The Month End Close As A Batch Problem

Ask a finance team why the close takes as long as it does and you will get a list: late entries from the business, intercompany balances that do not agree, accruals that need judgement, systems that do not reconcile, an audit requirement, a person who is on leave. Every item on the list is true. None of them is the reason.

The reason is that you decided to do a month's worth of work in a batch, at the end of the month, all at once. Everything on the list is a consequence of that decision rather than an independent constraint. Late entries exist because there is a deadline to be late for. Reconciliations are difficult because a month of drift has accumulated before anyone looked. The dependency on individuals is acute because the work is compressed into a window too short to absorb an absence.

The close is, in most organisations, the single largest batch in the business, and it has all the properties that batch size theory predicts: a long queue in front of it, a spike of overtime inside it, a high defect rate discovered at the worst possible moment, and an outcome that arrives too late to influence the decisions it was meant to inform. Finance leaders who have accepted a five, eight or twelve day close as a fact of nature are accepting the consequences of a policy that someone chose, usually before the systems they now run existed.

What a batch actually costs

The economics of batching are a trade-off between two costs, and the trade-off is well characterised.

Transaction cost is the fixed overhead of processing a batch, regardless of its size — the setup, the coordination, the checklist, the sign-off meeting. High transaction cost pushes you toward large batches, because you want to amortise the overhead.

Holding cost is what you pay for work sitting undone while the batch fills. For the close, holding cost is the value of decisions made on stale information, the cost of errors discovered a month after they occurred, and the cost of finance staff unavailable for anything else during the close window.

Most finance functions have optimised for transaction cost alone, and for a good historical reason: when closing required physical ledgers, manual journals and a mainframe run, the transaction cost genuinely was enormous. Monthly batching was the right answer to a real constraint. That constraint has largely dissolved, and the batch size has not moved, because nobody revisits a cadence once it has become a calendar.

The characteristic signature of an over-large batch is visible in the close itself. Work arrives in a flood rather than a flow. Defect discovery is concentrated at the end, when correction is most expensive and there is least time. Everyone is at full utilisation for a few days and materially underloaded for others. Queue length in front of the batch grows all month, invisibly, because nothing in the reporting counts unreconciled items as inventory.

Where the days actually go

Do not accept the checklist as a description of the process. Map it as a value stream and split every interval into touch and wait.

Take the last close. For each material workstream — bank reconciliation, intercompany, accruals and prepayments, revenue cut-off, payroll, fixed assets, consolidation, management commentary — record when it could have started, when it did start, when it finished, and how many hours of effort it contained. You will find three patterns with near-certainty.

Serialisation that is not required. A great many close tasks are run in sequence because that is the order in the checklist, not because the second depends on the first. Every unnecessary serialisation adds a full waiting period to the critical path.

Waiting on the business, not on finance. A large portion of the close is finance waiting for someone outside finance: an approval, a cost centre owner confirming an accrual, a submission from a subsidiary. Finance is held accountable for a duration whose largest component it does not control, which is also why exhorting the finance team produces so little.

A genuine critical path that is shorter than the close. Total the longest dependent chain of actual work. In most organisations it is a small fraction of the elapsed close, which means the close duration is set by queueing and coordination rather than by the volume of accounting to be done.

That last finding is the one worth presenting to an audit committee, because it reframes the close from a capacity question into a scheduling question.

Moving work upstream

Continuous accounting is the name usually attached to the remedy, and the label matters less than the mechanism. The mechanism is to stop treating the period end as the trigger for work that could have been done when the underlying event occurred.

Reconcile continuously rather than at period end. A bank reconciliation performed daily or weekly handles a small, recent, comprehensible set of differences. The same reconciliation performed monthly handles a large set of differences, many of them old, several of them interacting. The work is not merely deferred by batching; it is made harder, because investigating a three-week-old discrepancy costs far more than investigating a one-day-old one. Feedback delay raises the cost of every defect it delays.

Push journal entries to the point of the transaction. Recurring accruals, depreciation, amortisation and standard allocations do not require period-end judgement. Anything mechanical should be automated and scheduled away from the window entirely.

Move approvals before the deadline rather than after it. Cost centre confirmations, accrual sign-offs and variance explanations can be collected during the period, on a rolling basis, when the owner still remembers the transaction and is not being chased during a four-day crunch.

Make period-end cut-off real and early. A soft cut-off produces late entries, and late entries are what force the re-running of downstream steps. Soft deadlines are not kindness; they are a transfer of cost from the business to finance, paid in rework.

Report a preliminary position early and refine it. Many management decisions need directional accuracy on day two far more than they need precision on day ten. Publishing a flash position and then a final one splits a single large batch into two smaller ones, which is exactly what batch size reduction looks like in practice.

Reconciliation is rework

It is worth naming this plainly, because it changes where the improvement effort goes.

A reconciliation difference is a defect. Something was recorded wrongly, recorded twice, recorded late, mapped to the wrong account, or not recorded at all. The reconciliation is the inspection step that finds it and the investigation that corrects it. Finance spends a substantial share of the close doing work that exists only because an upstream process produced something incorrect — what John Seddon would call failure demand, and what Deming spent a career arguing should be designed out at source rather than inspected out at the end.

Treat it accordingly. For one close, categorise every reconciling item by its originating cause: a coding error at entry, a timing difference, a system interface failure, a master data problem, an unapproved transaction. Count them by category and by originating function.

The distribution is almost always concentrated. A small number of causes produce most of the differences, and those causes usually sit outside finance — in a procurement process that allows purchase orders without correct coding, in a sales system whose interface silently drops records, in a master data set nobody owns. Each of these is fixable at source, permanently, and each fix removes a recurring block of close effort rather than making it faster.

This is the difference between improving the close and shrinking it. Better checklists, more automation of the reconciliation itself, and additional temporary resource all make the inspection faster. Fixing the upstream cause removes the inspection.

The control and audit objections

Two objections arrive reliably, and both are answerable — but only if you answer the real version rather than the caricature.

"The auditors require this." Auditors require evidence that controls were designed appropriately and operated effectively throughout the period. They do not require that the control operate at month end, in a batch, on a spreadsheet, signed by a specific person. A control that operates continuously and leaves a timestamped, reviewable trail is generally stronger evidence than one performed once a month under time pressure, because it demonstrates operation throughout the period rather than at a point in it.

"Moving faster increases risk of error." The opposite is usually true, and the mechanism is batch size. Large batches concentrate defect discovery at the end, where there is least time to investigate and most pressure to book a plug. Smaller, more frequent operation means fewer items per cycle, shorter feedback loops, and errors caught while the context is fresh. The risky close is the compressed one performed by tired people against a hard deadline.

There is a legitimate version of the concern, and it should be respected: any change to the timing, ownership or evidence of a control needs to be documented, agreed with internal audit, and walked through with the external auditor before the period in which it takes effect, not discovered by them afterwards. Do that work early and in writing. Sequencing the conversation properly converts an objection into a project.

Objection	The real requirement	How it is satisfied
Controls must be performed at close	Controls operate effectively across the period	Continuous operation with a timestamped trail
Segregation of duties	Preparer and reviewer are distinct	Unchanged by moving the work earlier
Auditable evidence	Evidence is complete and retrievable	Systematic logging rather than period-end packs
Management review	Review is substantive, not perfunctory	More time for review, since preparation is spread

What shortening the close is worth

Build the case on decision latency rather than on finance effort, because the finance effort saved is real but comparatively small and easily dismissed as a headcount argument.

If management numbers land on the tenth working day, the business has spent roughly half a month operating on the previous month's understanding, and the decisions taken in that window — pricing, spend, hiring, inventory, collections — were taken blind. Bring the numbers to the third working day and you have not saved seven days of accounting; you have given every operational decision-maker in the business seven extra days of sight in every month of the year. That compounds, and it is the argument that persuades a chief executive.

Three further benefits are worth stating explicitly. Finance capacity is released from a compressed window into usable analytical time. Error correction becomes cheap because it happens near the event. And the dependence on a handful of individuals who know where the awkward bits are — a real operational risk in most finance functions — reduces as the work becomes routine rather than heroic.

What to do on Monday

Take the last close and build a timeline of every workstream with start, finish and effort hours. Compute flow efficiency for the close as a whole. Expect a number that makes the case for you.

Identify the longest dependent chain of genuine work. If it is materially shorter than your close duration, you have a scheduling problem rather than a resourcing problem, and you should say so in those words.

Categorise every reconciling item from that close by originating cause and originating function. Take the top two causes to the functions that own them with the counts attached.

Pick one reconciliation currently performed monthly and move it to weekly for a quarter. Measure the time it takes per cycle and in aggregate, and measure how many differences are found. Both numbers will make the batch argument better than any explanation.

Then schedule the conversation with internal audit and your external auditor about control timing and evidence, before you need it. That conversation is the long lead-time item in this entire programme, and starting it in month one is the difference between shortening the close this year and arguing about it for two.