Ask a practice principal why the BAS week runs long and the answer is almost always about volume. Too many transactions, too many clients, not enough hands.
Sit with the week and watch what actually happens, and the picture is different. The work does not run continuously and finish late. It stops, repeatedly, and each stop lasts a day or two because it is waiting on somebody who is not in your office.
That distinction is not pedantry. It decides which fixes are worth paying for. A tool that codes transactions faster does very little for a week that is mostly spent waiting.
Where the time actually goes
Four things consume a BAS quarter, in roughly this order of cost.
1. Statements that have not arrived
This is the big one, and it is almost never counted properly because it does not look like work. It looks like an email.
The cost is not the email. It is the round trip: send the request, wait two days, no reply, send a follow-up, wait, phone the client, get "sorry, I'll do it tonight", wait, receive a photo of a statement taken at an angle, ask for the PDF, wait.
The thing to notice is that this cost is roughly fixed per round of chasing, not per transaction. A client with 40 transactions and a client with 4,000 cost about the same to chase. That is why a book of small clients can be more expensive to service than a handful of large ones, and why the practices that feel busiest are often not the ones with the most transactions.
2. Statements that arrived but did not import cleanly
Australian banks do not produce one statement format. They produce dozens, and they change them without telling anyone.
The failure mode that hurts is not a file that refuses to open — that is obvious and gets fixed in a minute. It is the file that imports and looks fine, with some rows silently missing. A few of the ways that happens:
- A PDF exported by a system that writes no spaces between columns, so the amount and the description run together.
- A layout where the transaction amount and the running balance sit in the same visual column, so a parser takes one for the other.
- An export that writes debits with two minus signs.
- A CSV whose date format changes halfway through the year because the client upgraded their internet banking.
Each of those produces an out-of-balance that gets discovered on the 27th, when there is no time left to investigate it calmly.
The defence is simple and worth insisting on: check the imported rows against the statement's own running balance. The statement carries its own proof. If the closing balance implied by your imported transactions does not equal the closing balance printed on the statement, something did not come in — and you want to know that before you code anything, not after.
3. The one answer that unblocks four transactions
Every quarter has a handful of transactions nobody can code without asking. A $4,000 transfer that might be a loan, a drawing, or equipment. A recurring payment to a name that means nothing.
These are individually small and collectively expensive, because each one blocks a lodgement until it is answered, and each answer requires a human who does not work for you. Three of these across a client is a week.
What makes them worse than they need to be is that the question and the answer usually live in different places — the question in your working papers, the answer in an email thread. Six weeks later, nobody can reconstruct why a payment was treated the way it was.
4. Suspense, and everything parked in it
Anything you could not code went somewhere. Usually a suspense or holding account, occasionally a real account chosen under time pressure.
It has to come out before the BAS is right. G1 is built from coded transactions, so a transaction sitting in suspense is either missing from G1 or wrongly in it. Suspense is the most common late-stage discovery precisely because it is invisible until someone runs the report.
What follows from this
If most of the week is spent waiting, then the fixes that shorten the week are the ones that either remove a round trip or move a discovery earlier.
Remove a round trip. Anything that gets the statement in without a human writing an email is worth more than it looks. Per-client email addresses that a client can forward a statement to. Standing requests that go out on a schedule rather than when someone remembers. Requests that carry the specific question attached to the specific transaction, so the reply lands where it belongs.
Move a discovery earlier. A balance check at import time costs nothing and saves the 27th. A suspense report run at the start of the quarter rather than the end turns an emergency into an errand.
Do not buy speed you cannot use. Coding faster is genuinely valuable at volume — but if your week has four two-day stalls in it, halving the coding time saves you an afternoon and the week is still a week.
An honest way to measure it
Before changing anything, spend one quarter recording two numbers per client: the date you first asked for the statement, and the date you had everything you needed to start coding. The gap between them is your real cost, and it is usually the number nobody has.
Most practices are surprised. The coding is often a day. The gap is often three weeks.
Where software helps, and where it does not
ReconLink is built around the observation above rather than around coding speed.
It reads the PDF, CSV and Excel files Australian banks actually produce, and checks the imported rows against the statement's own running balance before you commit — so a format change surfaces at import, not at lodgement. It chases your clients for statements, receipts, and answers on specific transactions, and the replies come back attached to the transaction they belong to. It codes around 99% of lines before you look at them, and builds the BAS worksheet — G1 through G19, plus 1A and 1B — from the coded transactions.
What it does not do is decide. Every coded line shows the rule it used, so you can see why a line landed where it did. The judgement, and the signature on the lodgement, stays with the registered agent.
$99 a month, ex GST. Unlimited clients, unlimited seats. Thirty days free, no card.
Frequently asked questions
Why does a small client sometimes cost as much as a large one?
Because the dominant cost is chasing, and chasing is priced per round trip rather than per transaction. A client with 40 transactions who takes three requests to send a statement costs more than a client with 4,000 transactions who sends one on time.
What is the fastest single improvement to a BAS quarter?
Getting statements in without a manual request. It removes the longest and most repeated stall, and it is the only part of the week that scales linearly with the number of clients.
How do I catch a statement that imported incompletely?
Check the imported transactions against the closing balance printed on the statement itself. If the two do not agree, rows are missing. The statement carries its own proof, which is why this check is worth running before coding rather than after.
Should suspense accounts be cleared before or during BAS preparation?
Before. G1 is built from coded transactions, so anything sitting in suspense is either absent from the BAS or wrong in it. Running a suspense report early in the quarter turns a lodgement-week emergency into a routine task.
Does faster coding shorten the BAS week?
Only if coding is what your week is made of. If the week contains several multi-day stalls waiting on clients, faster coding saves hours while the stalls still cost days. Measure the gap between first request and data-complete before buying for speed.
The short version
The BAS week is not long because there are too many transactions. It is long because it keeps stopping.
Count the stalls before you buy anything. Then fix the two that remove a round trip and move a discovery earlier — those are the ones that give you the week back.