Avoid stockouts: the maths behind FBA restocking
A stockout costs more than the margin of the days without stock. It costs rank, the Buy Box and the recovery afterwards. This guide shows what actually happens on Amazon, what usually causes it, and the formulas you need to work out lead time, safety stock and reorder point yourself.
- Formulas with a worked example
- FBA capacity limits and inbound lead time
- Works without any tool
Last updated:
What a stockout on Amazon really triggers
Amazon ranks offers, not products. Once stock hits zero, the basis for that ranking disappears – and not only for the days you are out of stock.
- The Buy Box is gone
- Without sellable stock there is no offer that can win the Buy Box. If you are the only seller, the buy button disappears. If you share the listing with others, the next best offer takes over – along with the customers who would have bought from you.
- Search visibility drops
- Sales velocity is one of the strongest factors in Amazon's ranking. An offer without stock makes no sales, collects no conversions and slides down on its keywords. ASINs that cannot be bought are often not shown in search results at all.
- The BSR keeps falling afterwards
- Best Sellers Rank is a rolling sales rank over the past days and weeks. It therefore keeps dropping after the stock is back in – the low point sits behind the end of the stockout, not before it.
- PPC runs into nothing
- Sponsored ads need a Buy-Box-eligible offer. They stop with the stockout, the campaigns lose the history they had built up, and they restart afterwards at worse click prices.
- The restart effect
- After restocking, the listing does not resume where it left off. It has to rebuild sales velocity first, usually with extra ad budget and a sharper price. The longer the stockout lasted, the longer that restart takes.
- Customers change habits
- Someone who bought from a competitor once does not automatically come back. For consumables with a fixed repurchase rhythm this weighs more than the revenue lost during the stockout itself.
The damage is therefore systematically larger than the margin lost on the empty days. That is exactly what justifies a safety stock that looks too high on paper.
The usual causes
Almost every stockout traces back to a handful of causes. They rarely appear alone – usually two of them combine.
- Ordered too late
- The most common and most trivial case: the reorder starts when stock looks low, not when a calculated reorder point is reached. With a 65-day lead time, “looks low” is about two months too late.
- Lead time underestimated
- People calculate with the production time the supplier quotes. What they do not calculate is order confirmation, quality control, loading windows, customs clearance, transport to the warehouse and FBA receiving. For sea freight from Asia, the gap between “goods are finished” and “goods are sellable” is regularly several weeks.
- Demand from distorted data
- Averaging daily sales over a period in which the product was temporarily unavailable underestimates real demand – and leads to ordering too little again. The error compounds with every stockout.
- FBA capacity limits
- The goods are ordered, paid for and in the country, but Amazon will not accept them because the account's capacity limit is exhausted. The stock then sits in your own warehouse while the listing runs to zero – the most expensive kind of stockout, because the capital is already tied up.
- Seasonality planned on an annual average
- An annual average systematically underplans the peak. Plan September with July's average sales and you will be empty in November – in exactly the weeks where lost margin is highest.
- Supplier delays
- Delays are not an exception, they are a distribution. If your supplier is on time on average but two weeks late in one out of five orders, that spread is precisely the reason for safety stock – not the mean.
How to calculate the reorder point
Five figures are enough. They build on each other, and each can be derived from data that already sits in Seller Central and in your purchase orders.
- 01
Average daily sales
The basis for everything else: units sold divided by the number of days the product was actually available – not by the number of calendar days. Pick a window that fits the product: 28 days for stable demand, or the same period last year for seasonal items you are planning ahead for. Note the standard deviation of the daily values as well, you need it in step 3.
Average daily sales = units sold ÷ days in stock
Strip out the days without stock. Otherwise the forecast gets worse with every stockout instead of better.
- 02
Lead time
The time from the decision to order until the moment the goods are booked as sellable in your FBA inventory. It is a sum, not a single number from the supplier. Measure it against completed orders instead of estimating it – and record the spread across your last few orders too.
Lead time = order handling + production + transport + customs & prep + FBA receiving
The last term is the one most often forgotten. Several days pass between delivery at the fulfilment centre and the stock being booked in, more in the fourth quarter.
- 03
Safety stock
The buffer for everything that runs differently than planned between order and put-away. Two quantities vary at the same time: daily demand and lead time. Safety stock covers both together, weighted by the service level you want to run.
Safety stock = Z × √( lead time × σ_demand² + avg. daily sales² × σ_leadtime² )
Z is the service-level factor: 1.28 for 90 %, 1.65 for 95 %, 1.96 for 97.5 %, 2.33 for 99 %. The buffer grows disproportionately – going from a 95 % to a 99 % service level requires about 41 % more safety stock.
- 04
Reorder point
The point at which the reorder has to be triggered. It is compared not against available FBA stock but against your inventory position: available plus inbound plus your own warehouse plus goods already ordered but not yet delivered.
Reorder point = average daily sales × lead time + safety stock
Trigger as soon as the inventory position reaches or falls below the reorder point. Looking only at available FBA stock leads to double ordering on long transport routes.
- 05
Order quantity
The reorder point tells you when, not how much. The quantity covers the lead time plus the period until the delivery after next – minus what you already have or what is already on its way.
Order quantity = average daily sales × (lead time + order cycle) + safety stock − inventory position
With FBA there is an upper bound on top: anything beyond your free capacity limit cannot be stored and has to wait in your own warehouse or at a third-party logistics provider.
Five mistakes that void the whole calculation
- Averaging daily sales over calendar days instead of days in stock.
- Ending the lead time at delivery instead of at the booking into FBA inventory.
- Setting safety stock as a flat percentage uplift instead of deriving it from measured variability.
- Planning the seasonal peak on the annual average instead of on the same period last year.
- Checking the reorder point against available instead of total inventory position – open purchase orders and inbound shipments count.
Worked example: sea freight from Asia
One article with stable demand, production in Asia, sea freight, fulfilled by Amazon. All values are assumptions used to show the method – put your own numbers in.
Inputs
- Average daily sales
- 18 units
- Std. deviation of daily sales (σ_demand)
- 6 units
- Lead time
- 65 days (21 production + 32 sea freight + 4 customs & prep + 8 FBA receiving)
- Std. deviation of lead time (σ_leadtime)
- 7 days
- Service level
- 95 % → Z = 1.65
- Inventory position today
- 900 units
- Order cycle
- 30 days
- Volume per unit
- 2.4 litres
Calculation
What this example shows
- 87 percent of the variability comes from lead time, only 13 percent from demand. Making a supplier more reliable or cutting three days off FBA receiving reduces the stock you need here more than any improvement to the demand forecast.
- The reorder point is 1,393 units, the inventory position is 900 – this order is already overdue. At 18 units a day, 900 units last exactly 50 days; the goods need 65. Even if the order goes out today, roughly 15 days without stock are baked in, and the safety stock is long gone by then.
- The order quantity needs about 2.48 m³ of free storage volume. If your capacity limit for the target month does not cover that, the calculation is correct and still not executable – you split the shipment and store the rest elsewhere.
What FBA adds on top
The formulas above hold for any warehouse. FBA adds four constraints that do not change the method, but do change its inputs.
- Capacity limits
- Amazon caps how much volume your account may store per storage type. The limit is granted monthly in cubic metres and depends on, among other things, your IPI score, sales history and Amazon's own demand forecast. The practical consequence: your order quantity can be arithmetically correct and still impossible to store. Check your free limit before you order – not just before you ship. Seller Central shows the current threshold and your limit for the coming month.
- Inbound lead time
- Between shipping to the fulfilment centre and the goods being sellable sit transport, delivery windows, receiving and put-away. Those days belong in the lead time. They are also the part you can most realistically shorten yourself: correct carton labels, exact quantities per carton, early appointment booking and a split that matches Amazon's distribution requirements.
- Q4 seasonality
- In the fourth quarter sales rise and receiving times get longer at the same time – both inputs of the formula deteriorate together. Amazon publishes a cut-off date each year by which goods for the Christmas season have to be at the fulfilment centre. Plan backwards from that date, not forwards from today's stock, and calculate Q4 with last year's fourth quarter instead of the current running average.
- Storage cost as the counterweight
- Safety stock is not free: monthly storage fees, considerably higher rates in the fourth quarter and surcharges for aged inventory. The question is never “much or little buffer” but what an extra month of coverage costs per unit – measured against what a day without stock costs.
- Several channels, one stock pool
- If you also sell on eBay, Kaufland or TikTok Shop, average daily sales have to be calculated across all channels while the reorder point is kept per storage location. Otherwise one channel drains the shared pool and creates the stockout on another.
Short answers
How long can a stockout last before the ranking suffers?
There is no grace period. Sales velocity feeds into the ranking continuously, so every day without sales has an effect. How strong an effect depends on competition in the category: in crowded niches the position is taken faster than where there are few sellers.
Which service level makes sense?
For most A items, 95 to 98 percent. 99 percent only pays off when a stockout is disproportionately expensive – for instance for an item that carries a bundle or a subscription. C items with thin contribution margin are deliberately run lower, otherwise you are financing storage fees for goods that barely earn anything.
Isn't a fixed 30-day buffer good enough?
As a stopgap yes, as a method no. For an item with a 14-day lead time and stable demand, 30 days is capital tied up for nothing – and for 70 days of sea freight with an unreliable supplier it is far too little. The buffer has to come from the variability of the individual item, not from one number applied to all of them.
How often should the reorder point and safety stock be recalculated?
Monthly at the very least, and always when an input changes: new supplier, different transport route, change of season, changed capacity limit. A reorder point that is a year old describes a product that no longer exists in that form.
What can you do once the stockout has already happened?
In this order: keep the offer alive instead of letting it run dry. Switch to FBM where possible so the listing stays buyable and the Buy Box does not disappear entirely. In parallel, push a partial quantity by air freight even though it costs more per unit – the number to compare it against is not the normal freight rate but the cost of the restart without it. Only then go looking for the cause in your calculation.
Calculating is step one. Keeping it current is step two.
The formulas above work in a spreadsheet – for ten articles. From the fiftieth onwards, what fails is not the maths but the fact that sales data, open purchase orders, inbound shipments and capacity limits live in four different places and nobody reconciles them daily. That is what Octuno is built for: forecast, reorder point and restock suggestion are recalculated every day, the FBA limit feeds into the suggested quantity, and suppliers and order lead times come from the same system.
Octuno is currently in closed beta with a maximum of 30 testers. Taking part is free.