row before any rule ran. A payee you edit by hand is followed too: the app keeps the payee the import stored next to the one you see, and compares the row with that. For a transaction imported with an earlier version of the app, the kept payee is the one it had when you updated, so an edit you make after updating is followed, but a payee you edited before updating can't be, and such a row may come in a second time. Only transactions an import put in the account are compared, never one you typed in, one from bank sync, one carrying the bank's own transaction ID as OFX files do, or a payment the app booked for a bill, and each one stands for a single row of the file, so two genuinely identical purchases are both kept.

To deliberately import rows that are flagged as duplicates (for example, genuinely repeated payments your bank exported without a unique reference), tick Import flagged duplicates too in the preview step. The option applies to the whole batch and resets for every new import; saved import templates can also store it.

Note: Duplicate detection is based on exact matching. If your bank changes the description text between exports, the same transaction may not be recognized as a duplicate. If your export has a unique sequence number per transaction, map it to the Reference column — it is stored on each transaction and makes duplicate detection exact. (OFX imports use the bank's own unique transaction IDs automatically.)

Bills, Transfers and Income in a Statement#

An imported transaction that matches a bill's, a recurring transfer's or a recurring income's pattern pays it, the same as marking it paid by hand. Where the app had already booked that payment, because you marked it paid or auto-pay did, the bank's transaction takes the place of the booked one rather than being added beside it, and anything you put on the booked payment, such as receipts, tags and notes, moves across. This holds whatever order the statement lists the payments in. See Bills, Transfers and Income for the details. A bill with no pattern is matched by its name, to the penny and within a few days; see Bills with no pattern.

Import History#

The Import History tab on the Import page lists your recent imports, showing the date of each import, the account it went into, how many transactions it added and the range of dates those transactions cover.

Removing an Import#

There is no single button to undo an import. If you imported a file into the wrong account or with the wrong column mapping, use the history to find the account and the dates it covered, then remove the transactions from the Transactions page:

  1. Filter the Transactions page to that account and date range
  2. Click Bulk Actions, select the page and choose Select all N matching transactions if the import runs over more than one page
  3. Delete the selection

Check the filter before deleting, as anything else in that account and date range will be selected as well. See Bulk Operations.

Tips#

  • The first row of a CSV file must contain column headers. Files without headers cannot be mapped correctly.
  • CSV encoding is auto-detected for common Western encodings (ISO-8859-1, Windows-1252, ISO-8859-15). For other encodings, save as UTF-8 before importing.
  • For large imports, the preview shows a sample of rows. Scroll through to verify different transaction types are parsed correctly.
  • Import into the correct account before clicking Execute Import -- transactions cannot be moved between accounts after import.
  • Use Import Rules to automatically categorize transactions after import, saving you from manually categorizing each one.
  • Import Rules -- Create rules to auto-categorize imported transactions by matching description or vendor patterns
  • Transactions -- View, edit, and manage all your transactions including imported ones
  • Accounts -- Set up accounts that correspond to your bank accounts before importing

Settings#

  • Auto-apply import rules -- Sets whether Apply import rules for categorization starts ticked for each new file. Turn it off if you prefer to review transactions before categorizing; you can still tick the box for a single import.
  • Import flagged duplicates too -- Per-import option in the preview step. Off by default, so flagged duplicates are skipped unless you explicitly opt in.