EOQ vs Periodic Review: Which Inventory Policy Fits

Your trims store has 4,200 open bins and one planner. Half are reordered on gut feel ("we're getting low on the black 12L buttons") and the other half sit on a spreadsheet formula nobody remembers writing. Both halves run out at the same rate.

That is not a forecasting problem. It is an inventory policy problem: nobody decided, SKU by SKU, whether stock should be checked continuously or on a calendar, and nobody sized the buffer against how unreliable the supplier actually is, versus how unreliable the textbook assumes he is, the same structured trade-off a decision tree for capacity investment forces onto paper for the bigger capital calls.

The cost shows up in three places: expedited air freight on a zip that was two days from stockout, a cutting room held for buttons "in the system" but not on the shelf, and a finance director asking why trims inventory value climbed 18% this year while on-time ship rate didn't move.

This article gives you the actual decision: which inventory policy to run, on which class of item, and how to size the safety stock so it survives a supplier who ships nine days late in July and four days late in November. It also covers the apparel-specific mistake that inflates working capital more than any formula error: putting fabric on the same policy as buttons.

The Four Inventory Policies You Actually Choose Between

Most inventory policy conversations collapse into "EOQ or periodic review" as if there were two options. There are four, and they answer two questions: do you check stock continuously or on a calendar, and does each trigger a fixed quantity or a fixed target level.

Most trims stores should run a mix of (s, Q) for the handful of high-value fasteners and (R, s, S) for the long tail of low-value items ordered from the same two or three vendors. Running one inventory policy across all 4,200 SKUs is the default that got you here.

How Much Safety Stock Do You Actually Need?

Safety stock protects against variability during the exposure window: the time between deciding to order and the replenishment landing on the shelf. Under continuous review, that window is the lead time L. Under periodic review it's R + L, which is why periodic policies need more buffer for the same service level.

The standard formula is Safety Stock = z × σ_L, where z is the service factor (1.65 for 95%, 2.33 for 99%) and σ_L is the standard deviation of demand over lead time. The part planners routinely get wrong is σ_L itself: most compute it from demand variability alone and ignore lead time variability entirely:

σ_L = √( L × σ_d² + d̄² × σ_LT² )

where d̄ is average daily demand, σ_d is the standard deviation of daily demand, both ideally pulled from an exponential smoothing demand plan rather than a flat historical average, L is average lead time in days, and σ_LT is the standard deviation of lead time.

Worked example: woven labels, single supplier, 95% cycle service level: average daily consumption 800 pieces, standard deviation of daily demand 150 pieces, average lead time 20 days, standard deviation of lead time 5 days.

Ignoring lead time variability (the common shortcut): σ_L = √20 × 150 = 671 → Safety Stock = 1.65 × 671 ≈ 1,107 pieces.

Including lead time variability (the correct calculation): σ_L = √(20 × 150² + 800² × 5²) = √16,450,000 ≈ 4,056 → Safety Stock = 1.65 × 4,056 ≈ 6,692 pieces.

The second term, lead time variability, accounts for roughly six times more of the required buffer than demand variability alone in this example. If your ERP's safety stock module doesn't ask for a lead time standard deviation as an input, it is running the wrong formula.

Field note: A knitwear plant in Chattogram ran uniform 15-day safety stock across all sewing thread SKUs for two years. A stockout review found the actual lead time on dyed-to-order thread ran 18 days average with a 6-day standard deviation, against the 12-day quote the planner had used since vendor onboarding. Recalculating σ_L against the real 18/6 figures nearly tripled the required buffer on that item alone, and explained four of the previous year's six thread-related line stoppages.

Trims Follow a Policy. Fabric Follows a Purchase Order.

This is the costliest mistake in the whole subject, and the most common: applying EOQ or continuous review to fabric.

Fabric consumption is not a steady draw from a shelf; it is order-linked. A style consumes exactly the fabric its cutting plan requires, at the yardage marker efficiency dictates, on the timeline its PO sets. There is no meaningful "reorder point" for a print that will never be rebought once the style ships.

Trims and accessories, buttons, zips, labels, thread, poly bags, hangtags (itself a frequent source of packaging decisions that quietly inflate landed cost), are the correct home for EOQ and the four policies above, because they are genuinely reusable across styles and seasons. Fabric, by contrast, should be planned against the order book: buffer sized against fabric-specific risk (shade band, shrinkage allowance, cut-and-sew loss), not a demand-variability formula built for a shelf item. Getting this backwards is how factories end up with warehouses of committed, single-style yardage they can't redeploy and are still calling "safety stock."

Inventory Policy Selection Checklist

Final Word

None of this requires new software: see our take on what to build first in a manufacturing analytics stack if you're wondering what would. It requires deciding, item by item, which inventory policy applies, and refusing to let fabric and buttons share a formula built for neither. The plants that get this right aren't running better math; they're running the right math on the right item, checked against real lead time variance.