Gemini reported 92.3% schedule reliability across all arrivals over November and December last year, with Asia–North Europe and the Transatlantic above 95%. The global figure for July was 56.4%. Whatever else you think about the alliance reshuffle of early 2025, that gap is not a rounding error.
Maersk and Hapag-Lloyd bet the network on a design most of the industry had spent a decade moving away from, and the number says the bet worked.
What hub-and-spoke actually changes
Around 340 vessels and 3.7m TEU of capacity run across 29 shared mainline services and 29 shuttle services, coordinated through Hanseatic Global Terminals and partner facilities.
The mainlines call at a small number of hubs and nowhere else. Everything beyond a hub — and that is a lot of ports — is served by a shuttle that feeds it.
The reason this produces reliability is structural rather than operational. A mainline with six calls has six opportunities to lose its window; a mainline with three has three. Cut the calls and you cut the accumulation of delay, and you also make recovery possible: a hub with a dedicated berth window can absorb a late arrival in a way a shared terminal on a crowded rotation cannot. Both carriers recorded the lowest average delay days among twelve global carriers, which is the same finding stated from the other end.
Compare that with what the rest of the market has been doing in North Asia this month — ships waiting ten days to berth, calls being merged and omitted after the fact. Gemini merged two Asia shuttle services into one string and dropped Shanghai calls too. The difference is that on their design, that is a shuttle decision. On a direct network, it is a mainline decision, and the mainline is where your ETA lives.
The part that costs you
Your box gets transhipped.
Every transhipment is a discharge, a yard dwell and a reload, and each of those is a chance for the container to miss its connection. Network reliability of 92% is measured on vessel arrivals, not on your specific box completing two moves and a connection. Those are different questions, and only the first one is in the published number.
So the honest framing is: the mainline is far more likely to be where it said it would be, and your cargo has more handoffs to survive on the way to it. For a full-container programme with predictable volumes into a major gateway, that trade is usually worth taking. For urgent or time-definite cargo into a secondary port, a slower but direct service on a less reliable carrier can still beat it, because you are removing a failure mode rather than a delay.
This is the calculation I see forwarders skip most often. They compare reliability percentages carrier to carrier without noticing they are comparing a direct call against a call plus a connection.
What to check before you route on it
Find out whether your port is a hub or a spoke on the service you are quoting — it is the single fact that determines whether that 92% applies to your booking or to the leg before it. Ask for the connection buffer at the hub, in days, and ask what happens when it is missed: next shuttle, or next week.
Two things are moving that will change this picture. The IMX service linking India, the Middle East and the Mediterranean went back through the Red Sea and Suez under naval protection from mid-February, which shortens the rotation and changes where slack sits in it. And Hapag-Lloyd has signalled it will fold some capacity from the ZIM acquisition into Gemini, though how much has not been confirmed. Added capacity has to be absorbed somewhere, and network integrations are exactly where reliability records get tested.
You can compare the current service picture on the carrier pages, and see which strings are actually holding their windows this week on sailing schedules.
If the 92% survives both the Suez re-routings and the ZIM integration, the rest of the market will have to copy the design. If it does not, we will find out that the number was mostly a function of running a young network below its capacity.
