Express fuel surcharges reset for Sep 7-13 week: DHL 43.5%, UPS 46.75%, FedEx 46.00%

The Analysis

Tuesday, 11:40 in the morning, a sales lead sends a quote to a customer. The quote has a fuel line in it, one percentage sitting in a small box near the bottom of the page. Monday the invoice comes back with a different number in the same box. Nobody lied to anybody. The table flipped at midnight, the quote was written against the old one, and the gap is small enough per piece that no one argues and large enough per month that somebody should. I have watched this run four weeks straight inside one account. The percentage was never the problem. The date was.

For the week of 7 to 13 September the three tables read like this. DHL, Hong Kong and mainland origin, 43.5 percent, up from 43.25 the week before. UPS 46.75, down from 48.25. FedEx 46.00, down from 47.75. Two down, one up, and the top of the range sits 3.25 points above the bottom.

Most people will average those three numbers, put the average in a slide, and move on. I would not. Three carriers, three indexes, three effective moments, and a shipper with volume spread across them is running three clocks at the same time. Then add capacity. Air space is tight because of the Middle East disruption, tight space keeps the base rate up, and fuel is charged as a percentage of that base. A percentage that falls on a base that rises still costs more money at the door. That is the trap sitting inside any headline that says surcharges eased.

Start with the clock, because the clock is the part I actually touch. A weekly table does not reset when the advisory reaches your inbox. It resets at a fixed hour, and the advisory is a courtesy that lands midweek, often two or three days after the table has already gone live. If your pricing person waits for that email, they are writing Wednesday's quotes off Monday's number and calling it current. The number that governs an invoice is the one live at the moment the charge is stamped, and that moment sits later than the day you agreed the rate and earlier than the day the bill lands on your desk.

So measure a quote in hours. One written at noon on Tuesday against the previous week's table has exactly 132 hours of life in it before the next flip, counted to the following midnight. Most quotes in this trade go out with a month of validity, which is 720 hours and four table flips, all priced off one fuel number. Somebody carries that difference and it is not the carrier. When I ask a sales team how long their quote lives, they usually answer in weeks, and then I show them the calendar and they answer in hours.

The spread is 3.25 points. Assume a 20 dollar base, with fuel applied straight as a percentage of that base, and I will use exactly that assumption for every number below. Cheapest table against dearest is 65 cents on one piece. Sixty-five cents sounds like nothing until it gets multiplied by a month of volume, and then it sounds like a line item somebody has to explain.

Assumption set for the small seller: 3,000 express pieces a month, average base 20 dollars, volume split evenly across the three carriers. Blended fuel on this week's tables is 45.42 percent, which is 9.08 dollars a piece and about 27,250 dollars a month in fuel alone. The same blend last week was 46.42 percent, 9.28 a piece, roughly 27,850. The easing across an entire month is worth about 600 dollars. That is real money and it is smaller than the damage one stale quote does in a heavy week.

Now leave every assumption where it is and move the same 3,000 pieces onto the cheapest table instead of splitting them. 43.5 against 45.42 is 1.92 points, 38 cents a piece, 1,150 dollars a month. That is what diversification costs in a week when the tables diverge. I am not recommending anybody concentrate volume on a single carrier to chase 38 cents, because space, cut-off and service are worth more than 38 cents on any week of the year. I am recommending that you know the number, because in a week when the spread opens to five points the same arithmetic stops being small and starts being a decision.

Here is where the headline misleads. UPS eased 1.5 points. FedEx eased 1.75. DHL tightened a quarter of a point. If the base under those percentages went up because space is tight, your all-in number can rise while two of the three percentages fall. The percentage is a rate. The base is a price. You pay a price, and the invoice does not care what the rate did.

The thing nobody prints is this. The surcharge does not attach to your booking, your quote, or your ship date. It attaches to a timestamp, and that timestamp is a scan. Which scan is a question each carrier answers for itself, and most shippers have never asked it. Label created 23:50 on Sunday. Pickup scanned 06:20 Monday. First hub scan 22:40 Monday. Three different hours, and if the table flips at 00:00 Monday, two of those three belong to different weeks. Until you know which scan your carrier uses, you do not know which week you are paying for, and you cannot plan a cut-off around a boundary you cannot name.

Pull the internal cut-off forward one shift and most of this evaporates. If the rate-setting scan is the first hub scan at 22:40, a 20:00 cut-off sits comfortably inside the week you planned. A 21:30 cut-off does not, because the truck waits at the dock for seventy minutes and the scan lands on the far side of midnight. Warehouse staff know this in their bones. Sales does not. The two of them are quoting different weeks and neither one of them is lying, which is exactly why the argument never gets settled.

The last ninety minutes at the hub is where I would put a camera if anybody let me. Between 22:00 and 23:30 the belt is full, the manifest is closing, and the person holding the scanner is deciding, without knowing it, which week four hundred parcels belong to. A box that reaches the scanner at 23:56 is priced on this week's table. The same box ten minutes later is priced on next week's. Nobody in that building is thinking about fuel. They are thinking about the truck that has to pull out at 23:45 to make the flight, and that truck, not the table, is what set your cut-off in the first place.

There is a second lag, smaller and more irritating. Forwarder advisories arrive midweek and they are usually close, but they are a retelling. The carrier's own published table is the source, and on the day they land the two are not always identical to the same decimal. When they disagree, price off the carrier and treat the advisory as a prompt to go and check. Chasing the advisory alone is how a quote ends up a tenth of a point off and a whole week late, and the tenth of a point is the part people notice while the week is the part that costs.

When you ask a carrier which scan sets the rate, a vague answer is the same as no answer and you should treat it that way. A usable answer has three parts: the timestamp, the system that records it, and the hour the table turns over in your origin time zone. If the account manager cannot put all three in writing, escalate once and then assume the earliest scan and pull the cut-off accordingly. Assuming early costs you a little capacity. Assuming late costs you the spread on every parcel in the batch, and it costs it without a single alarm going off.

Layered by parcel profile, because profile is what shows up at the belt. Lightweight cross-border e-commerce goes first. A 300 gram parcel on a 20 dollar base is almost all fuel and very little freight, so 3.25 points is a bigger share of the total than it is on a 40 kilo carton. These sellers also reprice most often, which means they touch the table most often, which means they get it wrong most often. The error is almost never the percentage. It is the date somebody wrote next to it, or the date somebody forgot to write.

Heavy and dense shippers feel the same move in absolute cash. A pallet with a large base does not care about 3.25 points in proportion, it cares in dollars, and the dollars are big enough to argue about with an account manager. Their exposure is different in kind: fewer shipments, bigger invoices, and a reconciliation done once a month after the fact. By the time finance sees it, four tables have gone by and nobody can say which week a line belongs to. That is not a pricing problem, it is a filing problem, and it is fixable in one afternoon with a date column and a rule that no invoice is booked without it.

Categories feel this differently and it is worth naming them, with the assumption stated: I am holding the 20 dollar base constant across all three, so the percentage spread is identical and only the share of freight in the ticket differs. Small fashion and accessories, the volume that fills cross-border e-commerce, sit closest to the pure fuel case, because freight is a thin slice of the total and the percentage moves almost the whole ticket. Consumer electronics with declared value sit in the middle, where insurance and handling start to carry weight. Industrial samples and spare parts sit furthest away, because the base is large enough that 38 cents is a rounding error and the missed flight is the actual problem.

Split book against single source is the other cut worth making. Three carriers means three clocks and one spreadsheet, which is work, but at least the work is visible. One carrier means one clock and no comparison, and in a week like this the single-clocked shipper is frequently the exposed one, because the table that rose is the Hong Kong and mainland table. If that is your origin and that is your only carrier, your lever this week is timing rather than routing. Move what can move onto the earlier uplift, and let the rest ride at the new number instead of pretending it did not change.

Transmission, in hours. Hour zero is the flip. Between hour zero and hour six the first manifest of the week closes and the first parcels get priced on the new table. By hour twenty-four the first invoices exist in the system. By hour forty-eight you can reconcile them, and if you have not, you will reconcile them at month end, when the person who wrote the quote no longer remembers why that number was chosen. By hour 132 a quote written the previous Tuesday is stale. By hour 168 the cycle restarts with a fresh set of percentages and the same argument in the same meeting room.

The contract wave runs on a slower clock and it is where the real money sits. A fuel clause written as a fixed percentage for a quarter is a bet that the table holds still. The table moves every week. If the agreement says 45 percent for the fourth quarter, somebody has fixed a price on a floating input, and the outcome will be settled at the close by people who never saw a single manifest. What I would ask for instead is a clause indexed to the weekly table with a cap on how far it can move in one direction, so the swing is shared instead of assigned to whoever signed last.

Assumption set for the large seller: 20,000 express pieces a week, average base 20 dollars, split evenly. The 1.92 point gap between the blend and the cheapest table is 38 cents a piece, 7,667 dollars a week, roughly 30,700 across a four-week month. Put all of that volume on the dearest table for one week and the full 3.25 points costs 13,000 dollars in that week alone. Nobody fires a carrier over that number. Plenty of people would like to see it on a screen at 09:00 on Monday, before they make a routing decision they cannot take back at four in the afternoon.

Now size the error rather than the spread. Assumption: one heavy week ships 1,200 pieces on quotes written before the flip, average base 20 dollars, and the gap between the quoted table and the live table is 1.92 points. That is 38 cents a piece, 456 dollars, gone in a week, and it very nearly erases the 600 dollars of easing the whole month produced. Two numbers, same week, opposite directions. This is why I keep saying the discipline of dating a quote is worth more than the skill of reading a table.

When the table moves in the middle of an order, somebody has to say something to the customer, and the wording matters more than the number. What works is a short message with two facts and one hour: which table applies, and by what time the parcel has to be scanned to stay on it. What does not work is an apology with a percentage inside it, because the customer hears a price increase and stops listening before the hour arrives. Two lines have saved an order that three paragraphs lost, and the only difference was that the two lines told the customer what to do next.

Reverse the whole thing. Say you decide the tables are noise and you stop repricing weekly. In a week when all three move together you lose nothing and you save the labour. In a week like this one, one up and two down, you lose the spread on every lane whose quote predates the flip, and you lose it quietly, because the invoice matches the table and the table is correct. Nothing looks broken. The margin thins by 38 cents a piece, which is the sort of number that never trips an alarm and never stops arriving.

The conclusion flips under three conditions and you should know which one you are standing in. It flips if the base falls while the percentages rise, because then the percentages look bad and the invoice looks better, and anyone who repriced in a panic paid for a signal that pointed the wrong way. It flips if the spread closes under a point, at which point the analysis costs more in analyst hours than it saves in fuel. It flips again if the book is small enough that 1,150 dollars a month does not cover the hours spent chasing it. Running this exercise on 500 pieces a month is a hobby, not a control, and pretending otherwise is how a small team burns a Friday.

Traps, in the order they bite. Fuel charged on top of other surcharges, where the percentage is applied to a base that already contains handling, and the customer never sees the compounding because the line item is one number on one page. Quotes with a fuel line and no date beside it, so the percentage is right and the week is wrong. DDP quotes where fuel is buried in a landed cost and nobody can say which table was used on which day. Disputes filed after the reconciliation window, which most forwarders will not reopen once the month is closed. And the advisory lag I complained about earlier, which will still be there next Monday, and the Monday after that.

What the coverage misses is simpler than any of this. There is no such thing as the fuel surcharge. There is your surcharge, on your origin, on your carrier, in the week your parcel was scanned. Every conversation that starts from a market average ends with somebody guessing, and the guess is usually generous to whoever is doing the guessing. This week the three numbers are 43.5, 46.75 and 46.00. If you can name the one that applies to the parcel in front of you, and the hour that decides it, you are ahead of most of the people quoting against you, and you did it without a single forecast.

The first Monday of that screen will be ugly and you should plan for it. Three defects surface in a predictable order. Nobody can say which scan sets the rate, so the all-in column is an estimate wearing the clothes of a fact. Half the open quotes have no table week printed on them, so the re-quote list is a guess dressed as a process. And the base per lane lives in three files owned by three people, one of which nobody has opened since the last rate review. Fix them in that order, one per week, and by the fourth Monday the screen runs in ten minutes instead of an afternoon.

So put four numbers on one screen every Monday at 09:00. This week's table per carrier. Last week's table per carrier. The base rate on your top three lanes. The all-in cost per piece on those same three lanes. Four numbers, ten minutes, one named owner, and a rule that any lane moving more than 0.5 points week on week gets re-quoted that morning, before the first pickup rather than after the first complaint. The screen is not reporting. The screen is the trigger.

The window closes and it is gone. A table that ran Monday to Sunday cannot be renegotiated on Thursday afternoon, and a parcel that missed its scan cannot be scanned again, no matter how good the explanation is. Rescue the parcels first, argue about the percentage second, and do the arguing with the numbers already on the screen, because by the time the argument is settled the table has moved and everyone in the room is quoting last week again.

↑ Back to the key points

— By Nadia Pryce