A founder receives two quotes. Direct shipping is expensive per order. A US warehouse promises lower domestic fulfilment costs. The apparent decision is to move stock closer to the buyer.
That skips the most expensive assumption: enough buyers will keep the product to absorb the commitment. A fulfilment model buys a different combination of variable cost, fixed cost, inventory exposure and customer promise. It does not buy demand.
The decision is therefore two questions in sequence: which entry and service arrangements are available for this product now, and at what volume does one feasible model earn its additional commitment?
Start with the rule in force, not an old parcel habit
As checked on 9 November 2026, CBP's June rules suspend duty-free de minimis treatment for non-postal and postal shipments. A low value by itself is not a basis for budgeting zero duty or entry work. The non-postal rule and postal rule describe different processes; “send it by mail” is not a complete alternative entry plan.
The 22 October postal amendment moves compliance for two specified postal provisions to 20 January 2027. It does not restore the exemption. A headline about a delay cannot safely become a zero-cost assumption in a spreadsheet.
This article does not provide a current tariff rate. Classification, origin, applicable measures, entry method and product-agency requirements need a product-specific review and a dated quote. The useful commercial model keeps those inputs visible so a change can be recalculated without rebuilding the argument.
Compare two complete ways of serving the same order
Fictional planning case. All products, prices, costs, volumes and operating conditions below are invented. Customs-related dollar amounts are supplied scenario inputs, not current US rates or advice on a real entry. The example assumes both routes have first been confirmed feasible for the same product and customer promise.
A Southeast Asian accessories brand sells one adult travel organizer for $80. It is deciding between direct shipment from origin and a small US-stock pilot. Both figures below cover one retained, paid order, before acquisition spending, fixed overhead and income taxes. Sales tax collected for a government is outside revenue and cost here.
| Cost per retained order, USD | Direct from origin | US-stock pilot | What the input includes |
|---|---|---|---|
| Goods and packaging | 22 | 22 | Goods $20 + packaging $2, same in both routes |
| Transport and fulfilment | 18 | 14 | Direct delivery; or $6 allocated inbound plus $8 domestic pick/ship |
| Customs service and duty/tax input | 13 | 10 | Direct $5 service + $8 assessment input; stocked $2 allocated service + $8 assessment input |
| Payment cost and expected return/failure reserve | 9 | 8 | $2 payment cost; reserves of $7 or $6 |
| Total variable cost / contribution from $80 | 62 / 18 | 54 / 26 | Contribution is not business profit |
The reserve is a probability-weighted allowance for refunds, reverse logistics and unrecovered goods across the served cohort, expressed per retained order. The goods line covers the retained item; the reserve covers additional losses from failed or returned orders. Those losses are not added again later. A real team must rebuild the reserve from its own definitions and records, rather than importing these fictional dollars.
The US-stock route appears $8 better per retained order. But it also requires $900 of incremental, non-recoverable setup for the pilot. That fixed cost is not included in the $54. Ordinary inbound and entry costs are already allocated in the table and must not be charged a second time as “setup.”
The threshold is $900 ÷ $8 = 112.5: 113 retained equivalent orders to exceed the extra setup cost, assuming every input holds. At 60 retained orders, the incremental comparison is 60 × $8 − $900 = −$420. At 150, it is +$300. Neither number is total profit; both describe only the difference between these two simplified routes.
A smaller shipping bill can therefore accompany a worse launch result. That is an arithmetic consequence of commitment, even before considering stock that does not sell.
Add the inventory decision the unit table cannot answer
Suppose the pilot requires buying 200 units at $20 goods cost each: $4,000 leaves the bank before customer receipts, with packaging and other commitments additional. The $20 remains the goods cost in the unit table. The $4,000 is a cash requirement, not a new expense to add on top of all 200 unit costs.
If only 60 units are retained, 140 units remain available, returned, damaged or otherwise unresolved depending on the actual operating record. Do not assume all 140 are worthless—or that all are immediately saleable at $80. Their condition, storage period, rework and recoverable value require a separate stock scenario. Any write-down absent from the earlier reserve must be added once, with its basis stated.
This is why the 113-order hurdle is necessary but insufficient. A brand could clear it eventually and still run out of cash while waiting. It could also discover the wrong variant, packaging or promise after committing most of its stock.
| Route | What it buys | What makes it unattractive | Evidence needed before commitment |
|---|---|---|---|
| Direct shipment | Smaller stock commitment and learning by order | Higher variable cost; delivery and remedy limits | Dated landed-service quote and a supportable customer promise |
| US stock | Lower assumed variable cost and local fulfilment options | Setup, cash tied in stock, ageing and forecast error | Retained-demand evidence, stock disposition and cash plan |
| Marketplace route | Access to a particular demand environment and service system | Fees, eligibility and operational duties can change the comparison | Actual market/category eligibility and complete fee/service scope |
| Delay launch | Time to resolve an invalid assumption | No sales learning during the delay | A specific unresolved blocker, owner and decision date |
The marketplace row is deliberately unpriced. A marketplace fee does not necessarily include fulfilment, entry, returns or customer support. Add only services actually purchased and avoid charging an included service twice. Availability and requirements must be checked for the actual seller and market.
A decision memo that makes the disagreement useful
For the fictional brand, the founder can write:
Pilot decision: retain direct shipment for the first bounded demand test. Authorize US-stock setup only after a refreshed comparison supports at least 113 retained equivalent orders within the cash planning window, and the team can fund the stock commitment and a downside disposition. Do not use paid-order count alone: reconcile cancellations, delivery failures and returns before calling demand retained. Refresh customs and service inputs before each commitment; a cheaper route that cannot support the advertised delivery/remedy promise is not eligible.
This is conditional, not a claim that direct shipping is generally superior. A retailer with committed replenishment orders, a costly delivery failure pattern or reliable local resale may choose stock earlier. A low-volume seller with high value per order might tolerate a higher parcel cost. A product with unresolved entry or safety requirements should not treat either route as ready.
The next useful conversation with a logistics or customs partner is not “what is your cheapest rate?” It is: for this product, origin, quantity and delivery promise, what is included, which costs vary, which cash commitments survive a failed test, and when does this quote expire? The business model becomes reviewable when those answers are separate.