Inventory Forecasting: How a Multi-Channel Brand Decides How Much Inventory to Ship

A forecast is a decision about cash and about stockouts. It is not a property of the spreadsheet. International brands miss this in a specific way: the factory’s production lead time, the ocean transit, customs under the current de minimis rules, and Amazon’s receiving time are four different clocks, and the purchase order has to respect the slowest one.

AEG plans replenishment with partners who are selling on more than one US channel. The method below is the one we will defend. It does not include a formula that claims to predict your fourth-quarter units. Nobody honest has that formula for a brand the US market is still learning.

The clocks

Write them down for the hero SKUs before you argue about a growth rate.

• Production. From a confirmed order to goods ready at the factory. Use the last two orders, not the quotation.
• International freight and customs. Ocean or air, plus brokerage. Duty-free de minimis treatment is suspended, so a small parcel is not a way around this clock. The legal status is in the market-entry essay.
• US receiving. A 3PL’s appointment, or an FBA check-in. FBA receiving is not the sailing time. It is additional. Amazon’s inbound placement options change the cost, and Amazon Warehousing and Distribution is a different building with placement included in its own rates. The published figures are in the inbound and AWD essay. Do not inherit a setting from a 2024 screenshot.
• The channel’s cover policy. FBA will charge storage, and aged-inventory surcharges, on units that sit. A 3PL will charge its own storage. Walmart Fulfillment Services has its own inbound rules, including a requirement that goods ship to WFS from inside the US or clear customs first. See the fulfillment essay.

The reorder date is today plus the sum of the clocks you do not control, subtracted from the day you would otherwise stock out. If that date is already in the past, you are air-freighting or you are accepting a stockout. Decide which, on purpose.

One number, then the reasons it is wrong

Start with recent weekly sales of the hero SKU, by channel. Multiply by the weeks of cover you are willing to finance. That is the base. Then adjust, in writing, for the only things you actually know:

• A promotion you have already scheduled, with an end date. Not a hope that ads will “scale.”
• A retail set or a wholesale order that is confirmed.
• A season you have now seen once in the US. One season is a hint. It is not a curve. The retail markers are in the launch essay.
• A stockout in the history. Units you did not sell because you were dark are not a demand of zero. They are missing data. Mark them. Do not average across them as if the brand was unwanted that week.

If you cannot name the adjustment, do not type it. A 40 percent uplift because a similar brand had one is fiction. We will not supply a benchmark to make the cell less empty.

Where the units should sit

Split the base across the places that ship, not evenly.

• FBA gets the cover that protects the featured offer on Amazon, and no more than you can sell before storage dominates. The featured-offer stakes are in the protection essay.
• The 3PL gets the replenishment for FBA, the store, Walmart if you are not using WFS, and wholesale.
• WFS gets its own inbound if you use it. Do not assume FBA units can be redirected to Walmart. They cannot, not quickly, and often not at all.
• The factory gets the next production only. Finished goods sitting in the home country are not US cover. They are a different SKU from the point of view of a US customer who wants the item this week.

Too many buildings is a common failure. Every extra location is a forecast of its own, and a chance to have the right total and the wrong building. Start with two: production, and one US pool that feeds FBA.

What to review, and when

Monthly is enough until a SKU is a real share of the business. Weekly during a launch or the fourth quarter. The review is four lines.

1. Weeks of cover in the building that ships, not in transit.
2. Units in transit, and the week they can actually be picked.
3. The next reorder date, given the clocks.
4. Any SKU that will stock out before that reorder can land. Those get a decision: air, a smaller ad budget, or a substitution you announce on the listing.

Amazon’s restock recommendations are an input. They see Amazon demand, not your wholesale order, not your store, and not the container you already booked. Use them as a check. Do not obey them against your own sheet.

Cash is the constraint the sheet forgets

Units you cannot afford to sit on should not be in the base “because FBA likes depth.” Depth you cannot fund becomes aged inventory and a distressed price, which then damages the brand you are forecasting for. The pricing essay is what that distressed price costs.

A factory has a temptation here. Production is “free” in the sense that the plant is already running. It is not free in a US brand. Every extra case is cash, duty, and a storage bill in someone else’s building. Make the US brand’s forecast the one the plant runs to, not the other way around.

The first forecast is allowed to be rough

A new US brand does not have a year of history. Use the smallest cover that still survives the clocks, advertise inside that cover, and rewrite the sheet when you have eight real weeks. Expanding the purchase order because the first two weeks were encouraging is how brands create the aged inventory they then discount. Wait for the reorder date you already calculated. If the product is selling through faster than the clocks can refill, that is an air-freight decision, made once, not a new permanent growth rate.

Book a strategy call

If the US purchase order is still a guess, send a brief or Book a strategy call.