How to Calculate Your Reorder Point With Variable Demand

Your reorder point is the inventory level at which you must place a purchase order so the replenishment arrives just before you run out. With constant demand the formula is simple: daily demand times lead time. But your sales are not constant, which is why the version that actually works adds a third term: reorder point = (average daily demand × lead time in days) + safety stock. That last term is what absorbs the spikes, and without it your reorder point looks prudent on the spreadsheet and leaves you out of stock in real life.

The reason is that variable demand breaks the average exactly when it matters most. If you sell an average of 10 units a day and your supplier takes 15 days, a naive calculation tells you to reorder at 150 units. But if a strong week hits during that 15-day wait and you sell 14 a day, you needed 210, not 150. That 60-unit gap is a guaranteed stockout, and it happens on the good days, when losing the sale hurts most and you also risk the Buy Box.

This article shows you how to calculate each term with real data, how to size safety stock according to how variable your demand is, and why doing it by hand in Excel across several SKUs on Amazon and MercadoLibre falls apart by the second day. The goal is not a pretty number, it is a reorder point that survives the week when everything sells.

what a reorder point is and what it is not

A reorder point is not your minimum inventory or the level at which “you start to worry.” It is an exact trigger: the number of available units at which, if you place the order right then, the new stock enters the warehouse just as the old stock is about to run out. Crossing that point without having ordered means that, whatever you do, you will spend a stretch with no sellable product.

Notice the key word: trigger. The reorder point answers “when do I order?”, not “how much do I order?”. That second question is answered by your order quantity or your EOQ, and those are separate calculations. Mixing them is a common mistake: people set a minimum of 50 units thinking it is their reorder point, when that 50 has no relationship to their sales velocity or their lead time. A good reorder point is derived from those two figures, never picked by eye.

And because it depends on sales velocity and lead time, it changes by SKU and by channel. A product selling 25 a day on Amazon with local replenishment in 5 days lives at a completely different reorder point than an imported one moving 3 a day with 90 days in transit. A single number for your whole catalog is the fastest way to overstock half of it and stock out the other half.

Glossary: a stockout is running out of sellable units in a channel; the reorder point exists precisely so that replenishment arrives before that stockout happens.

the three ingredients of the calculation

To build a reorder point you need three numbers, and each has its trap. The first is average daily demand: how many units you sell per day in that channel. It sounds trivial, but the trap is which window you use. Demand over the last 7 days captures the seasonal acceleration; 30 or 90 days smooths the noise. A seasonal product at its peak is better calculated with the short window, or you underestimate demand; one just out of a promotion is better read with the long window, so you do not mistake a passing spike for the new normal.

The second is lead time: the days that pass from placing the order to the unit being available to sell, not when it “leaves the factory.” That nuance is expensive. Real lead time includes production, transit, customs, receiving and listing in the marketplace, which in FBA can add several days of check-in before the unit is sellable. Using the supplier’s optimistic lead time, instead of the one you measure end to end, is one of the most frequent causes of a reorder point that arrives late.

The third is safety stock, the buffer that covers variability. It is the term that turns a theoretical calculation into one that survives reality. If demand and lead time never varied, it would be zero. Since they do vary, its size depends on how erratic your demand is and how punctual your supplier is. We size it in the next section.

For lead time it is not enough to ask; you have to measure it. How to do it well is covered in what is lead time and how to measure yours accurately, because this number is half of your reorder point and almost nobody has it documented honestly.

why variable demand forces a safety stock

With perfectly stable demand the reorder point would be just demand × lead time and you would never stock out. The problem is that “10 a day on average” hides days of 4 and days of 16. The average tells you the center, not the extremes, and it is the extremes that leave you out of stock. If you reorder at exactly the average, you are basically betting that the wait window lands in a quiet stretch, and half the time you lose that bet.

Safety stock is what you buy to stop depending on luck. Its correct size grows with two things: the variability of your demand and the variability of your lead time. A SKU that sells evenly every day needs little buffer; one that lives on unpredictable spikes needs a lot. Same with the supplier: one that always delivers in exactly 15 days lets you relax, one that sometimes takes 15 and sometimes 28 forces you to cover the bad scenario.

The simple way to size it is against the reasonable worst case: instead of “10 a day for 15 days,” calculate “14 a day for 18 days” using your recent high demand and your recent long lead time. The difference against the average calculation is, in practice, your safety stock. The statistical way uses the standard deviation of demand and a factor based on the service level you want, but the principle is the same: the more your sales jump around, the more buffer you need to avoid stocking out in the good week.

Glossary: days of inventory measure how long your stock will last at the current sales pace; they are the right unit for reading a reorder point, because they translate “units” into “time before you stock out.”

an example with numbers that move

Take a case from SPORTIFY. A resistance band sells on Amazon with this data from the last month: average daily demand of 12 units, but in its best week it hit 17 a day. The lead time with the supplier is 20 days in theory, though the last two replenishments took 24 due to customs issues. With those numbers, the naive calculation gives 12 × 20 = 240 units as the reorder point.

Now the honest calculation. Demand × lead time with the average: 12 × 20 = 240. Safety stock at the reasonable worst case: high demand (17) times long lead time (24) gives 408; against the average cycle of 240, the buffer is 168 units. Realistic reorder point: 240 + 168 = 408. The difference between 240 and 408 is not excess caution, it is exactly the gap that leaves you without product in the strong week, right when the marketplace algorithm punishes you most for going out of stock.

Notice this number is not fixed. If the band enters high season and the average rises to 18 a day, the reorder point recalculates itself upward; if variability drops because the supplier stabilizes, the safety stock can shrink. That is why a reorder point written once into an Excel cell ages badly: it is correct the day you calculate it and drifts further off every day demand changes without you recalculating it.

why doing it by hand does not scale

The whole calculation above is perfectly sustainable once, for one SKU, in a good spreadsheet. The problem shows up when you have 80 or 300 products spread across Amazon, MercadoLibre, self-fulfilled shipments and a 3PL. There the manual update breaks by the second day: each reorder point depends on a demand that changed yesterday and a lead time that shifted with the last replenishment, and recalculating all of that by hand every morning simply does not happen.

You know the pattern. You open Seller Central to check FBA sales and stock, then MercadoLibre for Full and Flex, then the 3PL sheet that arrived by email on Monday, paste it all into an Excel, calculate sales averages, look up your lead times in another file and build the safety-stock subtraction SKU by SKU. By the time you finish, the data is already from yesterday and you decide with that uncertainty on top. Multiply it by hundreds of SKUs and four channels and the reorder number you use is almost never the one that fits today.

Worse still: each channel has its own inventory and its own velocity. FBA stock only serves Amazon; Full stock only MercadoLibre. A reorder point calculated on the total lies to you just as an average lies about the extremes: you can have 600 units combined and still be about to stock out in the one channel where that product sells fast. The calculation has to be per channel, and by hand that multiplies the work until it becomes impossible to sustain.

what changes with real-time data

A reorder point calculated in real time exists precisely so that manual work disappears. Instead of you gathering the information, the system reads the real available stock in each channel, measures the current sales velocity with the window you choose, crosses your measured lead time and calculates the reorder point with its safety stock, recalculating it every day against fresh data. When the available stock crosses that threshold, the alert fires on its own, not when you finally sit down to check tabs.

The foundation of all this is starting from the honest number. If the calculation starts from the warehouse’s physical stock instead of the real available stock per channel, your reorder point comes out inflated and the alert arrives late even though the threshold looks prudent. Real available stock subtracts reserves, internal transfers, returns pending processing and blocked units, and it is the only base on which a reorder point actually warns you in time.

That change, from manual recalculation on yesterday’s data to an automatic threshold on today’s data, is what connects the reorder point to the rest of your operation. The same discipline of reading fresh numbers per channel is what lets you protect margin when it is time to replenish: if you are going to rush a purchase to avoid a stockout, it pays to cross it against your profitability and ad spend, something we explain in what is ACoS, because saving the sale with express freight and the margin already eaten sometimes costs more than the stockout itself. A good reorder point does not just warn you in time: it gives you the days of margin to replenish well, not at any price.

Glossary: real available stock is what you can sell right now, once reserves, returns pending processing and blocked inventory are subtracted; it is the honest base on which a reorder point that does not arrive late is calculated.