UPS and FedEx Add PIN and QR Delivery Checks as DHL Fuel Surcharge Hits 30 Cents a Pound

The Analysis

A client rang me yesterday at twenty to five in the afternoon and skipped hello. He moves phone accessories and small audio into the north-eastern United States, around eleven thousand pieces a day once November gets going, and what had him swearing was not a rate rise. It was thirty seconds at a front door in New Jersey. The driver scanned a parcel holding two hundred and sixty dollars of earbuds, asked for the code the buyer had never opened, waited, closed the van and drove off. The second attempt went the same way two days later, and on the third strike the box started back. He paid freight out, he will pay freight out again, the unit comes back scuffed if it comes back at all, and the review left on Friday says never arrived.

That is what the week of 12 to 16 October 2026 really put on our desk. UPS is rolling out PIN protected delivery and FedEx is standing up its own authenticated delivery process, both built for high value shipments with consumer electronics named out loud, and both requiring a PIN or a QR code to be verified before a parcel may be left. On the money side of the same page, DHL took its fuel surcharge from twenty nine cents a pound to thirty, a per pound structure that no competitor uses, and the regional carrier SpeedX stacked a five per cent peak rate adjustment on top, effective 20 October and running well into next year. UPS had still not published its 2027 general rate increase by the middle of the month, and last year that announcement landed in the third week of October.

I am not going to read the press release back at you, you have already read it. What I want is the clock. From the second you manifest a parcel to the second it hits a porch, which hour goes first, whose promise breaks, and which thousand pieces you can still rescue this week. Everything here is measured in hours, because this is decided in hours. And the clock started already: the regional adjustment went live on the twentieth, and the announcement window everybody is staring at is measured in days.

Start with why now, because security upgrade is the lazy reading and it sends you down the wrong road. The threat did not change this month. The calendar did. The surface freight market was described that week as building rather than arriving: blended truckload tender rejections were back to roughly thirteen per cent after a September peak near fourteen point six, spot rates per mile softened a touch, the networks were quiet. Quiet is when you fit locks. You do the painful changeover while the shop is empty, not when there is a queue out of the door and temporary drivers at every stop.

The second push is arithmetic on the carrier side of the ledger. A fourth quarter parcel is worth more than a July parcel, and the claims that really hurt are never the cheap ones. A missing forty dollar item is a refund and a shrug. A missing handset is a claim file, an investigation, sometimes a chargeback, and an hour of somebody here and there that nobody budgeted. Both networks are staffing up at a cadence described as consistent with prior peak seasons, which is them telling you they expect the volume, and it is also them telling you that at six in the evening in December somebody new will be standing at your buyer holding eight hundred dollars of someone else's stock. A four digit code costs less than a thousand investigations.

And then there is diesel, sitting on the very next line of the same report, and almost nobody read the two items together. The national average was six dollars seventeen point nine cents a gallon for the week of 12 October, down two cents on the prior week and up two dollars fifty one point four cents against the same week a year ago. Two and a half dollars a gallon year on year is not noise. It is the reason there is a fuel line on your invoice at all, and it is why a carrier that prices fuel per pound rather than as a percentage of freight wakes up one morning and decides it wants another penny.

Now take it hour by hour, because this either costs you nothing or costs you the quarter, and it turns on about six hours of the day.

Minus seventy two hours, somebody buys something. They are on a phone, half watching something else, they type an email into a checkout form, and a phone number goes into a field that quite possibly nobody validates. If that number is missing, mistyped, a landline, or a virtual line that cannot receive texts, the parcel you are about to manifest already has a hole in it, and the hole stays invisible until a van is standing at a door ninety hours later. Every failure I describe below starts at that keyboard and nowhere else.

Manifest time is cheap time and nobody treats it that way. Right there, for the cost of a few lines in your order management system, you can check the destination against every address that has failed twice in the last ninety days. You can flag a ship-to that is a business with no reception, a box number that cannot take a signature, an apartment block that ate three of your parcels in September, a dorm in week two of term. Changing a routing decision here costs keystrokes. Changing it in ninety six hours costs a reship and a review.

After that the parcel goes somewhere that will not save you. A hub, another hub, a linehaul truck, an automated sort that reads a barcode and throws the box down a chute. Nobody in that chain is thinking about whether a human being will be standing behind that front door. I say that because clients ring me and say the carrier should have known better. There is no mechanism by which the building knows. The building has no idea there is an authentication requirement attached to your account, and only the last fifty yards know, by which point there is nothing left to change.

Here is the hour. It runs from about ten in the morning to four in the afternoon local time, and if you pull any residential book apart you will find the same shape: most of your buyers are not at home. That has not mattered for twenty years, and that is the whole point. Delivery has worked all that time not because people are at home but because nobody was ever required to be. The parcel gets left, the buyer picks it up at half six, everybody wins. A PIN or a QR code converts an asynchronous network into a synchronous one. From the moment the scan says out for delivery, somebody has to physically be there, holding a code, inside that window, or the machine stops.

Walk the next seventy two hours. Attempt one fails at eleven. A driver enters an exception and it shows up on the tracking page that evening, worded exactly as it was worded in July, because the scan vocabulary did not change when the policy did. It says delivery attempt failed, or delivery exception. It does not say your presence was required and here is what to do about it. Your buyer reads that sentence, finds no instruction in it, and opens a ticket with you, or opens a dispute with the marketplace, and either way it lands on your own people at the worst possible moment of the year.

Attempt two is not the next morning. It is whenever that van next has slack, and in the third week of November the van has no slack. Budget twenty four to thirty six hours in an ordinary week and forty eight plus inside a peak week, and count the touches: every loop through exception status is one more handling event on their side and one more contact on yours. Attempt three usually ends it. The parcel gets flagged and starts travelling backwards through the same network that has just spent three days failing to deliver it.

If you ship domestically out of a local warehouse, that return has salvage value. If you crossed an ocean to get there, call it what it is, a write down. Nobody pays to bring a sixty dollar item back across the Pacific, clear it back in, and rebox it. You reship from the destination country or you refund, and either way the unit economics of that order are gone. Which is why the right instinct never changes: save the parcel first, argue about who is right afterwards.

Now the part that did not make anybody's headline, and the reason I sat down and wrote this rather than forwarding the newsletter with one line attached.

The carriers changed what they require at the door and changed nothing about how that requirement reaches the person standing behind it. The tracking page still speaks the old language. The notification still says out for delivery. There is no sentence anywhere in the standard flow that tells a buyer today is the day they must be standing in their hallway between ten and four holding a four digit number. So this is not really a security failure at all, it is an information failure, and an information failure is a thing you can fix yourself this week without waiting for anybody's product roadmap.

Fixing it means owning the message. You know at label creation exactly what will be demanded at the door. So send your own automated note within fifteen minutes of that label existing: this delivery needs a one time code, it will be sent to this number, if nobody can be here between ten and four reply with the word pickup and we will move it to a collection point before it leaves our dock. That one message, costing about two cents, is the difference between a ninety five per cent first attempt rate and whatever you are about to get. Do not sit waiting to be told how your carrier will handle this. Ask them, and be doing something else while you wait for the answer.

There is a second thing hiding in here that will bite the people who assumed a code is a code. Most operations generate one label per physical box. A three item order is three labels, and if each label carries its own authentication, your buyer now has three numbers arriving in three separate messages, and the driver needs all three before any of them can be left. I do not know yet whether your account groups them, and neither do you, so put that question to your account team today, in writing, before somebody finds out the expensive way in the last week of November.

And here is the piece of money that has already left the building while everybody watches and waits for a general rate increase that has not even been announced. SpeedX added five per cent on 20 October. DHL added a penny a pound. Both are live. None of it is hypothetical. The announcement everybody is braced for might land this week or might land in three weeks, and I want the model built before it does, but I also want somebody walking back to their own invoices from the twentieth and reconciling them, because that money is in there right now, and the only thing standing between you and it is whether anybody opened the file.

Let me talk about who gets hurt and by how much, because this does not land evenly, and pretending it does will waste your week.

The worst hit is high value business to consumer freight going to a residential address: consumer electronics, anything with a declared value above roughly three hundred dollars, anything the carrier has decided warrants protection. That is the population these programmes were built around. Right behind it comes a band that gets hurt almost as much for a different reason: expensive items going to buildings with no staffed desk. Gated communities without a guard, walk up apartment blocks, university halls during term, rural routes where nobody is inside five miles at noon.

Then there is a segment that quietly does fine, and I mention it so you do not spend money fixing something that is working. Anything going to a staffed business address is probably better off than it was. A reception desk is open during exactly those hours, it signs all day, and whoever sits there is already used to being asked for things. If a slice of your supposedly residential book is actually a business with somebody behind a counter, take the win, say nothing about it, and put your hours into the twenty per cent of destinations that generate eighty per cent of your exceptions.

Which brings me to geography, and this is a postcode problem far more than it is a category problem. Once you have fourteen days of exception data in front of you, ranked by destination, you will see clusters that make immediate sense the moment you look at them: the apartment heavy ring around a city centre, the university towns, the industrial parks that are dead after six. Rank them, then route them differently. Same product, same carrier, a different last mile instruction, triggered by a postcode list you own and update every week.

Now get the tape measure out, with every assumption written down where you can shout at it.

Assume you ship twelve thousand pieces a day into the United States over the forty five days from 8 November to 22 December. Assume fifteen per cent of that, eighteen hundred pieces a day, is electronics or anything else above three hundred dollars in declared value, since that is the population these programmes aim at. Assume your first attempt delivery rate on that population sits at ninety five per cent today and lands at eighty nine once a code is required, six points gone. Assume a failed attempt costs twenty four hours in an ordinary week and thirty six hours in a peak week.

Assume each extra attempt costs you four dollars fifty in redelivery or exception fees plus two dollars of your own customer service handling, call it six fifty all in. Assume eleven per cent of those failures eventually cross into return to origin. For a cross border shipper assume each return costs forty three dollars, thirty four of landed product and nine of outbound freight you never recover.

Eighteen hundred pieces at six points is one hundred and eight extra touches a day. At six fifty that is seven hundred and two dollars a day, and it is friction rather than freight, which means nobody invoiced you for it and nobody is going to notice it either. Eleven per cent of one hundred and eight is roughly twelve pieces drifting into returns every day, forty three dollars each, another five hundred and sixteen. Together that is one thousand two hundred and eighteen dollars a day. Across forty five days, fifty four thousand eight hundred and ten dollars, landing in the month your year depends on.

Against that, the alternative. An automated message fifteen minutes after label creation costs about two cents a piece. Eighteen hundred pieces is thirty six dollars a day, one thousand six hundred and twenty over the same forty five days. If that message buys back even three of the six lost points, you have recovered something over twenty seven thousand dollars for sixteen hundred. If it buys all six, the fifty five thousand comes back for less than a week of the loss. There is no version of this arithmetic in which waiting is the cheap choice.

There is also a second ledger, and it is the one that gets sellers suspended. Marketplaces do not judge you on the middle of your distribution, they judge you on the tail. If your on time delivery tolerance is four per cent, those hundred and eight failures are nine tenths of one point against twelve thousand pieces, which is nearly a quarter of your whole monthly allowance spent by one policy change inside one six hour window. You do not get penalised for being slow on average. You get penalised for the tail, and this change grows the tail.

Now the cents, because the fuel line is where patience gets expensive. Assume you move six hundred pieces a month on DHL at an average billable weight of forty two pounds, and assume your negotiated base after discounts is one hundred and eighteen dollars a piece. Twenty nine cents a pound was twelve dollars eighteen, ten point three per cent on top of base. Thirty cents is twelve sixty, ten point seven. In isolation that is forty two cents a piece, two hundred and fifty two dollars a month, and I can hear the yawning from here. Stay with me, because that small number is a decoy.

It is a decoy because that line does not move when your base moves. Assume you win another eight per cent off that base in the next negotiation, taking it to about a hundred and eight dollars fifty six. The twelve sixty stays exactly where it was, and it has quietly become eleven point six per cent of your invoice instead of ten point seven. Every point of discount anybody wins in the freight column is taxed by a fixed line nobody negotiated, and it is taxed harder the better you negotiate. Structure matters more than the penny.

And then let the diesel number do what it has been doing. Assume the per pound figure moves from thirty cents to thirty four over the next four months, which is nothing heroic given the fuel year this has been. That is another one dollar sixty eight a piece, roughly a thousand dollars a month on six hundred pieces, arriving while your contract may well be telling you your rates are held. Go and open that agreement today and look for one thing only: whether the sentence that holds your rates for twelve months covers surcharges, or covers transportation charges net of fuel and accessory charges. There is a very standard piece of drafting where the answer is the second one and nobody noticed at signature. I am not going to guess which version you signed; you have the file.

SpeedX is blunter and it is already running. Five per cent from 20 October, through into next year. Assume they carry eight per cent of your parcel book at an average six dollars eighty a piece, that is thirty four cents a piece. Assume twenty thousand pieces a week across everything you ship, that is six thousand eight hundred dollars a week, and the first week has been invoiced already while you were reading about something else. Reconcile from the twentieth onwards, this week, not at the quarterly review when it has quietly become your number.

Then UPS. Nothing published by the middle of October, and last year that announcement landed in the third week of the month. So this is a live risk window, and here is what matters about it: the announcement itself costs you nothing. Your reaction time afterwards costs you three weeks. If you start building lane level impact models the day the number goes public, you will spend ten days building them, another five arguing internally about which ones to accept, and for those fifteen days your commercial people will be quoting customers off a cost base everyone knows is wrong.

So build it now, before there is a number to attach it to. Top fifty lanes by spend, cost per piece and per pound, scenario columns at four point nine, five point nine and six point nine per cent, plus the residential surcharge, the oversize tiers and anything else that historically moves more than the headline figure does. None of that changes when the announcement lands, only one cell changes. Then the day it appears you are spending hours rather than weeks, and the customer conversation happens while there is still time for them to plan around it.

Where does all of this turn out to be a false alarm? Three places, and I would rather you tested them than took my word.

It is noise if the requirement is opt in per shipment rather than switched on by default. In that case you are running a value threshold decision rather than fighting a fire: below some declared value the coordination cost plainly exceeds whatever you save on fraud, and no carrier is going to volunteer where that line sits because their incentive runs the other way. Your account team will answer this in ten minutes. Ask the question in writing and keep the email, because the answer will change twice before Christmas.

It is noise if you route around it. The cheapest way to make a synchronous last mile asynchronous again is to stop using the porch. Hand twenty per cent of the high value book to a locker network or a shop counter and you remove the coordination failure and the theft exposure in one move, and usually pay less per piece for the privilege. What you pay instead is the people who will not walk to a locker, so you will see fewer collected parcels in that cohort and a slower final handover for those buyers. Test one metro this week with one family of products. Do not roll anything out nationally before you have read the results.

And it is noise on claims only if somebody has actually read the liability wording, which almost nobody has. The temptation is to read authenticated delivery as job done, the carrier owns this now. Then a parcel is delivered against a valid code at two in the afternoon and is not there at five, or somebody forwarded the number into a family group chat in nine seconds, and the shape of the argument changes completely. The old sentence was there is no proof anyone delivered it. The new sentence is somebody clearly had the code. You will still be in that room, you will simply be asked to prove something different. Find out what, before you are asked.

Everything above assumes you are still running a single option, and peak season punishes that assumption harder than any rate card does. Splitting volume across two nationals and one regional costs you an integration each and buys you somewhere to move when one network starts failing, and one network always starts failing. Do the integration in October while the volumes are boring. Nobody in this industry has ever successfully stood up a new carrier connection in the week after a failure.

The trap to avoid is the one everybody falls into: treating this as a procurement problem and handing it to whoever owns contracts. None of it gets fixed by a better rate. It gets fixed by a phone number validated at checkout, by a message sent fifteen minutes after a label exists, by a postcode list that routes twenty problem destinations to lockers, and by somebody looking at exceptions at eight in the morning every day rather than at the end of the month. Those are operational things, they belong to whoever runs your shipping floor, and that person should be given the authority today rather than after the next review.

↑ Back to the key points

— By Nadia Pryce