Express fuel surcharges hold high: DHL 38.5%, FedEx 39.75%, UPS 40.50%

The Analysis

Seventeen forty at the dock and the manifest is still open. Two pallets are wrapped, the third one is not, and somebody is arguing about whether the last carton goes out tonight or waits for tomorrow. That argument is never about tonight. It is about which clock the parcel starts on, because an hour lost at the dock is an hour you have to buy back further downstream, and you always buy it back at a worse price.

Here is the set of numbers I am working from this week. For the 20 to 26 July 2026 window, international express fuel surcharges were posted at DHL 38.5%, FedEx 39.75% and UPS 40.50%. They are down hard from the mid-June peak, which ran above 48%. They are still described as elevated against early 2026, and the tracker does not publish a figure for early 2026, so I am not going to invent one.

Two things about those lines matter more than the level itself. Each one is a percentage, so it scales with a base that somebody else chose. And each one is posted for a week, so it is a clock rather than a season. I read them as operating numbers, not finance numbers, because the only levers I have on a floor are hours and handovers, and both of those decide which week, which base and which attempt a parcel gets billed on.

A fuel surcharge is a percentage, and a percentage is not a price. It is a multiplier, and the base it multiplies is the part you actually control. Hand over a parcel that bills at 45 dollars of transport with a surcharge of 40.50% and you pay 18 dollars 23 on top. Hand the same parcel over on the same lane at 38.50% and you pay 17 dollars 33. Same box, same lane, same weekday, ninety cents apart. That is the entire prize on this number, and it is won or lost in the ninety minutes before the truck closes.

The spread across the three carriers this week is 2.00 points, from 38.50 at the bottom to 40.50 at the top. On a 45 dollar base that is ninety cents a parcel. On a 20 dollar base it is forty cents. On a 120 dollar base it is 2 dollars 40. So the ranking of DHL, FedEx and UPS by percentage tells you nothing about your own invoice until you know what each of them is multiplying. I have seen people switch carriers on a two point fuel difference and end up paying more, because the cheaper percentage was sitting on a dearer base. Rank them on all-in cost per parcel for your own mix, worked out on your own weights and zones.

The other half of the story is the fall. Mid-June these were above 48%. Call it 48.5% for the sake of the arithmetic, and let me say again that this is my assumption, because the data only says above 48. On a 45 dollar base that is 21 dollars 83 against 17 dollars 33 today on the lowest of the three. Four dollars fifty a parcel. That is real money and I am not going to pretend otherwise. But it is money that arrived because an index moved, and indexes move back. Bank the saving. Do not spend it, do not re-price a customer on it, and do not build a 2027 budget on the assumption that fuel percentages stay here.

What I actually care about is that this number resets every week. The window it was posted for runs 20 to 26 July. That is 168 hours of exposure, and if you are reading this on the Wednesday of that week you have somewhere around 120 hours left before it rolls into whatever gets posted next. A weekly clock reduces the whole problem to one operational question: which side of the reset does this parcel fall on. Everything else is commentary.

Start with the handover minute. Assume your carrier resets the weekly surcharge at 00:00 on Monday local time, and go and check that against your own contract, because I am assuming it. On that assumption, a trailer that pulls out at 23:40 on Sunday night and a trailer that pulls out at 00:20 on Monday morning are forty minutes apart in departure and a full surcharge week apart in billing. Forty minutes at a dock door is the entire difference. If your manifest closes at 17:30 and the truck leaves at 18:00, you have ninety minutes of slack to fight over, and every one of those minutes is cheaper than any rate negotiation you will sit through this quarter.

The trap inside that is the time zone. Your dock clock, the carrier hub clock and the origin country clock are three different clocks the moment you ship cross-border, and a 23:40 handover in your zone can already be Monday in the zone the billing engine runs on. Pick one reference zone, write the reset hour down in that zone, and put it on the wall where the dispatchers can see it. Three clocks arguing is how parcels get billed on the wrong week, and nobody notices for a month because the invoice arrives late.

Here is the hour model I use for an international parcel, and every number in it is mine, not the carrier's. Order to pickup cut-off: up to 6 hours. Pickup to origin sorting centre: 4 to 10 hours. Sorting and outbound: 3 to 6 hours. Linehaul including customs: 12 to 36 hours on a clean run. Destination station to the opening of a delivery window: 8 to 16 hours. Delivery window to completion: 4 to 8 hours. On the honest end of those ranges you are somewhere between 37 and 82 hours door to door before anything at all goes wrong. Most people quote their customers a number pulled out of the middle of that range and never measure which segment is actually eating it.

Now put failure into the budget. Assume a first-attempt failure rate of 6% in a normal week and 10% in a tight one. Assume each failure costs another 18 to 30 hours, because the parcel goes back to the station, waits for a sort, and comes out on the next window. Ten percent failure is not ten percent of your parcels running slightly late. It is a full extra cycle for one parcel in ten, and the ones that fail are disproportionately the ones already at the end of their hour budget, in the farthest metro, on the last window before a weekend.

You might reasonably ask what any of that has to do with a fuel percentage. Two things. Every extra cycle is extra handling hours at the station, and station handling hours in a tight week are exactly what makes a network miss its own published promise. And every parcel that rolls into the following week lands on the next surcharge posting, which you have not read yet. Failure does not just cost you the re-attempt. It moves the parcel onto a price you have not priced.

Let me run the full account with every assumption on the table. Assume 1,200 parcels a week. Assume a 45 dollar average transport charge per parcel. Assume you can move a quarter of your volume between the dearest and the cheapest of the three percentages without breaking a service promise. Assume 14 weeks between now and the start of peak, and hear me again, that 14 is my own planning horizon, not a published date.

Run it. If you moved everything from 40.50% to 38.50%, the saving is 90 cents on 1,200 parcels, which is 1,080 dollars a week. You will not move everything, so take the realistic version: a quarter of the volume, 300 parcels, 270 dollars a week, and over a 14 week horizon that is 3,780 dollars. That is the size of the prize on carrier selection at today's spread. It is worth having and it is not worth a re-organisation.

Against the June peak, on the same 45 dollar base, the lowest of the three has come down roughly 10 points. Ten points on 45 dollars is 4 dollars 50 a parcel, and on 1,200 parcels that is 5,400 dollars a week that has stopped leaking out of the invoice. Do not multiply it by anything. It is gone, it is already in your numbers, and the only useful thing to do with it is decide now what you would do if it came back. Mid-June was five weeks ago. That is how fast it can come back.

Now the hour side of the account, which is the side people never write down. Assume your dock loses forty minutes three times a week because the manifest runs late. That is 2 hours a week and 28 hours across a 14 week horizon. Assume each of those slips pushes 60 parcels past a cut-off, and each of those parcels then costs 20 extra hours to reach a door. That is 180 parcels a week and 3,600 parcel-hours of delay a week, bought for two hours of dock time. Fifty thousand parcel-hours across the horizon. The exchange rate is brutal and it runs the wrong way.

So what gets done, and by when. Today is Wednesday. Before 18:00 on Friday 24 July I want the reset time for each of the three carriers written on one page, in one reference time zone, signed off by whoever owns the rate table. Not next week. This week, because the window you are standing in closes on Sunday and you cannot test a clock you have not written down.

From Monday 27 July, the first day of the next posting, I want a handover log on the dock with a minute stamp on every trailer, and I want it compared against the carrier's own scan time every week for three weeks. Expect to find a gap. There is always a gap, and it is usually paperwork rather than transport: a manifest printed but not closed, a driver waiting on a signature, a trailer held for one pallet that could have gone on the next run. Find the gap in hours, not in excuses.

Set a threshold and put a name against it. If a carrier's posting moves by more than 2.00 points week on week, the routing table gets reviewed within 48 hours by the transport lead. Two points is the distance between the cheapest and the dearest carrier today, so it is the size of move that genuinely changes a routing decision. Below that, do nothing and let people get on with the work.

Targets, because a target nobody can miss is not a target. Ninety-five percent of trailers stamped out at or before the planned cut-off minute. First-attempt failure rate held under 7% through August. Handover-to-carrier-scan gap under 30 minutes on every lane by 31 August. Average door time for your top three destination metros measured weekly and reported in hours, not in service level names, because a service level name hides the segment that is failing.

Alternatives, and what each one costs. Drop a service level on the parcels that have hours to spare. You buy maybe 24 to 48 hours of transit and you cut the base that the percentage multiplies, which is the cleanest lever there is on a percentage surcharge: attack the base, not the multiplier. The cost is a promise somebody has to re-issue to whoever sold it, and a delivery window you now have to defend in the last mile. If your customer bought speed, this is not free.

Consolidate. Whether accessorials carry the surcharge is a contract question and your own table is the only thing that answers it, so check before you assume either way. But a percentage on transport is simple arithmetic: fewer, heavier parcels means fewer base charges to multiply. Two parcels billing at 22 dollars each carry more surcharge between them than one parcel billing at 40 dollars, because you multiplied twice. Consolidation costs you sorting time at the dock and it costs you granularity when one item in the box has to come back.

Re-weigh and re-measure your top twenty cartons. Carriers bill on the greater of actual and dimensional weight, and a carton that quietly grew two centimetres in a packaging spec change now bills in a higher bracket before any percentage is applied to it. Twenty cartons is an afternoon with a tape measure and a scale. It is the cheapest hour you will spend this month, and it is the one everybody skips.

Fix the address field before peak. A failed first attempt is not a small miss, it is a full extra cycle of 18 to 30 hours on my model, and a large share of failures come down to address quality: a missing apartment number, a company name with no street, a postcode that resolves to the wrong station. Take your last four weeks of failures, put a reason code on every one, and count them. If address quality is more than 30% of failures, the fix is a validation rule at checkout, not a phone call to the depot.

Audit the destination type flag in your system. The flag is what decides which rate table a parcel lands in, and it is usually set once at onboarding and never looked at again. Offices get flagged as homes, homes get flagged as offices, and the invoice quietly disagrees with your quote for months. Pull a sample of 200 addresses, check them against what the carrier actually classified, and fix the ones that are wrong before volume rises.

Build an exception queue with fixed clearing times. Assume exceptions are cleared at 09:00 and 16:00 every day. On that assumption, an exception raised at 16:30 does not get touched for seventeen hours, and by then the parcel has missed the window it was supposed to be on and moved to the next one. Two clearing times a day is the minimum. Three, at 08:00, 13:00 and 17:30, costs you one person for an hour and saves you the seventeen.

And when a parcel is already late, get the parcel back first and argue about who pays afterwards. That sounds obvious and it is not, because the first instinct on a floor is to find out whose fault it was. Fault takes hours to establish and the parcel does not have hours. Reroute it, relabel it, move it to a pickup point, do whatever gets it to the recipient today, and run the post-mortem on Monday. A parcel delivered late by a carrier you are still arguing with is worth more than a parcel still sitting in a station while two departments exchange emails.

Capacity, because it is July and the clock to peak is already running. Assume 14 weeks to peak on my own horizon and call it 2,352 hours. Inside that time you need the peak parcel profile settled: carton sizes locked, address validation live, cut-off minutes agreed with every carrier, and any extra collection slots booked. Carriers allocate peak capacity in windows, and the window closes on their clock, not on yours. When it shuts, that is it. What is left afterwards is the spot market, and spot capacity in the last mile in December is not a rate, it is an apology.

The pitfalls, in the order people hit them. Comparing percentages across carriers without the base. Assuming the surcharge applies to the same charges in every contract, when the table in your agreement is the only thing that decides it. Reading the weekly posting after the week has started. Booking peak capacity as though a week were a small unit, when in the last mile a week is five cut-offs and five separate chances to miss one. And treating a two point easing as a trend, when the only honest thing you know about next week's number is that you have not read it yet.

One more that belongs to the commercial side rather than the dock. If you quoted a customer a landed price back in June when these were above 48% and you are now being billed at 38.5%, you are holding roughly 4 dollars 50 a parcel of margin you never priced for. Decide deliberately whether to hold it or give some of it back. Holding it silently is how a rate card turns into a renewal conversation you lose in January, and giving it back without saying anything is how a saving disappears without buying you anything at all.

Two more hour questions come up every time I run this exercise. One is whether the surcharge follows the week a parcel was tendered or the week it was delivered. Most tables I have seen bill on tender, but yours is the only one that counts, so go and check it. If your table bills on the delivery week, then a customs hold is not merely a delay, it is a repricing: a parcel held for 30 hours across a Monday reset gets billed on a number that did not exist when you quoted the job. Under tender-based billing the same hold costs you hours and nothing else. That single line in the contract decides whether you fight for clearance speed on cost grounds or only on service grounds, and it changes who you phone when a parcel is stuck.

The other is what a failed attempt costs next to a redirect. Assume a re-attempt costs 18 to 30 hours and still carries a failure risk on the second try, call it half the first rate. Assume a redirect to a pickup point costs 6 to 10 hours and almost never fails twice, because the recipient collects it on their own clock. On those numbers, offering a pickup point the moment the first attempt fails is cheaper in hours than insisting on a second attempt, and cheaper in station capacity too, which is the thing that runs out first in peak. The cost is a customer who wanted it at the door and did not get it, so offer it as an option, never as a default.

What I would actually do with this week's numbers is small and specific. Write down the three percentages and the base each one multiplies. Write down the reset hour in one single time zone. Stamp the trailers. Count your failures by reason and put hours on each reason. Then stop looking at the index for a fortnight, because a weekly number watched daily makes people do stupid things with trucks, and the trucks are the expensive part.

↑ Back to the key points

— By Nadia Pryce