On Tuesday afternoon at Leaders in Logistics: Autumn 2026, at the Business Design Centre in London, one line from a panel stayed with me. Isaac Arevalo, Director of Supply Chain at AG1, was describing what his team actually does all day. He said they joke internally that "our job is mainly relationship management and firefighting", sitting with carriers and 3PLs to understand the mapping and the data.
I sat in ten of the sessions over the two days. I think that one line explains most of the ones I heard. Almost every speaker told us to specialise, to partner or to mix carriers. Very few talked about the handover between them, the point where one partner passes a parcel, a scan or a status to the next. That handover is usually where the picture goes dark. I believe the speed of adding the next partner is now the number that decides who wins.
In short: The advice in London was to specialise and partner, and it is good advice. Every specialist you add is another handover. Somebody has to connect the partners. The winners will probably be the operators who can add a carrier, a customer or a market quickly, and still keep one promise and one picture. Almost nobody measures that speed today.
Because one network trying to do everything for everyone is getting harder to run at a profit. Aneira Pugh, Associate Partner at OC&C Strategy Consultants, put it as plainly as anyone did. In her words, "it is about specialisation rather than trying to do everything and doing nothing well".
Her argument was really about where the margin goes. In pick and pack, she said, "any element of non-standardisation immediately erodes efficiency, immediately hurts your margin". Her advice to the big volume players was probably not to chase the awkward few per cent of parcels that drag their economics down. The specialists, in her view, can double down on that work and charge properly for it.
She also gave a picture of where this could end up by 2035, and she was honest that not all of it will happen. It had a few national generic carriers, a handful of mission specialists, lockers as the default and the doorstep as the premium option.
James Edge, CEO of Landmark Global and a member of the bnode executive committee, described the same thing from the carrier side. Final mile now has far more options than it did, from regional carriers to delivery apps. "They're not necessarily all at the same service level, but you can start to mix them together," he said. That holds whether you are a logistics company running a multi-carrier strategy or a retailer doing it yourself.
I agree with all of it, to be honest. Specialisation is the right call for most of the businesses in that room.
It multiplies them. A retailer that follows the advice ends up with a national carrier, two or three specialists, an out-of-home partner and probably a fulfilment partner. Each one brings its own data, its own status codes and its own way of saying something went wrong.
Arevalo's own example made the point. The pan-European carriers, he said, "actually don't cover all of Europe", and they subcontract parts of it. His team spends its time working out which partner really covers which region and reading the data. They also try to get ahead of things like the flooding in Spain that week, before customers notice.
Paul Johnson-Barnett, VP of Operations EMEA at Centric Brands, described something similar in his interview. He works with five 3PLs across Europe and meets them daily to share the latest view. He was proud, rightly, that the relationship has moved from transactional to genuinely solving problems together.
He also said something I keep coming back to. Bringing a new brand into his European operation "takes a season", even with standard processes already written down. He was clear that much of that is real learning, because every brand's customers and factories are different. Nobody in the room asked how much of the season is the learning and how much is just getting the systems and the data talking.
I suspect that second part is where most of the time goes, for most of the companies I talk to. A daily call with five partners works, and it works because good people make it work. It just doesn't get any faster when the sixth partner arrives.
JD is the proof that control matters, from the one company in the room that can afford to own all of it. Natalie Frow, Managing Director of JD Logistics UK, was one of the most impressive speakers of the two days. I think the rest of us should take her seriously rather than explain her away.
Her reasoning was simple: "if you can control your supply chain end to end, then you can control your customer experience". JD builds its own automation, robotics and systems in-house. Frow said that lets them go from the keys of a new building to a fully operating automated fulfilment centre in about three months. She also said JD launched in the UK with the same kind of next-day and same-day promise it runs at home, from day one.
That only works because JD owns every handover inside its own walls. There is nothing to translate and nobody to call.
Even so, she was candid about the part that did not travel. "I think the biggest difference is probably our final mile, because that's much less replicable," she said. With lower density and fewer people at home, the UK final mile had to be built largely from scratch. Her next step, opening the capability to outside customers, is exactly where handovers JD does not own start to appear.
Mikael Redin of Ingka Group, IKEA's largest franchisee, showed a middle path that I found really interesting. IKEA's final mile used to be fully outsourced, on technology it did not own. It now runs a mix of its own fleet and outside carriers, with the allocation and routing decisions made in-house. IKEA did not take back all the vans, it took back the decisions and the ability to mix.
Redin described the change as moving "away from functional decision making into a more holistic approach, where we're able to actually tie customer offer and proposition with the logistics capabilities". That, to me, is what connecting the partners actually looks like for a company that is not JD.
There is one more concession to make, and it is a real one. Pugh gave a second route that avoids the problem entirely: impose standardisation, the way Fulfilled by Amazon makes sellers fit its system and turns messy flows into predictable ones. That works, of course. It works for whoever is big enough to make everyone else conform, which is not most of us.
Because when a customer cannot see a parcel, every partner in the chain looks like the problem. Mike Richmond, Chief Customer Officer at Royal Mail, was refreshingly direct about the order of priorities in his customer conversations. "They want price, price, price, price, price," he said, and he meant it.
Visibility came straight after, and not in the narrow sense. "They want to know where an item is at all times, but not just the tracking elements of it," he said. Retailers want visibility in bills and invoices, in claims, and an early warning when something is going to be late. It is very close to what retailers ask for whenever we talk to them.
Richmond also gave a sense of the scale behind that. He said Royal Mail has 1,400 people in its customer service function across ten locations, handling service and claims. The demand, he said, is huge.
He also warned about where this goes if nobody fixes the basics. AI businesses now sit on top of retailers to fight claims with carriers. You end up, in his words, "in this kind of crazy world where AI is talking to AI and sort of arguing with each other around parcel photos and visibility". His answer was first principles: don't lose the parcel, and fix the relationship.
Edge admitted Landmark's own version of the same gap with real honesty. The back end was strong, he said, but "the front end's not kept up with back end". Customers now expect the kind of data they get from their fitness apps.
One panellist on the new commerce panel put the customer's side of it very clearly. The job, they said, is "being proactive and making sure that people know when they're getting their product, even more importantly, when they're not".
This is where it comes back to the handovers. Every one of these speakers described visibility from inside their own network, and inside one network the data is usually rich. The retailer and the shopper see it across five networks at once. That is where the picture goes dark, and it goes dark at exactly the points where one partner hands over to the next.
In narrow, measurable jobs, mostly, and the speakers were more careful about this than I expected. Edge gave the clearest example. He said tariff classification tools that used to be about 80% accurate are now in the upper 90s. He also said Landmark has done some interesting things with AI agents in its call centres.
Sven Kukemelk, CCO at Omniva, described AI handling claims in writing rather than by voice. Voice models, he said, are not yet good enough in smaller languages like Estonian, Latvian and Lithuanian. He also described routing parcels on the recipient's instruction rather than the address on the label. "We don't care about the label on the parcel," he said, which sounds like a joke and is actually a serious design choice.
Edge was also the one who added the caution. On the return from AI, he said, "It's not well measured in larger organisations".
It looks slower and more people-heavy than the technology pitch suggests, and that is the concession I most want to make out loud. Edward Hill-Male, Head of Growth at Evri Fulfilment, talked about moving a live fulfilment operation to a new site while bringing new customers on at the same time.
His list of the ways transformation fails started with putting technology first and people second. He was clear that the measure is not the go-live date. "The actual metric for success during transformation is business as usual," he said, and he closed on "transformation is a capability, not a project".
Johnson-Barnett had the line on why this is hard to fund. "Flexibility costs, and having resilience is a benefit that you only ever see if it goes wrong," he said. The speed of adding a partner has exactly that problem. Nobody notices it until the day a carrier fails in peak and you need a second one by Friday.
Probably more of the same pressure, sooner, although nobody knows how soon. Richard Lim, CEO of Retail Economics, opened the event with agentic commerce, where an AI agent compares options on the shopper's behalf. In his view the agent weighs price, brand, delivery and returns before the shopper ever reaches a website. As he put it, "all of that decision criteria is being pulled to the very beginning of the customer journey".
Pugh went a step further in her 2035 picture, with agents choosing delivery providers for consumers and carriers picked by something close to a live auction. Edge's point about cross-border fits here too, since duties and taxes have to be calculated and charged before checkout, not after.
All of that assumes carriers are already connected, already speaking the same language and able to state a promise a machine can trust. Lim was honest about the timing. "I think the big unknown in my mind is not whether it's going to happen; it's just the speed of it happening," he said.
I wouldn't plan a business around a date. I would just notice that every version of that future rewards the operator whose partners are already connected.
The time it takes to add the next partner, measured honestly, is the one I would start with. Most operators know their cost per parcel to two decimal places. Very few could tell you how long their last carrier took to go live.
These are the measures I would take into Monday's meeting:
None of these needs a new system to measure. Connecting the partners does not mean buying a system and hoping the problem goes away. It means the people who know which partner is weak in which region spend their week on that, rather than reconnecting the same partner by hand. A spreadsheet and an honest afternoon will probably do it. If the answers come back in months rather than days, that is likely where the next few years will be won or lost.
This is the part of the problem we work on at GN TEQ, mostly for logistics companies that want to offer parcel without running a parcel network themselves. We help them connect new carriers in hours rather than months, onboard customers quickly, keep one tracking language across every carrier and stretch for peak without rebuilding the network.
Leaders in Logistics was a really good two days, and I'm grateful to every speaker for being so open about what is hard. The earlier version of this argument, from the Parcel Forum analyst day in Orlando, was that the information layer has to come together as networks split apart. London made me think the more useful question now is how quickly you can add the next piece to it.
Most speakers at the London event argued for specialisation, partnering and mixing carriers. JD Logistics UK made the case for owning everything end to end. Across sessions, retailers' second ask after price was visibility, and the gap nobody fully addressed was how quickly an operator can add and connect a new partner.
For most businesses, specialising and partnering is the realistic route, as OC&C's Aneira Pugh argued in London. Owning everything works for companies with JD's scale. In between, IKEA's approach is useful: it mixes its own fleet with outside carriers but keeps the allocation and routing decisions in-house.
Count the days from signing a carrier, customer or market to the first parcel tracked end to end without manual fixes. Add the people time the onboarding took, and how many of the new partner's events never reach your tracking. Most operators have never measured this, so the first number is usually revealing.
Because a parcel they cannot see becomes a complaint they cannot answer. Royal Mail's Mike Richmond said in London that retailers ask for price first and visibility second. That means not only tracking, but bills, invoices, claims and early warning of delays, across every carrier they use.