Ask a finance team why the management accounts take three weeks and the answer is usually that there is a lot to do. Watch the close and that turns out to be false. There is not much more work than in a team that closes in five days. The work is arranged differently.
A long close is mostly waiting, and waiting is fixable.
Stop treating the close as a single block
The default mental model is that the month ends and then the close begins. That guarantees everything is sequential and everything competes for the same few days.
Most of what happens in a close does not need the month to have ended. Fixed asset additions can be processed as they occur. Prepayment schedules can be rolled forward mid-month. Payroll is known before month end. Recurring journals can be prepared and dated in advance. Intercompany balances can be agreed on the twenty-fifth and adjusted for the last few days.
Move everything that can be done before the month ends into the month, and the close becomes the genuinely period-dependent work rather than all of it.
Set a cut-off and defend it
The single biggest cause of a long close is waiting for supplier invoices that have not arrived.
Set a date, two or three working days after month end, after which nothing new is accepted into the period. Anything not received by then gets accrued from the purchase order or from a reasonable estimate, and the difference washes out next month.
This feels wrong to people trained to wait for the document. It is the right trade. An accrual estimated within a few per cent and booked on day three is worth considerably more than an exact figure on day nineteen, because only one of them arrives while the information still matters.
Set a threshold under which you always estimate, publish it, and stop debating it monthly.
Reconcile monthly, properly
Balance sheet reconciliations done every month, reviewed by someone other than the preparer, are the foundation of both a fast close and a cheap audit.
A team that reconciles monthly is confirming work already done. A team that reconciles at year end is reconstructing twelve months under deadline, and will find errors from February in January, by which point the explanation is gone and the correction is an audit adjustment.
The review matters as much as the preparation. A reconciliation nobody checked is a schedule, not a control, and your auditor will treat it as one.
Kill the things nobody reads
Most management packs have accumulated pages. A report someone asked for in 2022, a schedule that supports a decision no longer being made, a variance analysis down to accounts that never vary.
Ask who uses each page and what they do with it. Anything without an answer comes out. This is faster to do than it sounds and usually removes a surprising amount of work from the critical path.
What remains should answer three questions: what happened, why it differs from what we expected, and what we are doing about it. Everything else is appendix.
Write the timetable down
A close that lives in people’s heads cannot be improved and cannot survive someone being away.
List every task, who owns it, which day it happens, and what it depends on. The dependency column is the valuable one, because it shows you the critical path and the critical path is the only thing worth optimising. Shortening a task that was never holding anything up achieves nothing.
Once it is written down, the close also becomes something a new joiner can run, which is a resilience benefit on top of the speed one.
Expect it to take a few cycles
Five days is not reached in one month. The usual path is three weeks to two, two to ten days, then ten to five as the pre-month-end work and the cut-off discipline embed.
Each cycle, pick the one thing that held you up longest and fix only that. A team trying to fix eight things at once fixes none and concludes the close cannot be improved.
The end state is worth the effort. Numbers on working day five are management information. The same numbers on day twenty are a record of something you can no longer do anything about.