The loop nobody closes

The definition has been stable for decades: unit cost plus every freight leg, insurance, duties and tariffs, brokerage, port handling, documentation, financing, and currency effects. The execution has not caught up. A supply-chain software whitepaper puts the state of practice plainly: ninety to ninety-five percent of importers run landed cost on a standard accounting application supplemented by spreadsheets — producing, at best, an average, from information that arrives too late to act on. Importers who do estimate landed cost at purchase-order time rarely close the loop when the forwarder's actual invoice shows up with different numbers; the small fixed fees often never make it into the spreadsheet at all. And the visibility numbers behind that are brutal: in one survey of 623 companies, six percent claimed full supply-chain visibility.

Read that as a market description: the industry quotes a landed cost, pays a different one, and mostly never learns the difference. Every percentage point of that gap is somebody's quiet margin.

Contract manufacturing's dominant commercial model makes the problem structural. Under turnkey, the manufacturer buys your components and sells them back with 10–25% folded into the unit price — so the inbound logistics cost is invisible to the customer by design. Outbound, the typical quote ends at the factory dock: EXW or FCA, freight and customs pushed onto the buyer. Freight forwarders fill the gap and live on it — 10–30% markup over carrier cost is the standard structure, though only 2–4% survives as net margin. The giants have noticed there is a business here: Flex and Jabil both sell logistics and fulfillment as adjacent services, evolving into what one industry analysis calls supply-chain orchestrators. But orchestration is sold as a service engagement, not as a number on your quote.

The most instructive data point is the smallest one. JLCPCB — the prototype-scale board house — offers delivered-duty-paid at checkout: estimated duties pre-collected, no broker fees, one number. Customers demonstrably love it. Nobody has done the same for serial production. The demand was proven at a fifty-dollar order size; the offer stops there.

The surprising finding is how much of logistics is already machine-bookable. One API layer quotes duty and tax to two hundred destinations by HS code. Rate platforms reach roughly seventy percent of global air capacity with instant quoting and booking. Maersk sells guaranteed-loading ocean bookings that integrators report closing in about thirty seconds. Digital forwarders expose booking, tracking, documents, and customs over APIs. The asset-light shape this enables has a name — the 4PL, a control tower that coordinates carriers and brokers without owning a ship — and a software-defined 4PL is now a realistic architecture, not a metaphor. What stays hard is instructive: origin drayage in messy geographies, LCL consolidation physics, rolled cargo and customs exams, and contract capacity in tight markets, which is still negotiated annually by humans with relationships.

So the estimation half of landed cost is commodity plumbing. The rare capability is the other half: truing up the quoted number against carrier, broker, and duty invoices per shipment, per product, continuously — the loop the 90–95% never close. That, plus consolidation: buyer-consolidated inbound freight documents 15–30% savings, and the trigger is not company size but lane repetition — which is exactly what a network that routes many customers' production through shared hubs generates as a byproduct.

The thesis routes production near demand, and that changes the customs math at the root: a product built in the country that buys it has a domestic outbound leg — no duty, no broker, no border risk on the way to the customer. The cross-border complexity concentrates on inbound components, where a network can consolidate and where the discipline of the intake standard already names every part, origin, and supplier. This is why the economics page commits to the shape this research points at: freight, duty, and brokerage passed through at documented cost, one published orchestration fee in the Amazon-FBA style, a quoted landed cost on every RFQ, and a reconciliation against actual invoices on every shipment — variance on the record. The FBA precedent carries one more lesson worth stating: even Amazon needed an explicit fuel surcharge and annual repricing. Transparent does not mean fixed, and promising fixed through a market that moved fourteen-fold in one cycle is selling someone else's bankruptcy.

Three things stay hard, and they are named here so the register can hold them later. Customs liability does not follow the software: the importer of record carries classification, valuation, and duty exposure — US penalties run to multiples of the revenue loss, and EU indirect representation makes the representative jointly liable, which is why it is expensive and often refused. Being the importer per hub country is entities, bonds, and real liability — legal infrastructure, built one country at a time. Freight volatility is not a rounding error: Asia–US containers went from under $1,500 to over $20,000 and back down more than 80% inside three years, which is why quotes carry validity windows and then index to a public benchmark. And exceptions are still human — APIs book cargo; they do not argue a customs exam or rescue a rolled container. The published fee has to fund an operations team, or the promise fails exactly when it matters. None of this weakens the case; it prices it.

Visco — importers on spreadsheets (90–95%) · ERP Software Blog — the unclosed loop · Supply Chain Dive — 6% visibility (Geodis) · PCBCart — turnkey materials markup · GoFreight — forwarder margin structure · JLCPCB — DDP at checkout · Logistics Viewpoints — CMs as orchestrators · Zonos — landed-cost API · Maersk Spot — instant guaranteed booking · Freightos/WebCargo — rate and booking APIs · C.H. Robinson — buyer's consolidation · Freightos — the 14x cycle · StoneTurn — the DDP trap and IOR liability · Avalara — EU indirect representation · ShipBob — the FBA fee schedule precedent