Frequency is the wrong sole criterion
The usual rule — automate what happens daily — is a decent heuristic and it excludes a category of work that genuinely deserves attention: infrequent, concentrated, deadline-bound tasks.
Three days of work once a month is 36 days a year, which is more than a task taking twenty minutes daily.
Three questions that decide it
- Is the effort concentrated? Three days in a block disrupts more than the same hours spread out.
- Is the deadline hard? Statutory filings and payroll do not move, so the risk of being late has a cost attached.
- Is an error expensive? Month-end errors propagate into decisions and are found late.
Two yeses and it is worth pricing, however infrequent.
The monthly work that usually qualifies
- Month-end close and reconciliation
- Payroll preparation and variance checking
- Statutory and regulatory returns
- Client or funder reporting packs
- Commission and bonus calculations, which are error-prone and disputed
- Stock counts and variance reporting
The trap: infrequency hides the rules
Nobody remembers exactly how they did something they do twelve times a year, which is why monthly processes are the least documented and the most dependent on one person.
That makes discovery harder and more valuable. Frequently the documentation is worth as much as the automation.
How we scope infrequent work
We watch it happen, which means the project timeline is shaped by your cycle: if it is a month-end task, we need to be there at month end.
That is worth planning for rather than discovering. A monthly-process project usually spans two cycles: one to observe, one to run in parallel.