What actually changed
The change is not the rate. It is that every shipment now needs a formal customs entry, and most of the cost of a formal entry is fixed — filing, documentation, brokerage — whether the shipment holds one unit or a thousand. Spread across a single parcel that fixed cost is punishing. Spread across a container it is close to noise.
So what has been removed is not "cheap". It is the one-parcel-at-a-time model itself. The arithmetic that made it work no longer holds, and no amount of shopping for a cheaper courier brings it back.
What the replacement model looks like
The mainstream route has three legs: consolidate domestically into a container, clear it formally once on entry to the US, then fulfil locally from a US warehouse. The clearance cost is spread across the whole container, and local delivery is faster than cross-border post — so the customer experience improves rather than degrades.
The barrier is not logistics, it is inventory. A container means committing stock in the US: how much, for how long, and what happens if it does not move all become cash questions. That is why the same route works for some factories and sinks others.
You do not have to arrive there in one step. Start with a third-party warehouse charging per unit, get volume and sell-through data, then consider your own space or a long-term commitment. Doing it in the other order spends money on an assumption you have not tested.
What an importer of record is, and why you cannot avoid one
The importer of record is the party that carries declaration and legal responsibility for the shipment in front of US customs. Every entry needs one. It can be your US customer, an entity you set up yourself, or a third-party service.
The three choices differ a lot. Letting the customer be IOR is easy and costs you control of the channel. Setting up your own entity is expensive up front and keeps the pricing power and the customer relationship with you. Using a service is quick to start, but read carefully how liability is divided and who carries it when something goes wrong.
Worth stressing: IOR is a legal responsibility, not an errand. Description, HS code, value, origin — if any of them is wrong, the IOR answers for it. So deciding who holds it is really answering the question of whose channel this is.
Marketplace, semi-managed, or your own channel first
Decide whether you want cash flow or brand equity. Marketplaces and semi-managed models bring volume quickly but cap your price band and keep the customer data. Your own channels and physical retail move slower, but the pricing and the relationships are yours. Most factories should prove the product in one channel before opening three.
A semi-managed model solves fulfilment and traffic for you; the price is that you have almost no say over the end price and the platform can change the rules whenever it likes. Treating it as a way to validate product and move volume is reasonable. Treating it as strategy is dangerous.
The test is simple: when this channel has run its course, what is left in your hands? If it is only cash that has already been spent, it was a tactic. If it is a customer list, repeat-purchase data and brand recognition, that is an asset.
Which categories suit a US warehouse, and which do not
Suited: high enough unit value, sensible size and weight, predictable sell-through, and not hostage to season or trend. Those four conditions decide whether stock sitting in the US turns over in a reasonable time.
Not suited: oversized, heavy, highly customised, or products with a two-month trend life. For those, the inventory risk exceeds the clearance saving — you have simply swapped a tariff problem for an inventory problem.
Anything in between should start with a trial: stock one month of demand, run it for three, and look at sell-through and returns before committing further. That money is well spent, and far safer than a container bought on a hunch.