Guide · Fulfillment models
Supplier-direct forwarding buys the goods after the customer orders and forwards the supplier’s parcel; stocked fulfillment dispatches from merchant-owned inventory already at the warehouse. The choice is per SKU, not per store: forward anything still being tested, stock only what repeats predictably. Routing is decided per order line, so one order can use both.
Once you rule out handing your suppliers to a sourcing platform, these are the two China-leg models that remain. The common mistake is treating them as a store-level either/or. Goods stay merchant-owned in both, and switching a SKU between them changes nothing about how the store runs.
| Dimension | Supplier-direct forwarding | Stocked fulfillment |
|---|---|---|
| Cash committed up front | None — you buy after the customer orders. | A full stock purchase per SKU, plus storage while it sits. |
| Dispatch trigger | The supplier parcel arriving at the warehouse and being received. | The order reserving stock already on the shelf. |
| What the customer waits for | Your supplier’s domestic leg, then the international leg. | The international leg only. |
| Storage charged | None. Nothing is held, so the storage layer is exactly zero. | 60 days free from receipt, then $0.50 per m³ per day, computed from the stock ledger and itemised monthly. |
| Fulfillment operation charged | Carried in the lane as a per-shipment handling component rather than a separate published fee. | $0.99 per order + $0.20 per item for picking, packing and measuring the outbound order. |
| Prep work (relabel, polybag, FNSKU) | Per parcel, as it passes through. | Once at intake, amortised across every order that draws on the lot. |
| Risk if demand vanishes | None beyond the single order you already sold. | The stock, plus storage accruing until it is withdrawn or sold. |
| Evidence trail | Receiving scan, photos and measurements per parcel. | A ledger entry per movement, plus cycle counts that reconcile physical stock against it. |
| Inventory truth | There is none to keep — the parcel is the unit. | Levels and the monthly statement are computed from the same immutable movement ledger, so they cannot drift apart. |
| Reversibility | N/A — it is the default state. | Let stock run out, or withdraw or transfer it. Each movement is recorded, and routing follows live inventory either way. |
One clarification that trips people up: “supplier-direct forwarding” does not mean your supplier ships to your customer. The parcel stops at your China warehouse address first — that stop is what makes inspection with photo evidence, consolidation and a measured quote possible.
Break-even is specific to you: freight is lane-based and your storage line depends on the volume you actually hold. What generalises is the signal. Measure each signal against your own order history, and the routing it points to follows.
| Signal you can measure | What it points at | Why |
|---|---|---|
| Fewer than a handful of sales a month, or still testing | Keep forwarding. | A SKU that never sells costs you nothing but the listing. Stock on an unproven SKU is the single most expensive mistake available to a small store. |
| Repeat orders arriving predictably, and the supplier leg is your slowest step | Stocking candidate. | Stocking removes exactly one thing: the wait for a supplier shipment. If that wait is not your bottleneck, stocking buys you nothing. |
| Bulky or heavy relative to its price | Keep forwarding. | Storage is charged by volume held per day ($0.50 per m³ per day after the free window), so a low-value item that occupies a lot of m³ pays for shelf space out of a thin margin. |
| Seasonal, and the season is closing | Keep forwarding, or withdraw existing stock. | Storage accrues whether or not the SKU is selling. A seasonal item past its window is the classic case of paying rent on something that has stopped earning. |
| Prepped the same way every single time (relabel, bundle, sticker) | Stocking candidate. | Prep done once at intake is amortised across every order that draws on the lot, instead of being repeated per parcel. |
| Sells steadily but reorders are lumpy and hard to predict | Stock a small buffer, keep the tail on forwarding. | Routing degrades gracefully: if stock runs out mid-demand, lines fall back to forwarding rather than overselling silently. |
The full fee structure behind these numbers — which layers exist, what each is measured in, and which figures are account-scoped — is in What China fulfillment actually costs.
Zero commitment: list it, sell one, buy one, forward one. A SKU that never sells costs you nothing but the listing.
Sell-through per month and reorder frequency. When a SKU repeats predictably and the supplier leg is your slowest step, it is a stocking candidate.
Register the SKU, pre-alert the inbound, keep everything else on forwarding. The store changes nothing; routing follows inventory.
If stock runs out mid-demand, lines fall back to forwarding (or pause for your call in mixed orders) instead of overselling silently.
Every statement in this guide is about a mechanism, which means it is testable on a real account rather than something you have to take on trust.
| Claim | How to check it |
|---|---|
| Stocking does not transfer ownership of your goods | Read the terms for any transfer of title. Goods stay merchant-owned in both models, and withdrawal and warehouse transfer are supported operations rather than negotiated favours. |
| Storage is derived from the ledger, not estimated | Compare a monthly statement’s per-lot storage lines against the movement history for the same lots. They are computed from the same ledger, so they have to agree. |
| A mixed order is never split silently | Place an order containing one stocked line and one unstocked line. It should pause for a recorded decision — forward all, wait for stock, or re-evaluate — rather than dispatching half. |
| Routing follows live inventory, not a cached number | Let a stocked SKU run to zero and place another order. The line should route to forwarding automatically rather than reserving stock that is not there. |
| The storage layer really is zero when you hold nothing | Run a month of pure forwarding and read the statement. A storage line on an account holding no stock is a billing bug, not a policy. |
Decide per SKU. Forward anything still being tested, seasonal, long-tail, or bulky relative to its price. Stock only SKUs that repeat predictably and where waiting for the supplier is genuinely your slowest step. Because routing is per order line, the same store — and the same order — can use both.
It depends on the SKU, and any page giving you one number is guessing. Forwarding avoids storage entirely but every order waits on a supplier shipment; stocking removes that wait but ties up cash and accrues storage at the published rate of $0.50 per m³ per day, after 60 days free from receipt. The honest comparison is sell-through and reorder frequency against your own storage lines.
No. Goods stay merchant-owned the entire time, and inventory levels plus the monthly statement are computed from the same immutable stock-movement ledger — so what you are billed for and what you believe you hold cannot drift apart. You can withdraw or transfer stock, and each movement is recorded.
Yes — routing is per order line, not per store. A stocked bestseller and a forwarded long-tail item can share one order. When that happens the order pauses for your explicit choice — forward all, wait for stock, or re-evaluate — rather than being split automatically.
Register the SKU, pre-alert an inbound shipment so the warehouse receives against an expectation rather than a guess, and once it is counted in, orders for that SKU reserve stock automatically. Switching back is just letting stock run out, or withdrawing it — routing follows live inventory either way.
Anything still being tested, seasonal items past their window, and goods whose storage cost outweighs the dispatch-time gain — typically oversized or slow movers. The point of per-SKU routing is that these keep working as forwarded parcels with zero stocking commitment.
Lines fall back to forwarding, or pause for your call inside a mixed order, rather than overselling silently. Routing is evaluated against live inventory at the moment the order arrives, so an empty shelf produces a routing decision instead of a promise you cannot keep.