The question usually arrives as a rate comparison: what does air cost, what does sea cost, which is cheaper. Air is cheaper than sea on almost no shipment, so the comparison answers itself and settles nothing. The decision worth making is whether the consignment in front of you can tolerate a sea transit, and what that intolerance is actually worth in money. That is a four-step test, and freight cost is the last step rather than the first.
Start with the gate that overrides economics: can the cargo travel by that mode at all. Lithium batteries are treated very differently under the IATA Dangerous Goods Regulations than under the IMDG Code, and some cell types, packing configurations and states of charge are restricted or refused on aircraft while moving routinely by sea. Aerosols, flammable adhesives, magnetised material and several classes of chemical sit in the same territory. Establish acceptance before pricing anything, because a rate for a mode that cannot legally carry the goods is not an option.
Then compare freight cost per kilogram against goods value per kilogram. Freight representing a small fraction of the value of a kilogram of cargo barely moves the landed cost; freight that approaches or exceeds it changes the commercial case entirely. Instruments, components and finished electronics sit comfortably at the first end of that range. Furniture, ceramics, packaging and bulk consumables sit at the other, and no degree of urgency makes air economic for them beyond a single emergency tranche.
The two modes also price on different units, which is why identical cargo can look reasonable on one and absurd on the other. Air freight charges the greater of actual and volumetric weight, conventionally derived at 6,000 cubic centimetres per kilogram, so a light bulky carton pays for space it does not use in weight terms. Sea FCL charges by the container regardless of the contents, and LCL by the greater of cubic metre or tonne. Measure the packed cargo before either comparison means anything.
Next, price the time rather than describing it. Capital tied up in stock in transit, the safety stock a longer transit forces you to hold at destination, the margin lost on a stockout, and the discount a late seasonal delivery attracts are all quantifiable figures. Where the difference between a three-day and a thirty-day transit is worth more than the freight differential, air is the cheaper option and the rate comparison was measuring the wrong thing altogether.
Distinguish speed from reliability, because they are separate purchases. Air transit is short but not immune to disruption: capacity is finite, consolidations are built to aircraft and unit load device limits, and cargo is offloaded at peak. Sea transit is long but the window is usually broad enough to absorb a few days without consequence. If your delivery date has no tolerance at all, what you need is not the fastest mode but the mode whose variance you can plan around.
The answer is frequently not one mode. Air the first tranche to open a launch or restock a line, sea the balance behind it; or move the whole order by sea and hold an air allowance against the shortfall. For European and Central Asian destinations, rail sits between the two on both cost and transit and often removes the need to choose at all. Splitting the shipment also splits the risk, which matters when one consignment carries a whole season.
Run the test in order: the regulatory and physical gate, then value density, then the cost of time, then the variance your delivery date can absorb. Write down the assumptions behind the answer, particularly the packed dimensions and the value per kilogram, because the conclusion changes when either of them does. Rate cycles, peak surcharges and capacity shift through the year, so a decision taken in March deserves rechecking before a September shipment.
