Shelf-life constraints should change your safety stock math

Standard safety stock formulas assume symmetrical demand error. Chilled networks violate that assumption weekly when batches approach expiry. We model spoilage as a first-class constraint, not an afterthought. Service level definitions shift from “in stock” to “sellable days remaining above threshold.” That language aligns merchandising, QC, and operations in the same spreadsheet row. Simulations layer temperature excursion risk where data exists. When data is thin, we document the gap and widen confidence bands instead of pretending precision. The outcome is not prettier charts—it is an honest conversation about shrink versus availability before the next heat wave.