Field Notes · 2 April 2026
Reading tolerance tables without guessing
A calm way to interpret price and quantity tolerances before you widen or tighten them.
Tolerance tables look numerical and objective. In practice they encode old purchasing habits, currency rounding, and compromises struck after a painful close. Widening them “a little” without reading the sample is how controls erode.
Start from invoices, not from the setting screen
Pull a mixed sample: auto-matched, near-miss, and blocked. For each near-miss, write the percentage and absolute difference between PO, receipt, and invoice. Only then open the tolerance table. You will see which band actually decides the outcome.
Separate percentage from absolute
A 2% price tolerance on a ₩50,000 line is a different control from the same percentage on a ₩50,000,000 capital item. Absolute caps matter. Many invoice matching applications let both coexist; audits should state which one won for each sample item.
Quantity quirks
Partial receipts, catch-weight items, and service entry sheets create quantity stories materials clerks already know. If the matching application still expects full PO quantity, exceptions are policy — not software failure.
Change discipline
When tolerances do change, record who approved the change, which entities it affects, and what exception volume you expect next close. Online Reason includes that change trail in full application audits so future reviewers are not guessing again.