James Allman | JA Technology Solutions LLC
The Math Behind Inventory Decisions
EOQ, safety stock, and reorder point are standard supply chain formulas. Here is what each calculates, what it assumes, and where real inventory breaks the model.
Most inventory planning software has an order-recommendation engine under the hood. Set your service-level target, enter lead time, and the system produces a reorder point and a suggested order quantity. What is less clear, even to people who use these systems daily, is what math is running: what it optimizes for, what it assumes, and what happens when the assumptions do not hold. This article covers the five formulas at the center of inventory math, with enough detail to read the numbers your system produces, challenge the ones that seem wrong, and build or extend the ones your system does not provide.
The tools linked throughout are free, run in the browser, and cover each formula in isolation. The service that ties them together, pulling live demand and lead-time data and pushing reorder actions into an ERP or WMS, is custom integration work. Both paths are here.
EOQ: the order size that minimizes total cost
Economic Order Quantity (EOQ) answers one question: how many units should I order at once to minimize the combined annual cost of ordering and holding? The formula is EOQ = √(2DS/H), where D is annual demand in units, S is the fixed cost to place one order (purchasing time, receiving labor, inbound freight, paperwork), and H is the annual cost to hold one unit in inventory (capital cost, storage, insurance, shrink, obsolescence).
The logic behind the square root is a balance between two opposing costs. Ordering in large batches cuts the number of orders placed per year, so annual ordering cost (D/Q × S) falls as Q grows. But larger batches mean higher average inventory on hand, so annual holding cost (Q/2 × H) rises. The total of the two is minimized where the two curves cross, which algebra shows happens when Q = √(2DS/H). At that point, annual ordering cost equals annual holding cost exactly.
One property worth knowing: the total-cost curve is nearly flat around the minimum. Being 20% off the true EOQ typically changes total annual cost by only 1–2%. That flatness is why rounding the EOQ to a convenient case or pallet quantity rarely costs much. Use the result as a directional check, not as a mandate. The EOQ Calculator computes the formula and shows the ordering and holding cost components separately.
Safety stock: the buffer for variability
EOQ assumes demand is constant and lead time is fixed. Neither is true. Demand varies day to day and week to week; suppliers deliver early or late. Safety stock is the extra inventory held to absorb that variability and still fill customer orders at the target service level. Without it, the reorder point would be set to exactly the average demand during lead time, and any spike in demand or delay in delivery would produce a stockout.
The standard formula for safety stock is SS = Z × σ(demand over lead time), where Z is the service-level factor from the standard normal distribution and σ is the standard deviation of demand over the lead-time period. A 95% service level uses Z = 1.645; a 99% service level uses Z = 2.326. The σ term combines both sources of variability: if both demand per day and lead time vary, the combined standard deviation is √(L̄ × σ_d² + d̄² × σ_L²), where L̄ is average lead time in days, σ_d is the standard deviation of daily demand, d̄ is average daily demand, and σ_L is the standard deviation of lead time.
The practical tradeoff is direct: raising the service-level target raises Z, which raises safety stock, which raises the inventory investment. A buyer who wants to go from 95% to 99% service level adds roughly 0.68 standard deviations of safety stock per SKU. On a SKU with meaningful demand variability and a long or unreliable lead time, that is a real inventory commitment. The Safety Stock Calculator computes the buffer for any combination of demand variability, lead-time variability, and service-level target.
Reorder point: when to trigger the order
The reorder point (ROP) is the inventory level at which a replenishment order should be placed. It is calculated as: ROP = (average demand per day × average lead time in days) + safety stock. The first term is the expected demand that will be consumed while the order is in transit. Safety stock sits below it as the buffer. When on-hand inventory falls to the ROP, an order of EOQ units goes out; ideally, it arrives just as on-hand inventory reaches the safety-stock floor.
The three variables in the ROP formula each carry measurement decisions. Average demand per day should reflect a recent, clean demand signal, not one distorted by stockouts (when inventory was zero, demand was suppressed), promotions (demand was temporarily elevated), or seasonal peaks. Average lead time needs to be the lead time from the same supplier and lane the reorder will use, not a catalog number. And safety stock depends on the service-level target chosen for that SKU, which is typically higher for A items than for C items.
The Reorder Point Dashboard applies the ROP formula across a full SKU portfolio uploaded as a CSV. It flags any SKU whose current on-hand is already below its reorder point, which in a live operation means an order should already be in flight.
ABC analysis: where to spend the inventory control effort
Not every SKU deserves the same safety stock, cycle count frequency, or buyer attention. ABC analysis ranks the assortment by annual dollar usage (unit cost × annual units sold) and partitions it into three tiers. A items are typically the top 10–20% of SKUs that drive 70–80% of total inventory value. B items are the middle tier. C items are the long tail, individually cheap or slow-moving, often numerous.
The practical consequence of the ranking is that control effort follows value, not item count. A items get tight safety stock targets, high service levels, frequent cycle counts, and close buyer review. C items get looser safety stock, lower service levels, and less frequent counts. Treating a C item with A-item rigor wastes labor; treating an A item with C-item looseness produces expensive stockouts. The ABC Analysis Tool runs this ranking on a CSV of your SKU list and produces the tier assignments.
One nuance: annual dollar usage is a proxy for the cost of error, not a perfect measure of it. A slow-moving spare part that costs $4 per unit can still be a de facto A item if a stockout shuts down a line. Pure dollar-usage ranking misses those cases. Most operations add an override column for critical or single-source items, or run a separate criticality analysis alongside the ABC.
Inventory turns and GMROI: how hard the investment works
Ordering and safety-stock formulas optimize the flow of inventory. Inventory turns and GMROI measure the result: how productively the inventory investment is deployed. Inventory turns = cost of goods sold / average inventory at cost. A turns figure of 12 means the inventory turns over roughly once a month on average. Higher turns mean less cash tied up in stock relative to sales; lower turns mean excess inventory or slow movement.
GMROI (gross margin return on inventory investment) adds the margin dimension: GMROI = gross margin dollars / average inventory at cost. Where turns alone measure volume, GMROI measures profit productivity. A department generating $200,000 in gross margin on an average $100,000 inventory investment has a GMROI of 2.0, meaning each dollar of inventory investment returns two dollars of gross margin. Comparing GMROI across categories reveals which departments are working their investment hardest. The GMROI & Inventory Turns Calculator computes both from period sales, margin, and inventory figures.
The link to EOQ and safety stock is direct. A lower EOQ means you order more frequently in smaller quantities, which lowers average inventory and raises turns, but raises total ordering cost. Safety stock raises average inventory and lowers turns by definition. Every service-level decision is implicitly a turns and GMROI decision as well. Inventory valuation also feeds these numbers: the cost flow assumption (FIFO, LIFO, or weighted average) changes the average inventory cost figure that sits in the denominator. The inventory valuation article covers that detail.
Cycle counting: accurate counts without the annual shutdown
The formulas above are only as good as the on-hand counts that feed them. A reorder point calibrated for 30 days of safety stock provides no safety stock if the on-hand field shows 500 units when the shelf holds 50. The classic inventory audit is the annual physical count: every location, every SKU, operations suspended. For most operations it is too infrequent to be operationally useful and too disruptive to run more often.
Cycle counting solves this by spreading the counting work continuously across the year. Instead of counting everything once, a subset of locations and SKUs is counted each day or week, ABC-weighted so A items are counted several times per year and C items perhaps once. The physical count team keeps working, the results are reconciled against the system in small batches, and discrepancies are investigated while the evidence is fresh. Accuracy stays high without a shutdown. The Cycle Count Planner generates a count schedule weighted by ABC tier and count-frequency targets.
Where the formulas break
The EOQ model assumes constant demand, fixed lead time, no quantity minimums, and independent SKUs. Real inventory breaks every one of those. Lumpy or intermittent demand, common in industrial spare parts and seasonal goods, produces a demand standard deviation that is as large as or larger than the mean; safety stock calculated on that σ can be enormous, often larger than a reasonable EOQ. The right response is not a bigger buffer, it is a different model: min-max, base-stock, or a periodic review system tuned to the demand pattern.
Supplier minimum order quantities (MOQs) override the EOQ entirely when the MOQ is larger. If the EOQ for a slow-moving item is 20 units but the supplier's minimum is 144, you order 144. The holding cost of the excess is the real cost of the minimum, and the right comparison is the annual holding cost of carrying the extra 124 units against the cost of finding an alternative supplier or consolidating the order with something else. The formulas can model this; most tools just do not expose it.
Perishability and promotions distort demand history in opposite directions. A stockout during a promotion suppresses observed demand below true demand. Perishable items develop a forced-scarcity demand pattern near end of life. Both cases produce a biased demand signal that feeds incorrect safety stock and reorder points if used raw. Cleaning the demand history, removing the stockout zeros and isolating the promotional lifts, is data work that has to happen before the formulas run.
These formulas are well understood and widely implemented. The gap is usually not that an organization lacks the math, it is that the inputs are not clean, the formulas are not wired to live data, or the outputs are not integrated into the purchasing workflow. A safety stock number on a spreadsheet helps one buyer. The same number recalculated nightly from live POS or ERP data and loaded back as a reorder point into the purchasing system helps the whole operation.
That integration is what I build: extraction from ERP or POS, formula execution against cleaned demand and lead-time data, and results loaded back into the system that drives purchasing decisions. If you want to talk through how your replenishment math works today, or what it would take to automate it, get in touch or use integration services as an entry point. Ask James and describe your current setup; I read every conversation that comes through the chat.