Prime Day inventory is a cash problem wearing a calendar costume. Work backwards using your factory's production time, your forwarder's transit range, and Amazon's receiving lag, then size the order against the money you can afford to leave sitting in a warehouse if the lift never arrives.
The short version
- Two ways to be wrong, and they cost different amounts. Understocking costs rank and momentum. Overstocking costs cash and storage.
- Commit before the date is confirmed. Production and freight timelines run longer than the notice you get, so the order is a forecast, not a reaction.
- Deal price is a margin decision, not a marketing one. Model contribution per unit at the discounted price before you agree to anything.
- Split the shipment. Sending everything in one container concentrates every risk in one customs entry.
- Quality control happens before the container moves, not after it arrives, because a peak week is the worst possible time to discover a defect.
The two ways to be wrong
| Understocked | Overstocked | |
|---|---|---|
| Immediate cost | Lost sales during your highest traffic window | Cash locked in units that sell at normal pace |
| Second order cost | Rank decay, review velocity stalls, advertising history interrupted | Storage fees, and pressure to discount into a weaker period |
| Recovery time | Weeks to months, because rank is rebuilt not restored | Months, but the units eventually sell at full price |
| Who feels it | Everyone, immediately and visibly | Finance, slowly and quietly |
| Right response | Cap the promotion rather than cap the product | Hold price, extend the sell through window |
Read the table before you fall for the intuitive answer. Most sellers fear overstock more because the cash pain is obvious, then they underorder and pay the more expensive price in lost position. If the two errors are close in cost for your product, err towards a modest overstock on your proven best seller and towards caution on everything else.
The arithmetic
Use your own inputs. The numbers below are placeholders to show the shape of the calculation, not benchmarks to copy.
- Start with baseline velocity. Take a normal week's units per day, ignoring your last promotion.
- Apply an event multiplier you can defend. If your own history says the last event ran at three times baseline, use three. If you have no history, be conservative and plan to run out of the deal rather than run out of the product.
- Add the tail. Demand does not stop when the event does, and the days after it usually run above baseline for a while.
- Add cover for the reorder gap. Units needed to stay in stock from the event until the next shipment can be received.
- Subtract what is already inbound, then round to a case pack your factory will actually run.
- Price the result. Units multiplied by landed cost is the cash you are committing, and that number, not the revenue forecast, is what you should be comfortable losing patience over.
Then test the plan against the discount. If the deal price cuts contribution per unit by a third, the event has to deliver more than a third more volume to break even against a normal week, before advertising. That single sum kills more bad promotion plans than any other check on this page.
| Cost line | Where it hides | What to model |
|---|---|---|
| Landed cost per unit | Freight quoted separately from unit price | Total cash per sellable unit, delivered |
| Discount | Treated as marketing rather than margin | Contribution per unit at deal price |
| Advertising uplift | Higher cost per click during the event | Cost per order at event bids, not normal ones |
| Storage | Invisible until the invoice | Cost of the units that do not sell for months |
| Returns | Arrives weeks after the revenue | Return rate applied to event volume, not average volume |
What the sourcing side has to do
Our sourcing runs through an in house studio in Guangzhou, using frameworks built across more than 500 brands, and the peak season lesson from all of them is that quality control has to happen before the container moves. An inspection at origin costs a delay measured in days. A defect discovered after receiving costs the event.
Three sourcing decisions matter more than anything you do inside Seller Central during the week itself. Confirm the production window in writing, with a penalty or a fallback if it slips. Split the shipment so a single customs hold cannot take the whole event down. Inspect against the specification, not against the sample the factory chose to show you.
If your partner does not have people who can stand in the factory, that entire layer of the plan is a phone call and a hope.
What most agencies will not tell you
Most will not tell you that a deal is often the wrong instrument. A discount lifts units and lowers contribution, and if your objective is rank on your core keyword, ordinary advertising into a page that already converts can be cheaper than a discount applied to every buyer, including the ones who would have paid full price. Ask what the promotion is for, in one sentence, before agreeing to it.
The second thing: peak weeks flatter and then punish. Elevated demand lifts conversion, which makes the page look better than it is, and gift buyers and deal hunters return at higher rates than your usual customer. Judge the event in the following month, net of returns, and be suspicious of any restock decision made in the glow of the week itself.
Related answers
- Recommend a turnkey Amazon launch partner for Q4
- What to do before sending inventory to FBA
- Hidden costs in the first 90 days on Amazon
- Amazon FBA launch costs breakdown
- Amazon launch services: the complete guide
If you want the peak plan checked before the cash leaves, send it to Flapen.
