Almost every finance team we meet has a workbook named something like final_v7 or revenue_master_DO_NOT_EDIT. It is not there because the team distrusts spreadsheets less than software — it is there because the billing or revenue system stopped modelling a case the business actually has, and the spreadsheet is where that case gets fixed by hand every month.
The pattern is consistent enough to name: a deferred instalment on a one-time fee that the system bills on the wrong phase. A volume tier that changes mid-term and the system just re-rates the whole period instead of splitting it. A seat vendor's "increase" column that the system imported literally, so it broke at every customer boundary in the file. None of these are exotic. They are the normal texture of B2B contracts, and a system that cannot model them pushes the correction into a human's hands — and once it is in a human's hands, it goes into a spreadsheet, because that is the tool humans reach for under deadline.
The expensive part is not the spreadsheet itself. It is that the spreadsheet becomes the actual source of truth silently. Six months in, the "system of record" and the shadow sheet have diverged in ways nobody remembers explaining, and the close takes two extra days reconciling them instead of zero.
The fix is not a policy telling people to stop using Excel. It is closing the specific modelling gaps that created the sheet in the first place: deferred tranches resolved by arithmetic instead of a manual note, mid-term rate changes handled as a phase split instead of a re-rate, snapshot imports that reconcile at customer boundaries instead of vendor delta columns that don't. When the system actually produces the number finance would have hand-corrected to, the shadow sheet stops earning its keep — not because someone banned it, but because it has nothing left to fix.