Amazon Vendor Central Inventory Forecasting: How to Predict Amazon's POs and Plan 1P Stock

Inventory forecasting for Amazon Vendor Central is different from every other channel you sell in because you are not deciding how much to ship. Amazon is. As a 1P vendor you sell wholesale to Amazon, Amazon owns the inventory and sets the retail price, and Amazon's replenishment system issues you purchase orders based on its own demand forecast and its own target inventory levels. Your forecasting job is therefore twofold: predict the POs Amazon will send (size, ASIN mix, and timing) so you can have the units ready to ship inside the window, and predict the real consumer sell-through underneath those POs so you do not overbuild when Amazon front-loads or pulls back. Miss in one direction and you take shortage chargebacks and lose the Buy Box to a 3P reseller. Miss in the other and finished goods sit in a warehouse waiting for a PO that is not coming.

How Amazon's ordering actually works

Vendor Central POs typically arrive on a weekly rhythm, with each PO listing ASINs, quantities, a ship-to fulfillment center, and a ship window. You confirm the PO within a short acceptance window, indicating what you can fill and when. Confirmed units you then fail to ship, or ship late or short, create shortage claims and chargebacks. Amazon's forecast for your ASINs drives the PO quantities, and that forecast is visible to you in Vendor Central's forecasting reports, which show projected demand by ASIN over coming weeks, often as a mean estimate with probability bands.

Three consequences follow. First, Amazon's PO is a derived signal: Amazon's forecast minus its on-hand and inbound at its FCs, not consumer demand directly. When Amazon is sitting heavy after a slow month, POs dry up even if sell-through is steady. Second, Amazon's forecast reports are an input to your plan, not a replacement for it; Amazon's model does not know about your creator campaign launching next month. Third, the ship window is short relative to production lead times, so by the time a PO arrives it is far too late to start making the product. You are forecasting to have finished goods on the floor, not to react.

Reading Vendor Central reports as forecast inputs

Vendor Central exposes more demand data than most retail accounts. Use several reports together:

• Ordered units and shipped units by ASIN. Ordered units (consumer orders on Amazon) are your sell-through series. Shipped units lag it. Use ordered units as the baseline for velocity and seasonality.
• Amazon's demand forecast. Compare its mean and upper-band projections to your own baseline. Where Amazon is above you, expect larger POs than your model would predict, at least in the short run. Where it is below you, expect Amazon to under-order and run thin, which shows up as lost Buy Box days.
• Inventory health and weeks of cover. Amazon's on-hand by ASIN, sellable versus unsellable, and its implied weeks of cover tell you how far it is from needing to reorder.
• PO history. Submitted, confirmed, and received quantities by week. The gap between what Amazon asked for and what you confirmed is your fill rate history.
• Chargebacks and shortage claims by reason. Treat these as a cost of forecast error.

Predicting the next PO

The core of 1P inventory forecasting is a simple replenishment identity. Amazon orders enough to bring its position back to a target level of cover:

Expected PO units = (Amazon's forecast weekly demand x Amazon's target weeks of cover) minus Amazon on-hand minus Amazon inbound

You do not know Amazon's exact target, but you can back into it from PO history: divide recent PO quantities plus on-hand by weekly ordered units and the implied cover tends to settle into a range. Then project POs forward from your own sell-through forecast and see where you and Amazon's report disagree.

The second calculation is yours, not Amazon's: how much finished goods inventory to hold so that you can confirm POs in full.

Reorder point = (average weekly PO demand x production lead time in weeks) + safety stock

Weeks of supply = finished goods on hand / average weekly PO demand

Safety stock in 1P should be higher than in DTC for the same SKU, because the cost of a miss is a chargeback plus a confirmed-short record, and because Amazon's POs are lumpier than consumer demand. A common approach is to cover the gap between a normal PO and the largest PO of the last quarter.

A worked example: forecasting POs on one ASIN

Take a fictional snack brand we will call Ridgeline Provisions and its Sea Salt Almond 6-pack, which sells 1P on Amazon alongside DTC and grocery. The planner is building the December plan at the start of November.

• Ordered units over the last 8 weeks average 2,400 per week and are trending up 5 percent month over month. Seasonal index for December is 1.4, so forecast sell-through is roughly 3,500 units per week through mid-December.
• Amazon's forecast report shows a mean of 3,200 per week and an upper band of 4,100. Close to the planner's number, slightly conservative on the mean.
• Implied target cover from PO history: about 4 weeks. Amazon currently holds 6,800 sellable units with 2,400 inbound on a confirmed PO.

Expected PO this week = (3,500 x 4) minus 6,800 minus 2,400 = 14,000 minus 9,200 = 4,800 units. The following week, with Amazon having sold through 3,500 and received the 2,400, the position is roughly 5,700 on hand plus the 4,800 just confirmed, and the formula gives another PO near 3,500. Amazon's POs step up from a trickle to roughly a week of demand each through the peak.

Now Ridgeline's side. Over the 6-week co-manufacturer lead time, expected PO demand sums to about 22,000 units. Safety stock policy on 1P hero SKUs is 3 weeks of PO demand, or 10,500. Reorder point = 22,000 + 10,500 = 32,500 units of finished goods. Ridgeline has 18,000 on hand allocated to Amazon and 12,000 in production landing in two weeks, for 30,000 available. That is below the reorder point, so a production order goes out now, sized by the method in the purchase order forecasting guide. Wait for the first big December PO before ordering and the units arrive in mid-January.

Cash check: at a landed cost of $1.85 and a production order of 20,000 units, that is $37,000 committed in November. Amazon's terms on this account are Net 60 from invoice, and the invoice goes out at ship, so the cash from the December POs arrives in February. That is the working capital cost of growing a 1P account, and it belongs in the cash forecast before the production order is placed.

Chargebacks, shortage claims, and the cost of being wrong

Vendor Central penalizes supply execution through chargebacks and claims, and forecasting is the first line of defense against most of them. Confirm the current fee schedule in your agreement, since Amazon updates its vendor terms, but the shape is consistent:

• Shortage claims when Amazon receives fewer units than invoiced. Often a receiving dispute, but a true short shows up here too.
• PO-related chargebacks for confirming a PO and then not shipping, shipping outside the window, or ASN (advance ship notice) errors.
• Prep and compliance chargebacks for carton labeling, packaging, and routing violations.

Forecasting touches the first two directly. If you can see the December PO wave coming six weeks out, you confirm in full and ship on time. If you cannot, you either confirm short or decline units and watch Amazon's forecast for your ASIN drift down, which lowers future POs. Either way, model chargebacks as a percentage of 1P gross sales in the P&L; they move with forecast accuracy.

1P versus 3P economics, and why it changes the inventory plan

Many brands run both Vendor Central and Seller Central, and the economics differ enough that the inventory plan should too.

In 1P, your revenue is the wholesale cost Amazon pays, less contractual allowances (co-op, freight, damage, and marketing accruals), less chargebacks, with payment typically on extended terms. Amazon controls the retail price and can discount to match the market. You carry finished goods until Amazon orders them; Amazon carries them after. Your inventory risk is overbuilding ahead of POs that shrink.

In 3P, you sell at retail, pay a referral fee (typically a percentage of the sale that varies by category), pay FBA fulfillment and storage fees, and typically collect on a biweekly disbursement. You control price. You decide how much to send to FBA, subject to storage limits. Your inventory risk is stranded or aged FBA stock and long-term storage fees, plus stockouts if inbound slips.

The practical difference is who holds the stock and when cash arrives. A 1P-heavy brand needs more finished goods on its own floor and a longer cash cycle; a 3P-heavy brand needs tighter FBA inbound planning. If you run both, keep them as separate lines in the SKU forecast and model the channel P&L separately. The Amazon 1P P&L template lays out the gross-to-net for the vendor side, including allowances and chargebacks, so the inventory plan ties to a margin number you trust. For the broader Amazon forecasting method, including the 3P side, see our post on Amazon inventory forecasting.

Putting the plan together

A working 1P inventory forecast, refreshed weekly, has five pieces: a sell-through forecast by ASIN from ordered units; Amazon's forecast report alongside it as a check; an expected PO projection from the replenishment identity; a finished goods plan with reorder points and safety stock sized to PO lumpiness; and a cash line showing production outlays against Net terms collections. Score it monthly on two numbers: fill rate on confirmed POs, and forecast bias on your PO projection. If bias is consistently negative, you are under-forecasting Amazon and your chargebacks will say so. If it is consistently positive, you are carrying inventory for POs that never arrive, and your cash flow will say so. The SKU-level demand forecasting guide covers how to run the same discipline across every channel at once.

Where Drivepoint fits

Drivepoint is the AI finance platform for consumer brands, and it connects Amazon Vendor Central and Seller Central data alongside Shopify, retail, distributor, and 3PL sources through 100+ integrations into one Excel-native model. Ordered units, PO history, and Amazon on-hand flow into a SKU forecast; the forecast flows into finished goods planning, production POs, and the cash they consume on Net terms. When Amazon's forecast for an ASIN jumps or a PO wave arrives early, you rerun the scenario and see the production and cash impact in minutes, not days. It is the same connected model that helped Oats Overnight increase EBITDA by $4M.