How to categorise bank transactions automatically
On this page
- 1.Bank reconciliation without the headache
- 2.How to categorise bank transactions automatically
- 3.How to connect your bank to your accounting (and why it changes everything)
- 4.Stop doing your bookkeeping every week: what an AI bookkeeper actually does
Sorting 300 transactions a month by hand is the chore every founder hates. Here's how automatic transaction categorisation works in Ledgers — a daily close plus a short review queue, instead of one monthly mountain.
The chore: 300 lines, every single month
Open your business bank account at the end of a busy month and you'll see it — a wall of transactions. Maybe 100, maybe 300, maybe more. Each one needs a label before any of it means anything: what is it, what's it for, is there VAT on it, does it count as a business cost at all.
That's categorising. And it's the single most-hated job in small-business bookkeeping, for one simple reason: it's hundreds of tiny, boring decisions, and it never stops. Finish this month and next month's pile is already growing.
The worst part isn't the work. It's the falling behind. Miss one month and you've got 600 lines. Miss a quarter and you've got a wall you genuinely don't want to climb — so you don't, and now your VAT return is guesswork and your accountant is charging you to untangle it.
It doesn't have to be a monthly event at all. Here's how to categorise bank transactions automatically, so the pile never builds up in the first place.
The hard way today: a spreadsheet or a pile of rules
There are two versions of the manual slog, and you've probably done both.
The spreadsheet version. You export your bank statement, paste it into a sheet, and type a category next to every line. "Office supplies." "Software." "Subsistence." You squint at "SP * AMZNMKTPLACE" and try to remember what you bought. You do this every line, every month, and you're never quite sure you've coded the VAT right.
The rules version (Xero and similar). A bit better. You set up bank rules so that anything from "Google" lands as software, and you use "find & match" to pair payments with bills. But you have to build and maintain those rules, they break when a supplier's name changes, and anything not covered by a rule still drops to a manual pile. You've automated some of it, but you're still the engine — and you're still the one who falls behind.
Either way, the responsibility sits on you to remember, to be consistent, and to keep up. That's the part that fails.
How it's automatic in Ledgers: a daily close, not a monthly mountain
Ledgers flips the model. Instead of you sorting a month's worth of transactions in one sitting, the software sorts them a little every day — and gets smarter the longer you use it.
Here's what actually happens under the hood.
It categorises as transactions arrive. Your bank feed brings transactions in daily (no exports — see connecting your bank below). As each one lands, Ledgers categorises it automatically: the right account, the right VAT treatment, matched to a bill or invoice where one exists. This is automatic transaction categorisation done by an engine that reads the payment, the supplier, the amount and the pattern — not a rule you had to write.
It learns your business. The first time you confirm that "DGTLOCEAN" is your hosting cost, it remembers. Next month it doesn't ask. Over a few weeks it learns the shape of your spending — your suppliers, your recurring costs, your one-off patterns — so the share it gets right on its own keeps climbing. You're teaching it by doing almost nothing.
It runs a daily close. Rather than waiting for month-end, Ledgers closes the books a little each day. That means the work is always nearly done. There's no mountain to climb because the molehill gets cleared every morning.
It only asks when it's genuinely unsure. Every categorisation carries a confidence score. When the software is confident, it just files it. When it isn't — a new supplier, an odd amount, something ambiguous — it puts that one item in a short review queue (the Exception Inbox) for you to confirm. You're not reviewing 300 lines. You're glancing at the five it couldn't be sure about.
What you'd see and do
Open Ledgers and instead of a wall of unsorted transactions, you see a clean, categorised ledger — already done — plus a short queue of items asking "is this right?"
You work the queue in a couple of minutes: confirm, correct, or recategorise. Each correction teaches the system, so the queue gets shorter over time. Anything you change is fully traceable — the ledger is event-sourced (every change is recorded and nothing is overwritten), so you can always see what a transaction was, what it became, and why. Nothing disappears.
That's the whole job now. Not "sort 300 lines." Just "clear a short queue, twice a week." The categorising that used to eat an afternoon happens in the background while you run your business.
And because everything is categorised correctly as it goes, the knock-on effects are huge: your VAT return is already written, your reconciliation ties out continuously, and your year-end is clean. Categorisation is the foundation — get it automatic, and the rest of your books fall into place.
Ready to stop sorting transactions by hand? In Ledgers, every transaction is categorised automatically as it lands, with a short review queue for the rest — so your books stay current without the monthly slog. See your numbers without learning accounting → start free.
Want the bigger picture of what runs in the background? Stop doing your bookkeeping every week — what an AI bookkeeper actually does →
New to the term? Here's the plain-English version: What is reconciliation in accounting? →
Frequently asked questions
How do you categorise bank transactions automatically?
You connect your bank so transactions flow in by feed, then let software categorise each one as it arrives based on the supplier, amount and pattern. Good software learns your specific business over time and only asks you about transactions it's genuinely unsure of, instead of making you sort everything by hand.
Is automatic categorisation accurate?
It's accurate for the vast majority of transactions, and it gets more accurate the longer you use it because it learns from your corrections. In Ledgers, every categorisation carries a confidence score, and anything uncertain is flagged for you to review — so you get speed without losing control.
What happens if it categorises something wrong?
You correct it in the review queue, and the system learns from the change so it doesn't repeat the mistake. Because the ledger records every change without overwriting anything, corrections are fully traceable — you can always see what was changed and why.
Do I still need to check my transactions?
Only the handful the software flags. Instead of reviewing every line, you glance at a short queue of genuinely uncertain items. That's minutes a week rather than an afternoon a month.
See your numbers without learning accounting
Ledgers does the bookkeeping — bank feeds, VAT, year-end — and keeps your accountant in the loop. Free for pre-revenue founders.
Start free →Get the next guide by email
Plain-English accounting for founders — a couple of new guides a week. No spam, unsubscribe anytime.
Keep reading · Bank feed & categorising
Bank reconciliation without the headache
Month-end reconciliation — ticking every transaction against the statement — is the slog founders dread. Here's how continuous tie-out works in Ledgers, with a "Reconciled" badge you can trust, no late-night discrepancy hunting.
4 min readBank feed & categorisingHow to connect your bank to your accounting (and why it changes everything)
Downloading bank statements and uploading CSVs is the quiet chore behind every bookkeeping headache. Here's how a live bank feed via open banking works in Ledgers — connect once, and transactions flow in on their own.
4 min readBank feed & categorisingStop doing your bookkeeping every week: what an AI bookkeeper actually does
"AI bookkeeping" sounds like a buzzword until you see what it does day to day. Here's the real job — categorising, matching, reconciling, flagging — with human review and an explainable ledger, so nothing is a black box.
5 min read