Accounting
Closing the month in days instead of weeks
A long close is rarely a staffing problem. It is a list of things that were allowed to go unresolved during the month, collected and resolved at the end of it.
The Rocket ERP team
Rocket ERP3 min read
Ask a finance team why the close takes three weeks and the answer is usually a list of waiting: waiting for supplier invoices, waiting for a stock count, waiting for someone to explain a difference. Almost none of it is accounting work. It is the accumulated cost of ambiguity that was tolerated for thirty days.
Close continuously, not at the end
The structural fix is to move each resolution to the moment the ambiguity appears. A goods receipt with no invoice should be visible as a known gap on the day it happens, not discovered on the 3rd of the following month. A cash difference should be investigated the evening it occurs, while the person who handled the till is still in the building and still remembers.
The practical version of this is a short daily or weekly check of a handful of exception lists, rather than one long reconciliation. The work is the same size; it is simply done while it is still cheap.
The six lists worth watching
Documents not posted. Anything saved but not reflected in the ledger. This list should normally be empty, and anything in it is tomorrow mystery.
Goods received with no invoice, and invoices with no goods. Both are real and both need an accrual; what you cannot do is discover them at the end.
Items with negative stock. Each one is a costing error with a date attached.
Unallocated receipts and payments. Money that arrived against no invoice becomes a customer balance nobody can explain.
Documents with no branch or no cost centre. These vanish from the reports that filter by those dimensions, so they distort the comparison quietly.
Suspense and clearing accounts with a balance. A clearing account is supposed to clear; a balance there is work in progress by definition.
Each list is a question with an owner and a due date. Together they are most of the close.
Lock the period, and mean it
A close that can be reopened silently is not a close. The figures you reported have to stay reportable, which means that after the period is locked, a correction is a new entry in the open period rather than an edit to the closed one.
This sounds restrictive and is the opposite. Once nobody can change history, the question "has this number changed since I reported it" stops needing to be asked, and the time spent re-checking previously agreed figures disappears from the close.
Automate the arithmetic, not the judgement
Depreciation, recurring accruals and standard allocations are rules and should run as rules. What should not be automated is the judgement: deciding whether an unexplained difference is an error or a genuine loss is not arithmetic, and a system that posts the difference automatically has destroyed the information that would have let someone find the cause.
A reasonable target
For most trading and services businesses, the ambition is a close within the first week, with a trial balance nobody argues about and a stock figure that matches the ledger without an adjusting entry. Getting there is almost never about working faster at the end of the month. It is about allowing less to be unresolved during it.