Coverage — the replenishment sheet · Sixty prototypes, two engines

The artifact

The sheet the planner opens on Monday.

Same forecast, same file, same system of record. One column changed, and a footer appeared under it that no forecasting tool currently prints: what the interval above actually covered the last time anyone checked.

Replenishment plan · week of 2025-12-2295% service levelcalibrated on 624 held-out SKU-weeks

SKU	Forecast	Old order-up-to	Calibrated order-up-to	Change
SKU-1180 Cable, USB-C 2m	88	105	132	+27
SKU-1425 Cable, USB-C 1m	117	136	168	+32
SKU-2044 Filter cartridge	161	183	220	+37
SKU-2390 Stylus, replacement tip	196	221	262	+41
SKU-3312 Hub, 7-port	235	262	307	+45
SKU-3808 Stand, monitor arm	293	323	373	+50
SKU-4901 Desk mat, large	372	406	462	+56
SKU-5140 Keyboard, compact	440	477	538	+61
SKU-5507 Battery pack 10k	509	549	615	+66
SKU-6055 Webcam, 1080p	597	640	712	+72
SKU-6720 Charger, 65W	744	792	872	+80
SKU-7314 Charger, 30W	920	974	1062	+88

Coverage note. The interval this sheet used to print covered 65.5% of held-out weeks against a stated 95.0%, and demand ran past its upper bound in 104 of 624 SKU-weeks. The calibrated interval covered 95.8% and was exceeded in 24. Both numbers are counted on weeks that were not used to build either interval.

What the planner does with it

Nothing new. The order-up-to level goes into the same system it always went into. The difference is that the service level written on the plan is now the service level the warehouse experiences, so when someone asks why a SKU stocked out there is a number to point at instead of an argument.

What it will cost them

More stock. Every calibrated bound on this sheet is higher than the one it replaces, because the old one was too low. The honest version of this pitch is not that conformal saves money; it is that the missing width was already being paid for in expedites and lost sales, and now it is visible and choosable.

What this does not decide

It does not choose your service level, set reorder policy, or touch the point forecast. It answers one question, which is what a stated interval actually covers, and then makes the stated one true. If a 95% service level is too expensive for a SKU, that is a business call, and the right response is to pick a lower level deliberately rather than to keep a 95% label on a band that delivers 65.5%.