Freight is usually treated as a line item to minimise. It is better treated as a decision about inventory: a slower route means more capital sitting in containers and a longer gap between a demand signal and a response to it. Two companies with the same product can rationally choose different routes.
Sea
The default for anything heavy, bulky or low-value per kilo. A full container is the cheapest way to move goods that exists; a part container (LCL) is far less advantageous than the volume ratio suggests, because fixed handling charges at both ends do not shrink with your pallet count. Transit times are long and, more importantly, variable — plan on the distribution, not the average.
Air
Expensive per kilo and priced on volumetric weight, which punishes bulky packaging harshly. It earns its cost in three situations: a launch where being late costs more than the freight, a replenishment of something that is selling faster than the sea cycle allows, and a corrective shipment after a quality problem. Using air as a routine mode is usually a symptom of a planning problem rather than a logistics choice.
Rail
Faster than sea, cheaper than air, and geographically sensible for goods made inland rather than on the coast. It is more exposed to route disruption than either alternative and capacity varies, so it works best as a planned share of your volume rather than as the whole of it.
The questions that actually decide it
- What does a week of delay cost you, in lost sales and in stock-out risk? If the answer is « nothing much », sea wins every time.
- What is your product's value per cubic metre? This ratio, not the unit price, decides whether air is even discussable.
- How reliable is your demand forecast? Poor forecasting and slow freight compound into either dead stock or empty shelves.
- Where does the incoterm place the handover? It determines who chooses the route and who absorbs a disruption.
Chinese New Year distorts all three routes for several weeks: factories close, everyone ships before the closure, and rates and space follow. Build the calendar first, choose the route second.