The slide says double your orders and the safety stock cut is 30%. The math is sound only because the source page says the step out loud: doubling the frequency halves the lead time [1]. A purchase order does not shorten a supplier’s queue. Leave the supplier’s clock alone and the same doubling returns 13.4% on a one-week replenishment, 4.3% on 35 days, and 2.0% on a 90-day import [2][3]. The buffer shrinks with the square root of the window it covers, and the inventory line that moves with the calendar is cycle stock, which halves when the order quantity halves [2].

The big picture:

A safety buffer covers an exposure window: the days between reviews plus the days the supplier takes to deliver. Both terms sit under one square root, so the buffer grows with the square root of the total, not with either term by itself [2]. Doubling order frequency halves the first term and leaves the second alone. The result depends on how much of the window was the review interval in the first place.

The 30% headline comes from the shelf case. In that model the new buffer is 0.707 of the old one, which is where the 30% and its 41% mirror come from [1]. The page’s own example is explicit: “the maximum delay time between replenishments is 1 day; hence, the lead time is also 1 day” [1]. That describes a shop floor or a same-day depot, not a part crossing an ocean, where the gap between orders is a rounding error against a 90-day clock.

Run the same halving against a four-week supplier and the cut lands near 4.3%, not 30% [2][3]. I keep coming back to how few teams rerun it. Once the supplier needs a month or more, the order calendar is already a small share of the window. On an import it is close to nothing.

The second inventory line moves the other way. Average cycle stock is about half the order quantity [2]. Halve the quantity and that balance halves, whatever the lead time does. On a fast mover, cycle stock is often the larger balance. So the calendar still matters, through quantity rather than through the buffer.

By the numbers

  • 29.3% — The ceiling: halving a weekly review to three and a half days trims the buffer 29.3% only when the supplier’s lead time is zero, which is the shelf case the model was written on [1][2].
  • 13.4% — One week of replenishment: give the supplier a seven-day lead and the same doubling returns 13.4%; at 35 days it returns 4.3%, and on a 90-day import, 2.0% [2][3].
  • $721 — Cycle stock: a part moving 12,000 units a year at $50, with a 25% carrying rate, holds $5,769 of average cycle stock on weekly orders; twice-weekly halves that and cuts the carrying bill by $721 a year [2].
  • $13.87 — Break-even per extra order: $721 spread over 52 extra purchase orders is the most one more order can cost before the cadence change loses money [5].

What I’d watch:

The teams closest to this weigh the supplier’s clock first, because it decides whether the order calendar is a lever or a rounding error. Where the whole lead time is under about ten days, the review interval is a large share of the exposure window, and halving it moves real money. On a four-week replenishment the same change is small. On an import it is close to nothing [2][3].

What a receipt costs is the other half of the decision. At the 2025 median of $21.76 an hour for shipping, receiving, and inventory clerks, fifteen minutes on the dock is $5.44 a line [5]. Freight minimums, invoice processing, and fixed dock shifts land on top. That is why teams rerun the $13.87 break-even before they approve a second weekly run.

The next answer sits in the order quantity, not the order count, because average cycle stock is half the order [2]. A price break at the old weekly quantity is the trap in the other direction. Order half as much twice as often, and the unit cost can rise by more than the carrying bill falls.

One level down the building, the same trade is already visible: a shelf holding a week of stock costs ten miles of walking, while a face refilled every hour swamps the crew doing the refilling.

The catch

Every anchor here except the wage is a software vendor’s explainer. The headline cut is a model, not a measurement. The page producing it is dated 2 February 2023 and signed by the planning vendor’s chief operating officer; its 30% and 41% results are careful arithmetic, not field results [1]. Nothing in the fetched text was tested on a real network.

Lead-time variance swamps the calendar. A part with a ten-day lead time needs 104 units of buffer at a fixed lead. Add a three-day standard deviation to that lead and the same part needs 506, a 4.9× swing the order interval cannot touch [4]. If the goal is a smaller buffer, the supplier’s reliability is the bigger lever.

There is also a unit trap inside the math, and it matters for planners copying the example. One review-period example labels its spread as weekly demand and then places a day-denominated window under the root, which overstates the buffer by about √7 [2]. A planner copying that shape should convert the spread first; the other explainer puts the rule in one line: “always use the same time unit for lead time and demand” [3].

I used to think the arithmetic was the hard part. The harder part is quoting a model without carrying its assumptions into the room. Ordering more often is a cadence call, and its size depends on the clock the part is on. That is a smaller claim, and a defensible one.

At a glance

The shift: Ordering more often shortens the order interval, not the supplier’s clock. The number: The buffer saving is a square-root effect, so it shrinks as the lead time grows. The risk: Cycle stock is the saving that scales, and one more order can spend it.

Related reading

Sources

[1] Arkieva, Sujit Singh (chief operating officer), “Effect of Change in Replenishment Frequency”, 2 February 2023 — the 0.707 multiplier, the 30% and 41% results, the sentence that doubling frequency halves the lead time, and the shelf model in which the delay between replenishments is the lead time. https://blog.arkieva.com/replenishment-frequency [2] Netstock, “How to Calculate Safety Stock Using 3 Methods” — the review-period formula that puts the order interval and the lead time under one square root, the 1.65 × 25 × √(7 + 14) worked example, and the definition of cycle stock as the stock consumed between replenishments. https://www.netstock.com/blog/safety-stock-meaning-formula-how-to-calculate [3] AbcSupplyChain, “6 Best Safety Stock Formulas” — the review period listed first among the factors that make lead time uncertain, the 35-day average against a 40-day maximum with ten deliveries a year, the 427-piece buffer and 1,578-unit reorder point, and the one-line rule that lead time and demand share a time unit. https://abcsupplychain.com/safety-stock-formula-calculation [4] User Solutions, “Safety Stock Calculation for Manufacturing” — the 104-unit fixed-lead result against the 506-unit result once a three-day standard deviation enters a ten-day lead time. https://usersolutions.com/blog/safety-stock-calculation [5] O*NET OnLine, 43-5071.00 Shipping, Receiving, and Inventory Clerks — median wages of $21.76 an hour and $45,260 a year in the 2025 wage table. https://onetonline.org/link/summary/43-5071.00