Global air cargo demand up 6% YoY in August; spot rates hold at $3.13/kg, +24%

The Analysis

Three thirteen a kilo. That is the average spot rate Xeneta put out for August, and it is the number I have been staring at, because I am the one who signs the order that absorbs it. Twenty-four percent above where it was a year ago, three percent below where it was in July, and the volume underneath it grew six percent year on year after July managed five. Set that beside the other two figures in the same reading: capacity flat against last year, and the dynamic load factor at sixty-one percent, three points higher than August 2025. Read in that order, this is not a freight story. It is a note about what my shelves will look like in six weeks and what is left of my margin when the goods land.

I want to be plain about which chair I am sitting in, because the same release gets read very differently from different desks. I buy. I do not operate aircraft, I do not run a charter desk, and I am not going to tell anybody how to build a rate. What I do is decide, for a given SKU in a given season, whether these units fly or float, and how many of them. Every dollar of that three thirteen comes out of the same pocket as the fabric and the duty. So when a rate moves twenty-four percent in twelve months, the question I have to answer is not what the market is doing. It is whether the units I am about to commit to still make money at the price I can actually charge, and whether they reach the shelf in time to be sold at that price.

The part I keep coming back to is the shape of it rather than the level. Demand grew six percent. Capacity did not grow at all. The load factor went up three points. Those three sentences describe one physical thing: more boxes chasing the same space, and the space is fuller than it was. Fuller aircraft do something very specific to me as a buyer. They turn space into something I have to go and secure early instead of something I can buy at the last minute at a price I like. That, rather than the rate itself, is what decides my shelf-fill rate in the middle of a season. The rate decides my margin. The load factor decides whether I have anything on the shelf to earn that margin on. I need to price both, and most buying organisations only price the first.

Let me take the cost structure apart, because a kilo rate is not a number a buyer can use on its own. Assume a unit I know well: a mid-weight outerwear piece at ninety dollars retail, one point two kilos of chargeable weight per unit once it is boxed. At three thirteen a kilo the air freight inside that unit is three dollars seventy-six. A year ago, at the rate the twenty-four percent implies, which is three thirteen divided by one point two four, or about two fifty-two, the same unit carried three dollars two. The difference is seventy-four cents. That is the entire effect of this release on my unit cost sheet, and it is the only number in it I can act on.

Now push it through to margin, because seventy-four cents does not mean anything until you see where it lands. Assume the unit costs me thirty-four dollars twenty landed before air freight, which is FOB, duty and inland, all in, sitting at thirty-eight percent of the ninety dollar ticket. This year the unit lands at thirty-seven ninety-six and earns me fifty-two dollars four, a margin of fifty-seven point eight percent. Last year it landed at thirty-seven twenty-two and earned fifty-two seventy-eight, a margin of fifty-eight point six. Eight tenths of a percentage point, gone. If your whole season profit runs at four or five points of margin, eight tenths is not a rounding error. It is roughly a fifth of your profit, moved by a figure somebody else published.

Can I raise the ticket to get it back? Sometimes, and I want to be honest about how narrow that sometimes is. To hold last year's fifty-eight point six percent margin against this year's thirty-seven ninety-six cost, the ticket has to go to thirty-seven ninety-six divided by nought point four one four, which is ninety-one sixty-nine. A dollar sixty-nine, or one point nine percent. Assume for a moment the customer simply pays it and I lose nothing, and the margin is restored.

But customers do not pay it for free. At ninety-one sixty-nine my profit per unit is fifty-three seventy-three instead of fifty-two oh four, so I can afford to lose one minus fifty-two oh four over fifty-three seventy-three, which is about three point one percent of my units, before the increase costs me more than it earns. Three percent. That is the entire elasticity budget I have on this item. If it is the sort of thing a customer compares against three substitutes on a shelf, I will lose more than three percent of units for no reason at all and the increase is a mistake. If it is the sort of thing they came in for, I probably keep them. The decision is not whether to raise the price. It is which of my SKUs are in the second group, and that is a list I can actually produce.

The rate is the part everybody looks at. The sixty-one percent is the part that bites, and I want to be careful about how I use it, because I am not in the business of calling where rates go and I am not going to pretend to know. What I can read off sixty-one percent is physical. Aircraft are fuller than they were a year ago by three points and capacity did not grow, so on average the space I want on the flight I want is not sitting there waiting for the day I decide I need it.

For a buyer that converts into exactly one input: booking lead time. I treat the load factor as a lead-time number, not a price number. Every SKU I plan to fly gets its booking date pulled forward, and how far I pull it should scale with how full the market is. At sixty-one percent I am pulling critical SKUs forward by five days against what I did last year. That is not a forecast. It is a reaction to a measured number.

Here is the calculation I actually run before I authorise an air shipment, and I am going to write out every assumption so you can put your own numbers in and see whether it still holds for you. Assume a seasonal SKU with planned season volume of twelve thousand units at ninety dollars retail. Assume ocean lands the unit at thirty-four fifty-five, which is my thirty-four twenty plus thirty-five cents of ocean and inland; use your own contract figure, the structure is what matters. Assume air lands it at thirty-seven ninety-six, the number I built above. Assume that with no air freight at all this SKU sells through eighty-eight percent at full price, and I clear the remaining twelve percent at forty percent off, so marked-down units go out at fifty-four dollars.

Scenario one, all ocean. Ten thousand five hundred and sixty units sell at full price, earning ninety minus thirty-four fifty-five, which is fifty-five forty-five each, for five hundred eighty-five thousand five hundred and fifty-two dollars. One thousand four hundred and forty units clear at fifty-four, earning nineteen forty-five each, for twenty-eight thousand and eight dollars. Season gross profit, six hundred thirteen thousand five hundred and sixty dollars.

Scenario two, I fly two thousand four hundred units early and ship nine thousand six hundred. Assume that getting twenty percent of the season onto the shelf four weeks earlier lifts sell-through from eighty-eight percent to ninety-six percent, because I am not guessing on the size curve for those units and I am not conceding the first four weeks of the window. Now eleven thousand five hundred and twenty units sell at full price and four hundred and eighty get marked down. The two thousand four hundred flown units earn fifty-two oh four each, one hundred twenty-four thousand eight hundred and ninety-six dollars. Nine thousand one hundred twenty ocean units sell at full price at fifty-five forty-five each, five hundred five thousand seven hundred and four dollars. Four hundred and eighty units clear at nineteen forty-five each, nine thousand three hundred and thirty-six dollars. Total, six hundred thirty-nine thousand nine hundred thirty-six dollars.

The difference is twenty-six thousand three hundred and seventy-six dollars in favour of flying. That is my answer to whether this increase is worth carrying. Let me show where it comes from, because I do not want anybody signing off on the strength of a bottom line. Moving from twelve percent markdowns to four percent means nine hundred and sixty units avoid the markdown, and each of those is worth fifty-five forty-five minus nineteen forty-five, thirty-six dollars. Nine hundred and sixty times thirty-six is thirty-four thousand five hundred and sixty dollars of markdown avoided. The air premium I paid is three forty-one a unit on two thousand four hundred units, which is eight thousand one hundred and eighty-four dollars. Thirty-four thousand five hundred and sixty minus eight thousand one hundred and eighty-four is twenty-six thousand three hundred and seventy-six. The freight is not the point. The sell-through is the point, and the freight is what I pay to get it.

Now invert it, because this is the number I actually keep as a threshold. I paid eight thousand one hundred and eighty-four dollars of extra freight. Each markdown unit I avoid is worth thirty-six dollars. Eight thousand one hundred and eighty-four divided by thirty-six is two hundred and twenty-seven, so the shipment pays for itself once it saves two hundred and twenty-eight markdown units. On a twelve thousand unit season that is a sell-through lift of one point nine percentage points. Under two points. Not ten, not a vague improvement in sell-through. If the air freight does not move sell-through by at least about two points, I am paying a premium for a feeling. Every air decision this season goes through that test, and I want the test written down, because it felt urgent is not a test and it is what usually gets said instead.

There is a second place this lands that has nothing to do with margin percentage, and that is inventory turns. Assume the same SKU and assume I run about eight weeks of cover on it. Flying rather than floating does not change how much I buy; it changes when those units become sellable and therefore how many weeks of stock I am holding at any point in the back half of the season. Assume air shortens my inbound pipeline by twenty-eight days against ocean, which is four weeks. On a unit landed at thirty-seven ninety-six, four weeks at an assumed twenty percent annual carrying cost is thirty-seven ninety-six times nought point two times twenty-eight over three sixty-five, which is about fifty-eight cents.

That is real money and it is not the reason to fly; it does not even cover the seventy-four cent rate increase on its own. I raise it because people put carrying cost at the centre of the air versus ocean argument and it is almost never the deciding term. The deciding term is the selling window, and the selling window shows up as sell-through rather than as carrying cost.

One more figure in that release, and it is the one most likely to be misused: spot rates were down three percent month on month. Three thirteen against roughly three twenty-three in July is about ten cents a kilo, about twelve cents on my one point two kilo unit, and on the two thousand four hundred units above that is under three hundred dollars. It is noise at my volume. I am not rebuilding a season plan around ten cents, and I am not going to read a one month move as a direction, because a single month-on-month change in a flat-capacity market is not a trend and I do not have the data to call one. What I do with it is narrow: where I have discretionary air freight that can slide two weeks without touching sell-through, I slide it and take the ten cents. Anything with a sell-through consequence stays exactly where it is.

The alternatives, and what each actually costs me. Shifting volume from air to ocean is the obvious lever, and the arithmetic above tells you precisely when it is the wrong one. If the units you move are the ones carrying the sell-through lift, giving up air costs thirty-six dollars a unit in markdowns to save three forty-one in freight, and you only need to be wrong about two hundred and twenty-eight units to come out behind. Cutting the air share is not a cost decision, it is a forecast-accuracy decision, and when the forecast is weak the answer is usually to fly a smaller and better chosen set of units rather than none at all.

Shifting the sourcing country does nothing about a global air rate, and it does something else that is easy to miss: it moves the transit baseline underneath the whole calculation. If a new origin adds twelve days of ocean transit, the four week gap that made air worth three forty-one a unit may become a six week gap that makes it worth far more, or a two week gap that makes it worth nothing. Re-run the threshold, do not reuse it. Holding more inventory instead of flying is the third option and the one I like least in a market reading sixty-one percent, because it does not solve lead time. It moves the buffer onto my balance sheet at full landed cost and it does nothing for the SKUs where I am guessing on the size curve. That is the actual reason to fly: not speed for its own sake, but buying the right to decide later.

The traps, and these are the ones that have cost me money. A kilo rate is not comparable across SKUs until you have normalised for chargeable weight, and chargeable weight is where volumetric rules quietly eat the margin, because a bulky light unit can bill at two or three times what it actually weighs. Every air line in my book has to be expressed as dollars per unit landed, built from chargeable weight times rate plus ground handling and clearance, or it is not comparable and I cannot run the sell-through test against it.

Then there is treating the market rate as if it were my rate. Three thirteen is an average of what somebody paid; my contract, my lane, my forwarder's allocation and my commitment all move my own number away from it, sometimes by a wide margin. I use three thirteen to size the problem and my own invoice to make the decision. There is also the transit promise, quoted from wheels up rather than from factory gate to shelf ready, which hides the two or three days at each end that decide whether I hit the window at all.

I ask for the door-to-shelf number and I ask for it in writing. And there is the assumption that space exists because a rate exists. At sixty-one percent load factor a quoted rate without a confirmed booking is a quotation, not a plan, and I have watched people learn that in the middle of a launch week.

If I were the buyer signing this season, here is what I would actually do. I would not fight the twenty-four percent, because I cannot; I would go and find the two points of sell-through that make it irrelevant. I would write the sell-through threshold into the air freight approval rule so that nobody has to argue about it in September with a launch date hanging over them. I would pull my booking dates forward by five days on every critical SKU purely on the strength of the sixty-one percent, and I would treat a missed booking as a stockout risk owned by planning rather than as a transport problem owned by somebody else. I would insist that every air SKU carries a dollar-per-unit landed freight figure so the test is arithmetic rather than instinct. And I would ignore the month-on-month move entirely until it has pointed the same way for three months running.

One thing I would not do is wait for the rate to come back. Capacity was flat in this reading and nothing in it tells me space is about to get cheaper, and a plan that depends on it getting cheaper is not a plan, it is a hope with a spreadsheet attached. Where the cargo actually routes and which gateways behave better under this kind of loading is Ömer Kaya's ground, not mine. I only know what the three thirteen does to my margin and what sixty-one percent does to my shelf, and between the two of them I know exactly which units I am willing to pay for.

↑ Back to the key points

— By Grace Yang