How to import your bank statement

Most of a quarterly update already exists, sitting in your bank account as a statement: every transaction, dated and described. Importing it means the software does the typing and your job shrinks to confirming categories. The statement has to arrive as a CSV file.

First, what a CSV is

CSV stands for “comma-separated values”: a plain table of your transactions, one row per transaction, that any software can read. It looks like a bare-bones spreadsheet, and that’s what it is. Banks also offer PDF statements, which are made for reading, not importing. Given the choice, CSV is the one you want.

Getting the CSV out of your bank

Every UK bank can produce one, and nearly every bank hides it somewhere different. The pattern that finds it:

• Use online banking in a web browser if you can. The traditional banks (Barclays, HSBC, Lloyds, NatWest, Santander and friends) usually keep the export on the website, not in the mobile app.
• Go to the account’s transactions or statements page and look for “Export”, “Download” or “Download transactions”.
• Pick CSV as the format. It’s sometimes labelled “spreadsheet” or “Excel (CSV)”.
• Choose a date range that covers the whole quarter: 6 April to 5 July for Quarter 1, and so on. A few days of overlap does no harm, because overlaps are caught as duplicates later.
• App-based banks (Monzo, Starling and the like) export from the app itself: look for “statements” or “export” on the account screen.

Only offered PDF or Excel? An Excel file can be saved as CSV from inside Excel or any spreadsheet program (File → Save As). A PDF can’t be reliably imported by anything, so keep looking for the export option: it’s there.

The three columns that matter

An importer needs three things from each row: the date, the description (who was paid, or who paid you), and the amount. Everything else in the file (balance columns, reference numbers, transaction types) is ignored. Amounts come in two normal layouts, and both are fine: one amount column where money out is negative, or a separate “money in” and “money out” pair (Santander and NatWest export the pair). You just tell the importer which layout your bank used, and a good one guesses first.

When dates or amounts won’t read

A careful importer shows a preview and marks anything it can’t read before saving a single row. The usual causes are quick fixes:

• Unreadable dates: the date style is set wrong. UK banks write day/month/year; some exports use year-month-day; American-style files put the month first. Switch the style and the column comes back to life.
• Unreadable amounts: usually the wrong column is matched. A running-balance column looks a lot like an amount column, but it holds the account total, not the transaction, so point the matching at the real amount column.
• Pound signs, commas, brackets round negatives and “DR”/“CR” markings are all normal in bank exports and read fine. They aren’t the problem.
• The first few rows unreadable: that’s usually a summary block, covered next.

Balance and summary rows: delete them

Many bank CSVs open with a few rows about the account (the account name, “balance brought forward”, some totals) before the real table starts, and some add a closing-balance row at the end. Those rows aren’t transactions, and when they sit above the row of column headings they confuse the column matching. The fix takes a minute: open the CSV in any spreadsheet, delete every row above the column headings along with any total or balance rows, and save as CSV again. A stray one left behind does no damage: rows that can’t be read are skipped and counted, never saved.

What happens with duplicates

Overlapping exports are normal: you download April to July, then July to October, and a week turns up in both. A careful importer checks each incoming row against what’s already in your records (same date, same description, same amount) and against the rest of the file, then flags the matches instead of quietly importing them twice. In QuarterFile, flagged rows arrive unticked at the review step and you decide. One caution: two genuinely separate transactions can look identical (two £3.50 coffees at the same café on the same day), so glance at what’s flagged rather than trusting the flag blindly, and tick the real ones back in.

If nothing was imported, or only part

• “Nothing was imported” means exactly that: nothing was saved, so there’s nothing to undo. It almost always means the date or amount column is matched wrong, so fix the matching and try the same file again.
• An import that stops partway (usually a dropped connection) tells you how many rows were saved. Re-upload the same file: everything already saved is flagged as a duplicate, so nothing doubles up and the missing rows come in.
• Rows skipped as unreadable are counted at review. A handful is usually summary rows doing no harm; a big number means the column matching needs another look.

After the import, nothing is filed

Importing saves transactions to your records, and that’s all it does. In QuarterFile every imported line arrives as a suggestion: it doesn’t count towards a quarter until you confirm its category, and nothing goes to HMRC until you press file. What goes in each category is the guide for the confirming part, and the digital records rules explain why an import like this keeps your records digitally linked end to end. QuarterFile also remembers your bank’s column layout after the first import, so next quarter the matching step checks itself.

CSV import is part of QuarterFile Plus, £12.50 a quarter with your first quarter free. The by-hand version is free forever. You can see the import screen by screen on the how it works page, and filing your first quarterly update covers what comes after.