How to Build Checklists for Work You Repeat Monthly
I have sent the same client the wrong invoice twice, fourteen months apart, for the same reason. The invoice itself was fine.

I have sent the same client the wrong invoice twice, fourteen months apart, for the same reason. The invoice itself was fine. What I missed both times was attaching the timesheet export, which that particular client requires before their system will accept anything, which means the invoice sits in a queue for three weeks and then someone emails me about it. Both times I knew about the requirement. Both times I was doing the month end run from memory at half past five on a Friday.
Monthly work is where this happens. Daily tasks become automatic and yearly tasks are so obviously unfamiliar that you look up what you did last time. The monthly ones sit in the gap: frequent enough that you feel you should know them, rare enough that you genuinely do not. Four weeks is long enough to forget the order, the one odd exception, and the step that only matters in certain quarters.
Four weeks is past the point of recall
There is a straightforward reason this feels harder than it should. A procedure you perform daily gets consolidated into something you execute without attention. A procedure you perform monthly never gets there. Each run is a partial reconstruction, and reconstruction is where steps go missing.
The symptom to watch for is a specific kind of hesitation. If you open the folder and spend two minutes working out where you were last time, that is not a memory failure, it is a missing artefact. You are doing unpaid design work every month on a process you already designed.
The other symptom is error clustering. My invoice mistakes were both the attachment, never the amount, never the address. Repeat work fails at the same joints every time, usually the handovers between tools, and that predictability is exactly what makes a written list worth the trouble.
Read and do, not do and confirm
Aviation distinguishes two kinds of checklist and the distinction is the most useful thing I have taken from that field. A read and do list is worked through line by line as you perform the task. A do and confirm list is for things you already know by heart, where you do the whole sequence from memory and then verify it against the list afterwards.
Almost everyone writes do and confirm lists and then uses them as read and do, or the reverse, and both mismatches fail. A do and confirm list is short and names only the items that are consequential if missed. A read and do list can be long and includes the boring mechanical steps, because you are reading it anyway.
For monthly work, you want read and do for anything that touches money, external parties or anything irreversible, and do and confirm for the parts you genuinely have internalised. My month end is now split exactly that way: a nine line read and do list for the invoicing, and a four item confirmation at the end for the admin I do quickly.
This is also where the usual advice goes wrong. The standard recommendation is to write down everything, document the whole process, be thorough. I did that in 2023. The result was a 41 step document that I read fully twice and then abandoned, because reading 41 steps to do a 25 minute job is intolerable and once you skip part of a list you stop trusting all of it. A checklist with 40 items is not a checklist, it is a manual, and manuals are for training someone new rather than for running a task.
The working ceiling in my experience is somewhere around nine items per list. Beyond that, split it into two lists at a natural boundary, or push the detail into a separate reference document the checklist points at.
Writing a step that cannot be fudged
Most useless checklist items share one trait: you can tick them while thinking about something else. Check the invoices. Review the numbers. Make sure everything is correct. None of those have a defined end, so your brain supplies one, which is the same reconstruction problem the list was supposed to remove.
- Name the artefact, not the activity. Not review the export, but open the CSV and confirm the row count matches the dashboard total.
- Give each step a visible end state. Something you can look at and say yes or no about. A number matching another number. A file present in a folder. An email in the sent items.
- Put the exception in the step it belongs to, not in a notes section. My list says attach the timesheet export for the one client that requires it, with the client named, inside the step about sending. Exceptions filed separately get read once.
- Write it in the order you actually work, including the inefficient order. If you always do the awkward reconciliation first because the data is not ready otherwise, write that. Checklists that describe an idealised sequence get silently reordered and then partially skipped.
- Keep one step per line and no conjunctions. Any step containing the word and is two steps, and the second one is the one that gets missed.
- Date the list and note what changed. One line at the bottom. When a step stops making sense in eight months you want to know when it arrived and why.
That last point does more work than it looks like it should. Half the steps on my lists exist because of a specific incident, and knowing which incident is the difference between removing a step confidently and keeping it out of superstition.
Where to keep it, and why not in a checklist app
There are applications built specifically for recurring checklists and they do the job. I have used two and gone back to something simpler both times, for reasons that are less about the software than about where my attention is on the day.
A checklist you have to remember to open is a checklist you will skip during the week you most need it. The best place for one is wherever you will already be looking at the moment the work starts. For me that is the calendar: the monthly block for invoicing has the nine lines pasted into the event body, so the list arrives with the reminder. No app to open, no sync question, no second place to maintain.
The second workable option is a template note that you duplicate per run. You get a dated record of each month, which matters more than you expect, because the record of what you ticked last month is the fastest way to answer what happened. If your notes are already organised, this costs nothing to add, and the filing side of it follows the same logic as organising a notes app so you find things.
What I would avoid is keeping the canonical copy in a shared document that three people edit, unless the work is genuinely shared. Lists maintained by committee grow, because adding a step is easy and removing one requires an argument, and growth is the failure mode here.
A checklist and a text snippet solve adjacent problems and people conflate them. Snippets handle the words you retype; the piece on your first fifteen snippets covers that side. A checklist handles the order and the completeness of a procedure. The invoice email body is a snippet. Remembering the attachment is a checklist item.
Build it from failures, not from imagination
Do not sit down and design the list in advance. You will produce something plausible that misses the actual failure points, because the failure points are specific and slightly embarrassing and you do not think of them when nothing has gone wrong.
Instead, run the task once more the way you normally do, and keep a note open beside it. Every time you hesitate, backtrack, or look something up, write one line. At the end you will have five or six lines, and that is your first version. Mine took two months of this to stabilise and has not changed much since, except for two additions after specific mistakes.
Then add one rule: when something goes wrong in a repeat task, the fix is a line on the list before you do anything else about it. Not a resolution to be more careful. A line. That single habit is what turns the list from a document into something that improves, and it slots neatly into whatever you already do as a weekly review.
If you want somewhere to start this week, pick the monthly task you most dread and write the five things you always have to look up. That list will be useful on its first run, which is unusual for anything in tools.
FAQ
How is this different from a recurring task in my to-do app?
A recurring task tells you when. A checklist tells you what and in which order. The two work together, and the mistake is trying to make the to-do app hold the procedure, which produces eleven subtasks that reappear every month and get ticked in bulk.
Should I write checklists for daily work too?
Rarely worth it. Daily work gets internalised, and a list you have read two hundred times stops registering. The exception is anything irreversible, like a deployment or a payment run, where a short confirmation list earns its place no matter how often you do it.
What do I do when a step no longer applies?
Delete it on the spot, during the run, while you remember why. Steps that linger past their usefulness are the main reason people stop trusting their own lists, and a list you half believe is worse than none.
How long should this take to set up?
About twenty minutes of actual writing, spread over two or three runs of the task. The version you write in one sitting without doing the work alongside it will be wrong in the specific places that matter.
More in Productivity Tools
Productivity Tools
Why a Clipboard Manager Beats Another To-Do App
I tracked my own copy and paste actions for a week once, out of suspicion rather than curiosity, using a counter I bolted onto...
Productivity Tools
Website Blocker for Focus and How to Set One Up
You sit down to write a report, and twenty minutes later you have read three news articles, checked a forum twice, and written...
Productivity Tools
How to Use a Read Later App So Saved Articles Get Read
Your read later app has 400 saved articles in it, and you have opened maybe a dozen.