- Predictability beats speed: a close that always takes six days is better than one that varies between three and twelve.
- Completeness of documentation is the leading indicator of close duration.
- Segregation of duties is a design requirement, not a software feature.
- The ledger remains the system of record; everything else is a feeder.
- Auditors want evidence of controls operating, not just controls existing.
What makes a close predictable
A close is a dependency chain. Each step waits on the one before, so a single late input pushes everything downstream. The steps most likely to be late are almost always documentation-related: missing receipts, uncoded transactions, unresolved review queues.
Improving predictability therefore means attacking completeness continuously rather than at month-end. A programme where receipt completeness sits above ninety-five per cent on the last day of the month closes on schedule almost every time.
Metrics worth tracking
Four numbers explain most of the variance in close duration.
| Metric | What it tells you | When to act |
|---|---|---|
| Receipt completeness at month-end | Whether documentation habits are working | Below the low nineties as a percentage |
| Transactions in review at cut-off | Whether review is keeping pace | If the queue is not near zero |
| Manual journal entries per close | Where the integration is failing | If the count grows month over month |
| Days to close | Overall process health | If variance widens rather than the average rising |
Segregation of duties in a small team
Textbook segregation assumes more people than most finance teams have. The workable version separates the two combinations that actually cause losses: creating a vendor and paying it, and approving your own spend.
Where headcount genuinely does not allow separation, compensating controls — review by someone outside finance, or a spending threshold requiring a second signature — are an acceptable and auditable substitute.
- Whoever adds a vendor should not be the person who releases the payment.
- Nobody approves their own expenses, including the finance lead.
- Bank detail changes require verification by an independent channel.
- Where separation is impossible, document the compensating control explicitly.
Reconciliation without the marathon
Reconciling a busy account monthly is a long, error-prone exercise. Reconciling it weekly is a short, boring one, and it surfaces problems while people still remember the transactions.
The same applies to card reconciliation. If the platform export agrees with the card statement every week, the month-end version is a formality.
Evidence auditors ask for
Auditors test whether controls operated during the period, not merely whether they were configured. That distinction determines which evidence is worth retaining.
- The policy as it stood during the period, with the date it took effect.
- Approval records showing who approved what, when, and against which threshold.
- Supporting documentation for sampled transactions, retrievable without heroics.
- A change log for limits, roles and permissions.
- Evidence that exceptions were reviewed and resolved rather than merely flagged.
System-of-record decisions
Every number should have exactly one authoritative home. Where two systems both claim a number, reconciliation becomes negotiation, and the answer to “what did we spend?” depends on who you ask.
Write the decision down: the general ledger owns financial statements; the spend platform owns transaction detail and documentation; payroll owns compensation. Then hold the boundary.
See the accounting reference for how the boundary is usually implemented.
Frequently asked questions
It varies widely with complexity, but predictability matters more than the absolute number. A stable six-day close is more valuable to everyone downstream than one that swings between three and twelve days.
For transaction detail and documentation, yes. For financial statements, no — that belongs to the general ledger. Keeping the boundary explicit avoids most reconciliation disputes.
Separate the two highest-risk combinations first: vendor creation from payment release, and self-approval of expenses. Where headcount prevents full separation, document the compensating control so auditors can assess it.
Usually mapping gaps: a category with no matching account, a cost centre that exists in one system and not the other, or tax treatments handled manually. Each one is fixable at the mapping level rather than every month.
Keep reading
Sources and further reading
Every factual statement on this page is checked against primary documentation. Terms change frequently, so confirm details with the provider before acting on them.
- FASB Accounting Standards Codification Reference point for accrual, expense recognition and close-process statements.
- IRS Publication 463 — travel, gift and car expenses Used for statements about expense substantiation and record keeping in the United States.
- Brex Support Center Official help documentation, including account access and card administration topics.
- Nacha — ACH Network rules and resources Used for statements about ACH timing and payment rails.