Month end close checklist

Last reviewed 2026-09-18

Most published close checklists are ordered by department. Ordered by dependency instead, the list is shorter and the reason each step exists is visible — and the days the close actually loses are nearly always spent redoing a step whose input changed underneath it.

Before the period end

• Confirm the period calendar: start date, end date, and which fiscal period this is. Retail and 13-period calendars make “September” ambiguous.
• Agree cut-off with anyone who can post: last day for expenses, last payment run, last goods receipt.
• Get the rate table loaded, with rate type and rate date for each rate. A close that stops on day three waiting for a rate stopped on day one and nobody noticed.
• Send intercompany balance confirmations early — they depend on somebody else's week.

Days 1–2: get the data in, whole

• Close the subledgers: payables, receivables, payroll, fixed assets, inventory.
• Load every external file — bank statements, card feeds, payment processors, the payroll bureau's register.
• Prove each file arrived whole, using a control total from outside the file. A record count the sender reports separately, a statement closing balance, a batch header. A file whose own debits equal its own credits proves nothing about completeness, and a half-loaded population is the one failure that reconciliation cannot detect afterwards — because everything downstream ties against the rows that arrived.
• Reject a file that does not tie in full rather than importing the part that parsed.

Days 2–4: make the ledger correct

• Accruals and prepayments, each with a schedule and a basis per line.
• Depreciation and amortisation.
• Revenue cut-off: what was recognised, which is not what was invoiced.
• Provisions and allowances, with the judgement recorded rather than the number alone.
• Translation of foreign entities, at recorded rates with recorded dates.
• Every control account tied to its subledger. Not a reconciling item — an integrity check that must pass before anything downstream is meaningful.
• Every clearing and suspense account cleared to zero, or the residual explained by transaction.

Days 3–5: reconcile and substantiate

• Bank reconciliations, gross debits against gross debits, gross credits against gross credits.
• Balance sheet substantiation for every material account, with an evidence reference per line.
• Intercompany agreed in both entity currencies, gross in both directions, before translation.
• Residuals split into rounding, unmatched and unexplained — three figures, not one. Only the third is an investigation.
• Every unexplained item given an owner, a required-evidence list and an expected date before the ledger is locked.

Days 5–7: explain and report

• Check period comparability before running any variance: unequal spans produce movements nobody caused.
• Decompose the movements into drivers that sum exactly to the total, with the unattributed amount shown as unresolved rather than absorbed.
• Write commentary that cites the underlying records, and says plainly where a conclusion is blocked and by what.
• Review, with preparer and reviewer named and dated.
• Lock the period and record who locked it.

What to automate first

Ordered by the ratio of time consumed to judgement required:

1. File ingestion and control totals. Purely mechanical, and it is the step whose failure is invisible later.
2. Transaction matching. High volume, rule-based, and the rules can be improved by looking at what they missed.
3. Recurring journals. Straight-line amortisation and standing accruals.
4. Variance decomposition. Arithmetic, provided the residual is reported rather than absorbed.
5. Reconciliation status reporting. Who owes what, and how old it is.

What should not be automated: establishing the cause of an unexplained movement, and concluding that a judgement is reasonable. Both are claims about the business rather than about the data, and a system that generates them produces a close that looks finished earlier without being more complete.

Two habits that shorten a close more than tooling

• A hard cut-off actually enforced. Most re-work is caused by a posting arriving after a downstream step consumed the balance.
• Carrying the unexplained forward explicitly. A named open item with an owner costs one line in the pack. The same item unnamed costs the first two days of the next close, every time.

Common questions

How long should a month end close take?

Five to seven working days is typical for a mid-sized group, and three is achievable with clean subledgers and an enforced cut-off. The number matters less than whether the last two days are spent explaining results or still finding data.

What should be on a month end close checklist that usually is not?

A completeness control for every imported file, taken from outside the file itself; a comparability check on the period before any variance is run; and a split of reconciliation residuals into rounding, unmatched and unexplained rather than one difference figure.

Which close tasks are worth automating?

File ingestion with control totals, transaction matching, recurring journals, variance decomposition and status reporting. Establishing the cause of an unexplained movement is not one of them — that is a claim about the business, and it needs a person’s name against it.

About Valcenra

Valcenra decomposes a financial movement into drivers that tie to the underlying records, keeps rounding, unmatched and unexplained residuals apart rather than summing them, and refuses to state a conclusion it cannot support — naming the field it needs and the conclusion that field decides. It is read-only: it drafts and computes, and it does not post journals, move money or approve anything. Talk to us.