Key takeaways
What this article covers, in order:
- Two businesses, two sensible answers
- Side by side
- Where CSV import trips people up
- A simple rule for each account
- Mixing both without making a mess
- What about PDF statements?
Use a live bank feed for any account with regular activity, and use a bank statement CSV import for quiet accounts, banks that won't connect, or catching up on history. Both get accurate bank data into your books without retyping. The difference is who does the fetching: the feed does it for you, a CSV needs you to download and upload a file each time.
Two businesses, two sensible answers
Raj is a sparky in Parramatta. His everyday business account has 90 to 120 transactions a month: supplier payments, customer deposits, fuel, tolls. Downloading a CSV every week would be one more chore he'd skip. A feed is the obvious choice.
Sophie runs a small online stationery shop from Hobart. Alongside her main account she has a term deposit that earns interest once a quarter and a rarely used second card. Paying for, or using up, a feed slot on accounts that move four times a year makes little sense. She imports a CSV for those when she reconciles.
Neither of them is doing it wrong. The right choice is per account, not per business.
Side by side
| Live bank feed | CSV statement import | |
|---|---|---|
| Effort after setup | Almost none; transactions arrive on their own | Log in, download, upload, every time |
| Timeliness | Usually within a day | Only as fresh as your last download |
| Typing errors | None | None, as long as the file is clean |
| Risk of duplicates | Low, unless you also import a CSV for the same dates | Higher, because overlapping date ranges are easy to pick |
| Works for every bank | Most Australian banks and cards, not all | Almost any bank that offers a CSV download |
| Good for history | Depends on how far back the bank shares | Yes, you choose the date range |
| Stops working when | Consent expires or the account changes | You forget to do it |
Where CSV import trips people up
CSV is a plain file format, and plain files are easy to get slightly wrong. These are the traps we see most often.
Date format
Australian banks write dates day-first: 03/10/2026 is 3 Oct 2026. Some exports, especially from overseas-built tools, write month-first, which would read that same date as 10 Mar 2026. If your software guesses the wrong way, early-month transactions jump around the calendar and anything after the 12th fails to import at all. Check the first few rows after every import.
Debits and credits
Some banks export one "Amount" column with negatives for money out. Others use separate "Debit" and "Credit" columns. A few show money out as positive. If the import treats your $2,400 rent payment as income, your profit just went up by $4,800 (the $2,400 expense disappears and a fake $2,400 of income appears). Look at a known payment and make sure it came in on the right side.
Overlapping date ranges
This is the big one. Say you imported 1 Sep 2026 to 30 Sep 2026 last month. This month you download "last 60 days" because it's the default, which covers 7 Aug 2026 to 5 Oct 2026. Import that and Sep 2026 is now in your books twice.
The fix is boring but reliable: always download from the day after your last imported transaction to the end of the period you're working on. Write the last imported date somewhere you'll see it next time.
Running balance columns
Many CSVs include a running balance. It's useful for checking, but make sure your import isn't treating it as a transaction amount.
Opening a CSV in Excel first
Excel likes to "help". It may reformat dates, strip leading zeros from reference numbers, or turn long numbers into scientific notation. If you need to look at the file, open a copy, and import the untouched original.
A simple rule for each account
Here's a practical way to decide, account by account:
- [ ] More than about 20 transactions a month? Connect a feed.
- [ ] Customers pay into it? Connect a feed, so you can see who's paid without logging into the bank.
- [ ] Bank won't connect? CSV import, on a fixed schedule (fortnightly or monthly).
- [ ] Quiet savings, term deposit or loan account? CSV at month-end is fine.
- [ ] Catching up on last year? CSV for the old period, then a feed from your clean start date onward.
Mixing both without making a mess
Plenty of businesses run feeds and imports side by side. The thing to get right is the hand-over date.
Suppose Raj has been importing CSVs up to 31 Aug 2026 and connects a feed on 15 Sep 2026. If the feed pulls in history back to 1 Aug 2026, Aug 2026 now exists twice: once from the CSV and once from the feed. Before connecting, decide the first date the feed should own (here, 1 Sep 2026), and either set the feed's start date there or remove the feed's earlier lines.
Then reconcile. If the bank's closing balance at 30 Sep 2026 is $11,846.20 and your books say $11,846.20, you know the hand-over worked. If your books say something much bigger, look for a block of doubled transactions around the switch date first.
What about PDF statements?
PDF statements are made for reading, not importing. Copying lines out of a PDF puts you back to typing by hand, with the same typos. Most Australian internet banking sites let you download CSV for business accounts even if they push PDF by default; look for "export" or "download transactions" rather than "statements". Keep the PDF, though. It's the bank's official record, and its closing balance is what you reconcile to.
How HelloBooks helps
HelloBooks supports both. You can connect your bank account (we work with most Australian banks and cards) and you can import bank statement CSVs. The Free plan includes one live bank feed plus CSV import, so a business like Sophie's can run the main account live and import the quiet ones. If you want a feed on every account, Pro (A$30 a month) includes unlimited bank connections.
Either way, transactions land in the same review list for you to confirm or change categories, and the reconcile screen then matches statement lines to your ledger with an AI suggestion on each, so you're left with just the exceptions. See how reconciliation works or compare plans.
FAQs
Is a CSV import less accurate than a bank feed?
Not if the file is clean. The data comes from your bank either way. The accuracy risk with CSV is human: picking overlapping dates, or a file whose date or sign format was misread.
How far back can I import with a CSV?
That depends on how much history your internet banking lets you download. Many banks allow several years for business accounts; some limit it. If you need older data, ask the bank for statements.
Can I switch an account from CSV to a feed later?
Yes. Just choose the date the feed takes over, make sure nothing before that date comes in twice, and reconcile the first month after the switch carefully.
What's the best file format to download?
CSV is the most widely supported. Some banks offer OFX or QIF as well. If you're not sure, CSV is the safe pick.
My bank's CSV has extra columns I don't need. Is that a problem?
Usually not. Imports normally let you map the columns you need (date, description, amount or debit/credit) and ignore the rest.
Pick a feed where money moves often, a CSV where it doesn't, and write down your hand-over dates. That's most of the battle.
Start free, no card needed. Try HelloBooks Free
