Every Friday afternoon, somebody in your business opens a spreadsheet and starts matching. On one side, a list of transfers that came in during the week. On the other, a list of invoices you sent out. The job is to work out which is which.
It takes two hours on a good week. On a bad week it takes the whole afternoon, and you still end up with four payments nobody can place and two customers insisting they have paid.
This is one of those problems that feels like a discipline issue and is actually a structure issue. You can hire a more careful person and still lose the afternoon. Here is why it happens, and how to make it stop.
Why matching breaks
When every customer pays into the same account, the only thing distinguishing one payment from another is the narration. And narration is written by the customer.
So you get "Payment", or "For goods", or a surname that does not match the business name on the invoice. Someone pays for three invoices in one transfer. Someone else splits one invoice across two days. A customer sends money from their brother's account. Somebody pays the exact amount of a different customer's invoice, and now you have two candidates for one payment.
None of this is unusual behaviour. It is just what happens when people pay. The structure asked the customer to label the payment correctly, and customers do not label payments correctly.
The fix is to stop relying on narration
Give each customer their own account number.
The number belongs to that customer and nobody else. When money arrives into it, you already know who sent it before you read a single line of narration. There is nothing to interpret, because the destination itself carries the identity.
This is the whole idea. It sounds almost too small to matter, and it removes most of the Friday job.
What changes in practice
The obvious win is that payments arrive pre-labelled. The less obvious wins are the ones people notice after a month.
You stop chasing customers who have already paid. Most of those awkward calls come from a payment you received but could not attribute, so it never got marked against the invoice. When attribution is automatic, the reminder never goes out.
Part payments stop being a problem. If a customer owes 400,000 and sends 150,000, it lands against their record and the balance updates. You are not trying to remember which spreadsheet row that partial belongs to.
Your ageing report becomes real. Most businesses have a debtors list that is quietly wrong, because it reflects what was matched rather than what was paid. Once matching is automatic, the list is worth acting on.
And month end shrinks. The record was built as the money moved, so nothing has to be reconstructed from receipts at the end of the period.
Setting it up
The work is mostly in the first hour. Roughly:
- Clean your customer list first. Merge the duplicates, fix the spellings, decide whether "Adebayo Stores" and "Adebayo Stores Ltd" are one customer or two. Do this before you issue numbers, not after.
- Issue an account number per customer. In Niimbu this sits under Account Issuing, and the number is generated against the customer record.
- Put the number on the invoice. This is the step people skip. If the customer cannot see their number on the document asking them to pay, they will use whatever account they used last time.
- Tell your existing customers once. A short message saying you have a new account number for them, with the number in it. Expect a few weeks of overlap where some still pay the old way.
- Leave the old account open during the changeover, and keep matching those manually until the traffic dries up.
What to watch for
A few things catch people out.
Customers who pay through a third party will still arrive unattributed, because the payment genuinely did not come from them. You will always have a small manual tail. The goal is to shrink it from most of your payments to a handful.
Walk-in and cash sales do not benefit from this at all. If a large share of your revenue is over the counter, dedicated numbers solve the transfer half of the problem and you still need a point of sale record for the rest.
And do not issue numbers to one-off customers who will never buy again. You will end up with a customer list full of records that exist because somebody bought once. Issue them to the accounts you expect to see repeatedly.
The point is not automation
It is worth being clear about what this actually buys you.
The saved afternoon is real, but the bigger thing is that your records stop being a reconstruction. When matching is manual, your books are an interpretation made days after the fact by somebody working from incomplete labels. Small errors get in, and they compound quietly until a stock count or an audit surfaces them.
When the record is built at the moment money moves, there is no interpretation step. What the system says happened is what happened.
That is the difference between books you maintain and books you can act on.
