Guides · CSV vs Excel
CSV or Excel: it matters less than which version of Excel.
Both formats import cleanly from every UK bank worth the name. The format that doesn’t is the one worth knowing to avoid, and Excel specifically hides two failure modes CSV can’t.
What actually works
Two formats in, one refused on purpose.
CSV
Plain text, no formatting to strip. Delimiter is detected from the header line automatically, so comma and semicolon-separated files both import without asking the client which one they have.
Excel (.xlsx)
Converted on upload. Handles what most current banking portals export by default: the newer format almost every bank moved to, and the one worth asking for if the client’s screen offers a choice.
Legacy Excel (.xls)
Refused at the form rather than half-read. It’s an older binary format that’s harder to parse reliably, and a visible error here is safer than a file that imports with the wrong columns and nothing looking wrong.
The specific failure modes
What goes wrong, and in which format.
A second tab hiding the transactions
Only the first sheet in the workbook is read. A bank statement export is normally one sheet, so this only bites when a summary or cover page has been added in front of the transaction list. Delete the extra tab, or move transactions to the first one, before sending it on.
A date that only looks like one
A genuine Excel date cell converts automatically to the UK format the parsers expect. A column that’s merely styled to display as a date, without being a real date type underneath, comes through as a plain number instead. Worth a glance at the date column after import if the file has been reshaped by hand.
No header row at all
Some exports (more often CSV than Excel) drop straight into data with no column names. Handled by checking whether the first cell looks like a date rather than a header, and inferring columns by position when it does.
A delimiter that changed on the way
Opening a CSV in a spreadsheet program set to a different regional format and re-saving it can silently swap commas for semicolons. Send the file the bank produced, unopened, where that’s an option.
FAQ
Before you send a client a file request.
Which should I ask a client for, CSV or Excel?
Whichever their bank’s export screen offers by default. Both import cleanly. CSV is marginally safer because there’s no formula or formatting to strip out, but the difference is small enough that it isn’t worth a client fighting their online banking to get the “right” one.
Why does .xls get refused instead of imported?
Legacy .xls is a binary format that’s harder to read reliably than .xlsx, and a half-read file that imports with the wrong columns is worse than one that doesn’t import at all. Refusing it at the form is the safer failure: an error you can see, not a silently wrong trial balance. Ask the client to re-export as .xlsx or CSV, which almost every current banking portal offers.
Can a CSV file be missing its header row?
Yes, and it still needs to import correctly rather than being rejected. A file is treated as headerless when its first cell looks like a date rather than a column name, and columns are then inferred by position instead of by name. It’s kept narrow on purpose: inventing headers for a file that already had them would break something that currently works.
What actually goes wrong with Excel exports specifically?
Two things. Only the first sheet in the workbook is read, so a summary or cover tab placed before the transactions is a real problem, not a cosmetic one. And a date only converts automatically if the cell is a genuine Excel date type: a column that merely looks like a date because of number formatting, without actually being one, comes through as a plain number instead. Re-exporting straight from the bank rather than reshaping the file avoids both.
Does the delimiter in a CSV matter?
It’s detected automatically from the header line, where the character counts are unambiguous, so comma and semicolon-delimited files both import. It matters more if a file has been opened and re-saved in a spreadsheet program set to a different regional format, which can quietly swap the delimiter.
What about a file exported in a non-UK date format?
05/01/2025 is read as 5 January, the UK interpretation. A file that’s been through a US-formatted export or spreadsheet needs checking before import, because the wrong reading doesn’t error, it just quietly moves transactions between months.
More formats
What the other exports look like.
Founding firms · 22 of 25 places left
Send a real export and see what comes back.
Fourteen days free and no card. A sample client is set up when you sign up, so you can see the whole thing before any real data goes in.
-
Solo
£49£33 /month Up to 50 clients -
Practice
£129£86 /month Up to 250 clients -
Firm
£299£199 /month Up to 1,500 clients