Amazon FBA Inventory Forecasting: How Much Stock to Send Without Stockouts or Storage-Fee Shock

Every FBA seller knows the two bad weeks. One is the week your best seller goes out of stock and the ranking you spent months building slides backwards. The other is the storage bill for stock that is not moving. Both come from the same root cause: ordering by feel instead of by rate of sale. This guide is the alternative, a forecast built from three inputs you already have.

The Two Ways Inventory Goes Wrong

Stockouts cost more than the lost sales. When a listing goes unavailable, the sales velocity that feeds Amazon's ranking stops with it. Advertising campaigns keep spending for a day or two before most sellers notice, reviews stop accumulating, and when the stock finally lands again, the listing usually has to climb back from a worse position than the one it left. A stockout during your best season is the most expensive version of this: the demand was there, competitors took it, and no reorder speed fixes a missed peak.

Overstock fails quietly, then all at once. Amazon charges monthly storage fees per cubic foot on everything sitting in its fulfilment centres, adds surcharges on aged inventory that has sat too long, and raises storage rates in the busy fourth quarter. Stock you cannot sell eventually has to come back out through paid removals or disposal, which is a second freight bill for product you already paid to ship in. And underneath the fees sits the cash problem: money parked in slow units is money that cannot go into the products that are actually selling.

Sellers usually treat these as opposite problems. They are the same problem. Ordering the supplier's minimum quantity because it is simple, reordering when the shelf "looks empty", or sending six months of stock to save on freight per unit are all versions of deciding without a forecast. The fix is not complicated software. It is three numbers per SKU, reviewed on a rhythm.

The Three Inputs Every Forecast Needs

Input 1: your daily sales rate. Take the last 90 days of unit sales for the SKU and divide by 90. Then clean the number before you trust it. Days the listing was out of stock are not zero-demand days, they are days demand went unrecorded, so a period with a long stockout understates your rate. One-off spikes count too: a Lightning Deal week or a viral mention inflates the average, and ordering against it stocks you for a week that will not repeat. Most sellers land on a 30-day rate for fast-moving items and a 90-day rate for steady ones. Either is fine. What matters is that the rate is units per day, per SKU, not store revenue divided by hope.

Input 2: your true lead time. Lead time runs from the day you commit to the order to the day units are available for sale on Amazon. That includes production time, freight, customs, any prep work, and Amazon's check-in time at the fulfilment centre. Check-in is the segment sellers forget: a shipment that arrives at Amazon is not sellable stock until it is received, and receiving slows down exactly when you can least afford it, in the peak season. Use your actual last few orders to set this number, not the timeline your supplier quoted in the sales pitch. If the last three orders took 45, 52 and 61 days door to door, your planning number is closer to 60 than to 45.

Input 3: safety stock. Safety stock is the buffer that absorbs the two things you cannot schedule: demand running hotter than your rate, and lead time running longer than planned. A simple way to set it is in days: decide how many extra days of sales you want sitting at Amazon as insurance, and multiply by your daily rate. Short, reliable supply lines (a domestic wholesaler who ships in a week) need a thin buffer. A 50-day import lead time needs a thicker one, because a two-week slip somewhere in the chain is normal, not exceptional. Fourteen days of buffer is a reasonable starting point for imported products; adjust it after you have watched a few cycles.

With those three inputs, the whole system is two formulas. Reorder point = (daily sales rate × lead time in days) + safety stock. When your sellable stock plus stock already in transit falls to that number, you order. Shipment quantity = daily sales rate × the days of cover you want each shipment to provide. That is the entire engine. The worked example below shows it running.

A Worked Example: From 90 Days of Sales to a Reorder Quantity

Example with inputs shown, so you can re-run it with your own numbers. This is an illustration, not a sales figure for any real product.

Say you sell a kitchen gadget with the following inputs:

• Sales history: 1,080 units sold in the last 90 days, no stockouts in the period, so the daily sales rate is 12 units a day.
• Lead time: production 20 days, sea freight 25 days, Amazon check-in 5 days. Total 50 days.
• Safety stock: 14 days of sales, so 14 × 12 = 168 units.
• Cover target: each shipment should provide 60 days of stock.

Reorder point: (12 × 50) + 168 = 768 units. The day your sellable plus in-transit stock drops to 768, the next order goes in. Shipment quantity: 12 × 60 = 720 units per shipment. Order 720, sell roughly 600 of them during the 50 days the following order is in transit, and the safety stock stays untouched unless demand spikes or the shipment slips. That is the plan working as designed: the buffer exists to be borrowed from occasionally, not to be planned away.

Now run the same product through the failure mode. Suppose you ignored the forecast and sent 2,160 units, six months of stock, to save on freight per unit. If the daily rate holds, the last units sell half a year later, and every one of those months the full pile is being charged storage by volume, with aged-inventory surcharges joining in once stock sits long enough. If the rate drops, say a competitor launches and you fall to 7 units a day, you are now holding close to ten months of stock. Before you commit to a big shipment, price the slow scenario honestly: our free Amazon FBA storage fee calculator shows what a given quantity of your product costs to store month by month, and the break-even calculator shows whether the freight saving survives the storage bill. On many products, it does not.

Seasonal Spikes and Launch Stock: Adjust, Don't Guess

Seasonality tempts sellers to throw the forecast away and order big on instinct. Keep the system, change the inputs. If you sold through the same season last year, use last year's daily rate for those weeks as your seasonal rate, and order against that, not against an all-year average that flattens the peak away. If lead times stretch in the busy season (freight and check-in both do), order earlier rather than simply ordering more: moving the order date two weeks sooner protects the peak without leaving you overstocked after it.

Launch stock is the opposite problem: there is no sales history, so there is no rate. Send a deliberately small first shipment, sized on a conservative estimate from your product research, and plan the second order the moment real sales data exists. An unproven listing holding six months of stock is the single most common way new sellers freeze their cash. A stockout on a brand-new listing is a minor inconvenience; you lose a few days while the product is still finding its level. Ten months of dead launch stock is a business problem. Forecast launches in short cycles until the product has earned a longer one.

Promotions belong in the same conversation. If you know a deal or a price drop is coming, expect the rate to lift for that window and stock for it explicitly, then return to the base rate afterwards. What you should not do is leave the post-promotion rate in the forecast for months. Temporary lifts that nobody removes are how quiet overstock builds.

FBA or Your Own Shelf: When to Split Shipments

Not every unit you own has to live at Amazon. Splitting stock between FBA and your own shelf (or a third-party warehouse) makes sense when lead times are long and demand is uneven: keep a buffer outside Amazon, drip-feed FBA in smaller, more frequent shipments, and a demand spike becomes a fast top-up instead of an emergency air freight order. It also gives you somewhere to pull stock back to if a listing gets into trouble.

Sending everything to FBA makes sense when the product is small, sells at a steady rate, and your supply line is reliable: the double handling of a split setup costs more than the flexibility is worth. The trade-off to price is storage at Amazon versus handling and outbound postage from your own side, run through the same storage calculator linked above.

One practical point either way: units moving into FBA still have to arrive prepped and labelled correctly, whoever holds them first. Since Amazon ended its own prep service, that work sits with you, your supplier, or a prep center, and inbound defects are charged back to your account. Our guide to Amazon FBA prep centers and inbound defect fees covers how sellers are organising that step now.

The Review Rhythm: Weekly Check, Monthly Reset

A forecast is not a spreadsheet you build once. It is a short recurring routine:

• Weekly check (15 minutes). For each SKU, compare days of cover (stock on hand ÷ daily rate) against your lead time plus safety stock. Anything at or below that line gets ordered or expedited this week, not next. Note any SKU whose rate moved sharply, up or down, and ask why before you react.
• Monthly reset. Recalculate every daily rate from fresh data, retire assumptions that stopped being true, and review your aged inventory and last storage bill side by side. The SKUs costing the most storage are usually the ones whose rate fell months ago while the reorder habit carried on.
• After any disruption. A stockout, a deal week, a supplier delay: each one corrupts the averages. Mark the period, exclude it from the rate where it distorts, and rebuild the number rather than letting one strange month steer the next order.

Done consistently, this is most of what "inventory management" means at FBA scale: three inputs per SKU, one weekly glance, one monthly reset. It is also, frankly, the kind of unglamorous routine that gets dropped the moment a seller gets busy, which is why it sits inside the operating rhythm for accounts on our e-commerce operations management service. Whether you run it yourself or hand it over, run it on the numbers. Your storage bill and your ranking will both show the difference.