The month ends on the thirty-first and the books are not closed until well into the next month. Nobody is idle. The team is posting, chasing, matching and re-checking, and still the owner asks on day six when the numbers will be ready.
Most of the entries are quick once you know the figure. The days go on finding the figure: the difference nobody expected, the bill nobody entered, the answer from another team. Month end is slow because of what is found late.
Where the time goes.
Five causes show up in almost every small finance team. You will probably recognise most of them from last month.
- Accounts reconciled only at month end.If the bank and card accounts are matched once a month, a whole month of differences arrives together. A timing item, a duplicate and a missed fee are all in the same pile, and each one has to be traced before the next step can start.
- Supplier bills that reached an inbox, not the accounting system.A bill sent to one person, forwarded once and then forgotten is an expense that does not exist yet. It turns up on day four when someone searches their email, and the numbers you already reviewed move.
- Accruals estimated from memory.Which jobs are finished but not billed? Which deliveries arrived without an invoice? When the answer is a recollection, the amount is argued about instead of checked, and the argument takes longer than the arithmetic.
- The checklist rebuilt every month.Someone opens last month's spreadsheet, saves it under a new name, deletes the ticks and edits what they remember changed. Steps get dropped without anyone deciding to drop them, and the order is whatever the last person left.
- Questions for other teams asked on day five.Is this payment a deposit or a sale? Did sales agree that credit? The people who know are busy with their own month end, so a question asked late waits, and the close waits with it.
None of these is a failure of effort. Each one is work that could have been done earlier, in smaller pieces, if someone had the time to prepare it.
One close, day by day.
Take a 40-person firm that supplies and installs commercial flooring, with a controller and one accounts clerk. Follow one close.
- Day one: the bank.The clerk exports the bank and card activity and the open ledger items. The controller looks at the list of unmatched lines and decides which are timing items and which need an entry.
- Day two: the bills.The clerk collects supplier bills from three inboxes and a shared folder, and checks each against the accounting system. Seven are missing. The controller decides which belong to this month and approves their entry.
- Day three: payroll.The payroll register is compared with the payroll lines in the ledger. One department differs. The controller decides whether it is a coding error or a real difference, and the clerk posts the correction the controller approves.
- Day four: accruals.A list of delivered goods with no matching bill, and finished jobs not yet invoiced, goes to the project manager the same morning. The controller reads the replies, sets each amount and signs off the entries.
- Day five: the review.The controller reads the budget against actual figures and a draft note on each large difference, then rewrites the notes in their own words before sending them on.
The controller decided every posting and signed the close. What changed is when the lists existed. The differences were already sorted by the time the controller sat down.
What changes when preparation comes first.
This is where the AI you already pay for, whether ChatGPT, Claude, Copilot or Gemini, has a small and plain job. It prepares. It does not post, approve or sign anything. A person supplies the exports and a person decides what to do with the result.
The preparation is lists, comparisons and drafts, built from records your team exports:
- A checklist for this close.The steps, their order and the owners, built from your own list of what has to happen, with last month's open items carried forward. The month-end close checklist is one starting point.
- A reconciliation list.Bank and card lines compared with the ledger, with matches, likely timing items and unmatched lines separated. The bank reconciliation preparation template shows the shape.
- A payroll comparison.The register against the ledger, department by department, with the differences listed. See payroll register to ledger reconciliation.
- Draft commentary.A first note on each large movement between budget and actual, drawn from the figures you supply and marked where the reason is unknown. See budget versus actual variance commentary.
These four are free workflow templates. They are samples and a starting point, written to be changed to fit your accounts, your exports and your way of closing. A workflow built from one of them runs on the AI platform your team already uses.
The controller still decides. A person checks every comparison against the source records, chooses every posting and signs the close. The workflow does not touch your accounting system, and it has no authority over a number.
It also changes when the questions go out. A list of open items can reach other teams on day one, while the month is fresh, instead of on day five.
Where to start.
Look at last month's close and write down, for each day, what you were waiting for. That list is your map. The page on month-end close sets out the work, the records it draws on and what a prepared close looks like.
If you want help turning that map into a plan for your finance team, the first step is below. This is also what 10X the ROI of your existing AI looks like in practice: the tools you pay for, doing the preparation so your people can do the deciding.