Use whenThe afternoon before the first day of a monthly close, a quarter-end, an audit fieldwork week, or any recurring period where a backlog is produced by other people's schedules.

Month-End Close Reset

A close period fails in a predictable way: day one is spent discovering that three things nobody owns have not been done, and by day three the whole week is gone. This protocol runs in the eighteen minutes before the first day of a close, and its output is a short written list rather than a feeling of readiness. It is deliberately unglamorous.

Work rhythm · 18 min · Published 2 October 2026 · By The Calm Desk

The sequence

1. 01

   List every task as a verb and an owner

   300 s

   Five minutes writing every close task as a verb plus a name: "reconcile the two freight accounts — Priya". Anything that cannot be given a verb or a name is not a task, it is a worry, and it goes on a separate line at the bottom.

2. 02

   Find the three that have no owner

   240 s

   Look at the list for blanks. There are usually two or three tasks that everyone assumes someone else does, and those are the ones that will appear on day one as an emergency. Assign them now, including if the name is yours.

3. 03

   Write the dependency, not the deadline

   240 s

   For each task, note what it cannot start until — a file from another team, a bank statement, a system that closes on day two. Dependencies are what turn a close from a schedule into a queue, and knowing the queue in advance is the whole value of doing this early.

4. 04

   Block the two hours that will otherwise vanish

   180 s

   Put two hours of uninterrupted time in the calendar for the hardest task, and decline the recurring meetings that fall inside it. If you do not do this now, the close will be done in the gaps between other people's meetings, which is how closes run to day five.

5. 05

   One physical thing for the first morning

   120 s

   Two minutes setting up one thing that removes a decision from tomorrow: the folder open, the reconciliation template saved with the date in the name, the printer loaded. Small, and it is the step people skip.

Unowned work is what breaks a close

Closing a period rarely fails because a task was hard. It fails because a task was nobody's until the moment it became urgent. The exercise of writing every task as a verb and a name, then hunting for the blanks, is designed to surface exactly those items forty-eight hours before they would have surfaced on their own. The blank line at the bottom of the page — the worries that could not be made into verbs — is usually the most useful part of the list and should be read out to someone.

Dependencies turn a schedule into a queue

A to-do list implies that everything can be started at once. A close cannot: half the tasks are waiting for a statement, a system, or another team's file. Writing the dependency beside each task converts the plan from a schedule into a queue, and queues make the real constraint visible — usually one late-arriving input that four downstream tasks are all waiting on. Once you can see that, one phone call on day one can recover a day.

Blocking time before the period starts

Calendars fill from the outside in: everyone else's recurring meetings are booked first, and the close work is left to occupy the gaps. Blocking two hours before the period starts works because it changes the order of operations — your block exists before the invitations go out, so declining them is a normal thing to do rather than an escalation. It also produces a visible signal that close work is scheduled and protected.

Environment tweaks

• Printed list rather than a digital one for day one; a page on the desk cannot be closed behind another window.
• Notifications off for the two blocked hours, not just on do-not-disturb — close the mail client rather than muting it.
• One status message to the team the evening before, so the first morning is not spent answering "where are we".

Questions people ask

My close is not month-end. Does this still apply?

The pattern applies to any recurring period where most of the work depends on inputs that arrive on other people's schedules — quarterly filings, audit fieldwork, payroll runs, weekly reporting. The trigger changes; the four outputs — verb-plus-owner list, unowned blanks, dependency column, two blocked hours — do not.

Eighteen minutes seems short for planning a close.

It is short on purpose. Longer planning sessions for a recurring process tend to produce a better document and a later start, and the document is not the point — the assignments are. Eighteen minutes reliably produces a list with names on it; a half-day workshop reliably produces a deck. If the close includes something genuinely new this period, that new item deserves its own meeting, separately.

What if I am not the person who can assign the unowned work?

Then your output is a short list for the person who can, sent before day one: three items, no narrative, with a proposed owner for each. That is a much easier message to act on than a day-one escalation, and it puts you in the position of having predicted it rather than reported it.

Informational only. This is a personal planning routine, not accounting, legal, or compliance advice, and it does not describe any specific reporting standard.