By Luis Lopez, AI transportation consultant, CEO of Go Hub.io Holdings Corp and subsidiaries, and host of the Freight Guru Podcast
Ocean carriers publish schedules, and importers build plans around them: when to book drayage, when to staff the warehouse, when to tell customers their order will arrive. Then the vessel is late. Or it skips the port. Or the sailing is canceled outright, what the industry calls a blank sailing. Or the container never made it on board and was rolled to the next vessel.
None of this is new. What is new is how much data is available to see it coming, and how well AI can combine that data into a usable warning.
Why published schedules are only a starting point
A carrier’s schedule is a plan, not a promise. Vessels can be delayed by weather, congestion at earlier ports, equipment problems or network changes. Carriers sometimes skip a port call to recover time, cancel a sailing when demand is weak, or roll containers when a vessel is full. Each of these shows up in different places and at different times, which is why importers often learn about them late.
What AI can combine
The useful signals already exist. The challenge is pulling them together for each specific container:
- AIS vessel positions. The Automatic Identification System broadcasts vessel positions, and that data is widely used for tracking. Comparing a vessel’s actual position and speed with where it should be on its schedule gives an early read on delay.
- Carrier advisories. Carriers announce blank sailings, port omissions, schedule changes and rollovers through advisories and customer notices. These are often emails or web notices that need to be read and matched to the bookings they affect.
- Terminal berth schedules. Terminals publish berth schedules showing when vessels are expected to berth. Changes there often signal trouble before the carrier’s own schedule is updated.
- History by service and port pair. Some services and some port pairs run closer to schedule than others. Historical arrival patterns for the same service help judge how much to trust the current estimate.
- Booking and milestone data. Gate-in, load-on-board and transshipment events tell you whether the container is actually on the vessel you are tracking.
AI’s job is to read the unstructured pieces, such as advisories, match everything to the right booking and container, and produce one estimate with a confidence level and a reason.
Predicting late arrival and rollover risk
Two predictions matter most to an importer.
Late arrival. If a vessel is behind its expected position, its berth window has moved, and the service has a history of slipping on this leg, the model can flag the arrival as likely late before the carrier updates the ETA. For more on how arrival estimates are built, see predictive ETA and freight visibility explained.
Rollover risk. If a container has gated in but there is no load-on-board event as the vessel’s departure approaches, or if an advisory mentions capacity problems on that sailing, the risk of rollover goes up. Flagging that early gives the importer time to ask questions or adjust plans.
Arrival delays also connect to what happens after the vessel berths. If several delayed vessels are now expected in the same window, the terminal may get congested. That side of the problem is covered in AI port congestion forecasting.
What importers do with the information
A better arrival estimate is only valuable if someone acts on it. The practical uses are:
- Warehouse labor. Shift receiving crews to match when containers will actually arrive, rather than when they were scheduled. The method is covered in AI warehouse labor planning.
- Drayage booking. Hold off on locking in a pickup date until arrival is more certain, or alert the drayage provider early so capacity is ready when the container becomes available. Container milestone tracking after discharge is covered in AI container tracking and drayage milestones.
- Customer promises. Update delivery dates for downstream customers before they ask, with an honest range rather than the original date.
- Safety stock. For products where a late container means empty shelves, the history of how reliable a service has been can inform how much buffer inventory to hold.
- Booking decisions. Over time, history by service and port pair can inform which services to prefer for time-sensitive freight.
As an AI transportation consultant, the first thing I ask importers is what they would change if they knew about a delay a week earlier. If the answer is “nothing,” the forecasting tool will not pay off until the downstream process can react.
Limits to keep in mind
- Carriers make decisions that data cannot predict. A blank sailing announced tomorrow may not show up in any vessel signal today.
- AIS data has gaps. Position reports can be delayed or missing in some areas.
- Matching is error-prone. Linking an advisory to the right booking depends on clean vessel names, voyage numbers and booking references. Mismatches create false alarms or missed ones.
- Transshipment adds uncertainty. A container moving through a hub port can miss its connection even when the first vessel is on time.
- Estimates should be ranges. A tool that gives a single exact arrival date with no confidence level is overpromising.
Getting started
Start with the containers that matter most: time-sensitive products or customers with firm delivery commitments. Track their vessels, collect carrier advisories in one place, and compare predicted arrivals with actual ones. That record shows where the tool is useful and builds the trust needed to change warehouse and drayage plans based on its warnings.
For more conversations on freight and AI, subscribe to the Freight Guru Podcast.
About the author: Luis Lopez is a Miami-based AI transportation consultant and logistics entrepreneur, the CEO of Go Hub.io Holdings Corp and subsidiaries, and host of the Freight Guru Podcast.
