Your month end close is a dependency chain, not a checklist

Why a flat checklist hides the problem

A list tells you what to do. It does not tell you what you are blocked on, and it does not tell you which of the thirty items will still be sitting there on day four.

Close items fall into three groups that behave completely differently. Some you can do yourself the moment the period ends. Some depend on a system finishing something, like a feed importing or payroll posting. And some depend on a human at the client who has no deadline pressure and did not know you were waiting.

That third group is where closes go late. It is also the group a checklist represents identically to the other two.

The three kinds of close item

Sort your existing checklist into these before automating anything. The sort itself usually shortens the close, because it surfaces how much of day four was avoidable.

• Yours, unblocked. Journal entries you already know about, recurring accruals, schedules that do not depend on anything new. These can be drafted before the period even closes.
• System-blocked. Bank feeds, payroll posting, merchant settlements. Predictable timing, no chasing required, but everything downstream is waiting on them.
• Human-blocked. The receipt nobody sent, the invoice nobody approved, the question about what a payment was for. Unpredictable, and the single largest cause of a late close.
• The tell. If your close reliably takes four days and two of them are waiting, you have a chasing problem, not a closing problem. Those need different fixes.

Chase earlier, not harder

The useful shift is moving the human-blocked items out of the close and into the weeks before it.

Nothing about that requires new software. It requires knowing, on day 25, exactly which items will block you on day 2, and asking then rather than discovering it later.

• 1. Map the chain once per client. Which steps block which. You only need to do this once, and most of it is the same across clients with the same stack.
• 2. Identify the human-blocked items specifically. These are the ones worth automating a reminder for. The rest do not need chasing at all.
• 3. Send the ask before the period ends. A single short list, to one person, with what you need and why the close waits on it. Repeated on a schedule so it is not a favour you keep asking for.
• 4. Draft the unblocked work early. Recurring entries and schedules do not need the period to be over. Doing them on day 28 shortens day 2.
• 5. Track where the close actually stalled. After a few months, the same two items will be responsible for most of the delay, on most clients. That is your real close problem, and it is usually two items rather than thirty.

What to automate and what to leave alone

The close is a place where over-automating is genuinely dangerous. A reconciliation that quietly reconciles itself is not a control, it is a hole.

Automate the coordination: knowing what is blocked, who owes what, what is unusual this month, and drafting the recurring entries you already understand. Leave the judgment calls, the unusual transactions and the sign-off exactly where they are.

The test is simple. If a mistake in that step would be caught later by someone reading the output, automating it is fine. If nobody downstream would catch it, do not.

Where this sits in a practice with several clients

Running one close is a task. Running four is a scheduling problem, and they all land in the same week.

The structure that survives is one where each client has the same shape: a standing record of how their close works, a place the current month's work lives, and an append-only log of what happened. Same folder names for every client, so you do not spend the first ten minutes of each close remembering where this one keeps things.

The consistency matters more than the specific structure. A file should have exactly one home for its whole life, and you should always know which of those three things you are looking at.

What this does not do

It does not close faster if the client's bookkeeping is behind. Nothing does. It surfaces that on day 25 instead of day 3, which is worth a lot but is not the same as fixing it.

It does not remove the review. It shortens the gathering so there is more time for the part that matters.

It does not pay off in month one. The chain map and the client-specific detail take a cycle to get right. Month two is where it starts giving time back, and it compounds from there.

Questions