Most month-end closes are reconstructions
If your close takes two weeks, you are not closing the books. You are rebuilding the month. Where those two weeks go, what they cost, and what a close looks like when the data arrives verified.

A CFO told me her month-end close takes "close to two weeks". I asked what happens in those weeks.
She walked me through it. The revenue team sends the settlement data and finance pulls it into the system. It does not match what they expected. The Stripe settlement is lower than revenue, so they dig. Refunds have not settled yet. PSP fees are higher than modelled. One payment processor settles on Thursday, another on Friday. "By day 5, we think we understand it. By day 10, we’ve created adjustments, reversed transactions, reclassified entries. By day 14, the books balance and we sign."
I asked what would happen if the books were right from day one. "That’s impossible," she said. "The data doesn’t arrive clean."
That is the real question. Not how to close faster, but why the data does not arrive clean.
Two different jobs
A close is the moment you confirm the books are right. You review, you sign. A reconstruction is something else. You piece together what happened during the month, find the inconsistencies and correct the records by hand until they match reality.
Most companies think they do the first. In practice they do the second, and it follows a familiar rhythm.
- Days 1 to 2. Settlement files, processor reports and order data come in.
- Days 3 to 5. Differences surface. Revenue does not match payments, and settlements differ from the general ledger.
- Days 6 to 10. Manual journal entries, reversals, reclassifications and balancing bookings.
- Days 11 to 13. Every adjustment is documented and explained for the auditor.
- Day 14. Sign-off.
Where the two weeks go
None of this is accounting work in itself. Each step is a symptom of a setup that was not built for the data it receives. The same five causes come back every month.
- Timing. The PSP settles on one schedule, the order system records on another and the ledger expects a third.
- Fees. You expect the settlement at the gross amount, but Stripe pays out net of fees. Every month, a variance.
- Refunds. The order system records a refund on day 1 and Stripe settles it on day 4. They are expected to match, so someone adjusts by hand.
- Batches. The PSP pays out 50 transactions as one payout, while the ledger expects 50 entries.
- Volume. At high volume, the ERP cannot take in transactions fast enough. Some imports fail and have to be retried.
So if your close takes two weeks, your team is not slow. The way the data reaches the ledger is what makes it slow.
What it costs
A team of three to four people working two weeks on the close spends 240 to 320 hours a month on it. Over a year, that is 2,880 to 3,840 hours.
The hours are only part of the bill.
- Auditors see the manual adjustments and want to know why. They flag them as a control weakness, and investors and the board read that as a fragile close.
- Analysis waits. A CFO who signs on day 14 starts the real financial analysis on day 15, every month.
- Late errors are fixed under pressure. A large error found on day 12 needs a rushed correction. The same error found on day 1 is caught before anything downstream depends on it.
What a confirmation looks like
When the data reaching the ledger is already verified, the close changes shape. On day 1, the settlement report, the PSP reconciliation and the order data show the same picture. Finance reviews them, and the variances are small and already explained. On day 2, there is a final review and the books are signed.
The team is not faster. There is simply nothing left to hunt.
Interrail went through this shift. It processes two million orders a year, and its month-end checks took close to two weeks. In its own words: "Every system holds part of the truth. None of them hold the whole story." Then each order was matched to its payment, refund and settlement across all connected systems. Differences no longer disappeared inside aggregates. Month end became tighter and calmer, with fewer surprises. Interrail did not hire better accountants. The data arriving in its ledger was already verified.
A publicly listed betting company that processes millions of wagers a day saw the same. Month end became "tighter and more predictable, with far fewer surprises and less reliance on manual controls."
You pay either way
The usual objection is that fixing this takes months and costs a fortune. That depends on your setup. But you pay the cost either way. You pay it once, in a focused effort of three to four months to get the data clean upstream. Or you pay it every month, in two weeks of reconstruction, audit friction and decisions that wait.
Teams that made the shift tend to ask why they did not do it sooner. It was not cheap. It gave the team back the time it used to spend digging.
The Actuals scan takes 15 minutes and shows where your data first stops matching on its way to the ledger.
In practiceHow many days of your close go to explaining differences?


