Three to eight hours a week across a finance team of three to five, concentrated in the last week of the month. It exists because the ERP's own reconciliation report is not trusted, usually for a reason nobody has written down.
Last reviewed 30 September 2026
At the surface, live, owned by one person who built it and is the only one who understands the tabs. It pulls a bank export, a debtors export and a creditors export, and matches them with lookups and colour. It is where the month actually closes.
A reconciliation report in the ERP, dormant. It was run once, disagreed with the bank by a few thousand rand, and has been treated as wrong since. Underneath that, a chart of accounts that was restructured in a year nobody remembers, which is usually the real reason the report disagrees.
A process fix first: find why the ERP report disagrees and correct the mapping, which is often a morning's work. Then plain automation for the matching, with the exceptions listed for a person. AI has no place here; the rules are already written, in the spreadsheet.
Timed across a full month-end, not a normal week, then averaged. The spreadsheet's own edit history gives a second source. Where the two disagree, the lower figure wins. How EMBR measures. This layer is one part of workaround debt; The Transect measures all of them together.
Tell us what you're working around, and we'll come back to you within two working days.