The ledger and the close — where numbers stop being opinion

A trading company I worked with had a chief executive reading excellent profit figures off a spreadsheet and no cash in the bank to pay salaries. Both statements were true at once, which is the specific condition this cycle exists to prevent.

Three causes, and none of them was fraud. Sales closed orders and sent the paperwork to finance late, so invoices were raised weeks after delivery. Millions of dirhams of stock sat in the warehouse carried at no realistic value in the books. And operating expenses paid out of petty cash were recorded two months after the money left. The company was profitable on paper and insolvent in practice, and the gap between the two was entirely a recording problem.

The finance cycle, in the order it actually runs

1. Posting at sourceEvery operational transaction carries its journal with it, in the same second. Nobody remembers to do it, so nobody can forget.
2. ReconciliationBank statement against the books, and every sub-ledger against its control account. This is the only evidence the numbers are true rather than merely consistent.
3. Period closeDepreciation, accruals and prepayments run on a schedule — and the month locks, so a reported figure stops moving.
4. ReportingA profit and loss anyone can read on the day, and a cash forecast that names the shortfall before it arrives.

Closing time is the measure of all four. Cutting it comes from moving work out of the close, not from working harder inside it.

What changes when the ledger is automatic

In a spreadsheet company, the accountant is a data-entry role: he spends the day writing journals from paperwork that reaches him late from other departments. On a working ERP most of the recording happens without him, and his job becomes checking and explaining rather than typing.

The mechanism is simple and worth being precise about. Every operational transaction carries its accounting entry with it.

When the storekeeper confirms a goods receipt in the procurement cycle, the system posts inventory against supplier liability in the same instant. When goods leave against a delivery in order to cash, the system posts cost of goods sold against inventory, at whatever the costing method says that cost is. Nobody remembered to do it, so nobody can forget.

That is where the accuracy comes from. Not from a better accountant, but from removing the interval between the event and the record. An entry made in the same second as the transaction cannot be lost, backdated or argued about.

Two reconciliations that decide everything

Automatic postings make the numbers internally consistent. They do not make them true. Two checks do that.

Bank reconciliation. The system imports the bank statement and matches it against what the books say happened. Anything that does not match surfaces immediately: an uncleared cheque, a bank charge nobody recorded, a customer payment received against no known invoice. Where this runs daily rather than monthly, the unexplained items number in single figures. Where it runs quarterly, the reconciliation itself becomes a project.

Sub-ledger to control account. The total of all customer balances in the sales module must equal the receivables control account in the general ledger. Same for suppliers, same for inventory. If they differ, something is posting to the control account directly, which means somebody is writing manual journals into a balance that is supposed to be derived. That is worth finding, because it is usually where an error has been buried.

These two are not accounting ceremony. They are the only evidence that the system’s version of reality matches the outside world’s.

Closing the period

Month end stops being an ordeal once the recurring entries are the system’s job rather than a person’s.

Depreciation runs from the fixed asset register on a schedule, posting monthly without a spreadsheet. The register also tells you what you own, where it is, and what it is worth now, which most companies discover they could not previously answer.

Accruals and prepayments get spread automatically. An annual rent paid in January is allocated across twelve months rather than distorting January’s result. Without that, monthly profit is noise and nobody can tell a bad month from a payment cycle.

Period locking is the underrated one. Once a month is closed, the system refuses postings into it. That is what makes a reported figure stable — otherwise last month’s profit changes every time somebody backdates an invoice, and no report can be trusted twice.

From fifteen days to three

Closing time is the cleanest measure of finance function health, and it is a number every company already knows about itself.

I have taken companies from a fifteen-day close to a three-day close. The reduction did not come from working harder in the close. It came from moving work out of it: postings automated at source, reconciliations run continuously rather than at month end, and approvals that happen when the document is raised instead of in a review at day ten.

The value is not the twelve days of effort saved. It is that a decision made on day three of the following month is based on information that is still actionable, and a decision made on day eighteen is history.

The two design choices worth an argument

The chart of accounts. This is the skeleton, and it is designed once. Get it wrong and every report afterwards is wrong in the same way. The specific failure is a chart built for last year’s company: no dimension for branch, or project, or product line, so profitability by any of those is unanswerable without rebuilding. Design it for the structure you expect in three years, and use dimensions or analytic tags rather than proliferating account codes.

Approval thresholds. An ERP can enforce that a payment above a stated amount requires two electronic approvals, and that the person who created the supplier record cannot be the person who approves payment to it. Both are configuration, both take an afternoon, and both are the difference between a control that exists and a policy that is written down.

The line to hold onto

Profit is an opinion. Cash is a fact.

Reported profit depends on the costing method, on the depreciation policy, on when revenue is recognised, and on judgements about what is recoverable. Every one of those is defensible and every one of them moves the number. The bank balance does not move on judgement.

A good system does not flatter you with paper profit. It tells you where the money physically is — sitting with customers who have not paid, or stacked in a warehouse as goods that are not selling — which is covered in inventory and costing. Those two locations are where most companies’ cash actually is, and neither of them appears on a profit statement.

The four cycles read in order are on the learn ERP page, starting from what an ERP actually is. Which products handle the close, fixed assets and multi-entity consolidation themselves is set out in the systems comparison.