EOQ frequently asked questions
Twelve questions the SERP under-answers. Each is the kind of question a CFO or PE operating partner will ask after seeing a Q* on a board pack[1]. If your question is missing, the calculator and derivation pages are the next stops.
What carrying cost percent should I use?
Default to 25 percent for stable SMB goods. Decompose: capital cost (5 to 10 percent) plus warehouse (2 to 5) plus services (1 to 2) plus risk (1 to 3) plus obsolescence (variable, 2 to 20). Defend the components, not the total.
When does EOQ stop being the right tool?
When demand is non-stationary (fashion, perishables, electronics in sunset), when shortages are catastrophic and immediate (use safety stock + high ROP instead), when supplier MOQ binds (use MOQ, document the over-order penalty), or when production rate p is finite (use EPQ).
Why does my answer differ from the Shopify or Omni calculator?
Three usual reasons: (1) they use H as a flat dollar number where we use H = i * C; (2) they ignore the sensitivity strip; (3) they round differently for case-pack constraints. The underlying formula Q* = sqrt(2DS/H) is identical.
How do I pick service level for ROP?
A-class items: 97.5 to 99 percent. B-class: 95 percent. C-class: 90 percent. The Z value grows non-linearly: each step up the service ladder is roughly 19 percent more safety stock. Match service level to the stockout cost.
Can I run EOQ across 200 SKUs at once?
Yes. Use the CSV template (export-csv-template page). One row per SKU, columns D, S, C, i. Compute Q* in a formula column. The math is identical per row.
How sensitive is Q* to mis-estimated H or S?
Q* depends on the square root of S/H. A 25 percent error in H moves Q* by about 13 percent. The total cost at the wrong Q is approximately TC(Q*) * (1 + epsilon^2/2), so a 25 percent Q error costs about 3 percent of total annual variable cost. Forgiving.
Should I use EOQ for fashion or perishables?
Stable basics inside fashion: yes (denim, white tees). Seasonal fashion-cycle SKUs: no, use a newsvendor model. Perishables under shelf life: yes, but cap Q at d * shelf_life. Perishables with exponential decay: Ghare-Schrader extension.
How does MOQ override Q*?
If supplier MOQ > Wilson Q*, MOQ binds: actual_Q = MOQ. The carrying-cost penalty (H * (MOQ - Q*)/2 minus the ordering-cost saving from fewer reorders) is the cost of the MOQ. Document it on the PO and use it as the negotiation lever.
How is EOQ different from EPQ?
EOQ assumes instantaneous resupply (full batch drops at once). EPQ assumes production at rate p > d (inventory builds at slope p - d, drains at slope d). The (1 - d/p) factor reduces effective holding penalty. As p approaches infinity, EPQ collapses to EOQ.
Is JIT a replacement for EOQ?
JIT re-engineers EOQ's inputs (drives S toward zero via SMED) so Q* shrinks toward 1. The math is identical; the inputs change. JIT cannot run with long international lead times; a kanban + EOQ hybrid is the practical pattern.
How do I cite this in a board pack?
Wilson EOQ formula Q* = sqrt(2DS/H) (Harris 1913; popularised by Wilson 1934). Inputs documented per page. Sensitivity strip shows robustness. See the /sources page for the full bibliography.
Why is Q* a square root?
Setting dTC/dQ = 0 gives H/2 = DS/Q^2, which solved for Q is sqrt(2DS/H). The square root falls out of the calculus, not from an arbitrary modelling choice. It is the reason mis-estimated inputs are forgiving and the cost curve is flat near Q*.