The complete guide to shipment tracking for freight forwarders.

Why proactive beats reactive

The switching data says customers act on this. In a survey commissioned by Logixboard with FreightWaves — a vendor-commissioned study, so read it as directional — 81% of shippers called visibility table stakes, and 46% said they had already left a forwarder for one with better technology. Not threatened to leave. Left. It is the strongest switching statistic published in this category, and even discounted for its sponsor, the direction is hard to argue with: the box moves at the same speed for every forwarder on the lane, so the thing customers can actually compare is how they were kept informed.

The reactive version of tracking has a name in e-commerce: WISMO — “where is my order?” In retail, where it is actually measured, WISMO runs 20–40% of all support tickets at roughly $5 per contact — retail figures, so extrapolate to freight with care. But the freight version of the question is the same question with a container number attached, and it arrives all day, on WhatsApp, marked urgent. Logi-Sys, a forwarding TMS vendor, names exactly this loop — customers chasing status by email and WhatsApp — as a driver of customer churn. Vendors of notification systems claim proactive updates cut inbound status inquiries by 35–55% (vendor figures, again directional): the mechanism is simply that a question already answered doesn’t get asked.

What the reactive loop costs your own team is arithmetic — run your shipment volumes through the calculator and see.

The weather: why late is structural

Proactive communication would matter less if shipments mostly ran on time. They don’t, and the numbers are worth staring at:

The point of this table is not that carriers are careless. It is that lateness is structural: carriers manage capacity by blanking sailings, networks route through congested transshipment hubs, and a schedule is a plan sailing into weather, strikes and port queues. A forwarder cannot promise on-time. What a forwarder can promise — the only thing fully within your control — is that the customer hears about the change from you, first, with a plan attached. That promise is deliverable every single time, and it is the one this guide is about keeping.

How forwarders track today

The honest description of tracking at most forwarding desks is a loop, run by hand, several times a day:

• The portal round. One browser tab per carrier line, each with its own login, its own search box, its own event vocabulary — visited container by container, re-keyed result by result.
• The Excel milestone sheet. Statuses pasted into a spreadsheet that is out of date the moment it is saved. An entire market of Excel tracking add-ins exists because this sheet exists.
• The WhatsApp storm. Customers who long ago stopped waiting for the sheet and simply ask — “urgent, where is my container?” — forty times a day, answered by a coordinator alt-tabbing back to the portals.

Underneath the loop sits an identifier hierarchy that trips up customers and new hires alike. A shipment is trackable by three different numbers, and they answer different questions:

The exception taxonomy

Strangely, no canonical taxonomy of ocean-freight exceptions exists in public — every vendor gestures at “exception management” without defining the exceptions. So here is the working taxonomy, each entry with its cause, its cost, and the one thing that separates a good forwarder from the rest: the correct next message.

Two of these deserve a second look. Rollover is the exception that defines whether your tracking is real: because carriers rarely announce it, the only reliable detection is watching the events — the box discharged at the transshipment hub, and then did not load on the booked onward vessel. During the 2020–21 capacity crunch, rollover rates at major hubs ran 28.5–39% — crisis-era, historical figures, but the mechanism that produced them never went away; it merely receded. And the demurrage/detention pair is where exceptions turn into invoices: demurrage is the meter inside the terminal, detention the meter outside it, and both start from events your customer never sees until the charge arrives — which is why the correct next message names the clock, not just the delay.

Milestones: ocean vs air

Ocean tracking runs on a chain of physical container events: empty gate-out → gate-in at origin → ATD → transshipment → ATA → full gate-out → empty return. Each link is a fact about a box, reported (eventually, unevenly) by carriers and terminals. Air runs on a different alphabet: IATA’s FSU status codes, attached to the air waybill:

The instinct is to assume air data is better than ocean — flights are faster, airports more digital. The published claim runs the other way: Riege, a forwarding-software vendor (so treat the figure as vendor-published), puts 40–60% of air FSU milestones as simply missing from the data stream. A tracking practice that only re-broadcasts carrier feeds inherits every one of those gaps; a tracking practice worth the name has to notice the absence of an expected milestone — the RCF that never arrived, the load event that never followed the discharge — and treat the silence itself as a signal.

The truth about ETAs

A carrier ETA is not a prediction. It is a plan — the schedule the carrier intends to keep — and it gets revised late and reluctantly, often after the reality it describes has already changed. That gap created a market: predictive-ETA vendors — Windward, FourKites, Shippeo among them — publish accuracy claims for machine-learned arrival estimates. Every one of those accuracy figures is vendor-published; no neutral benchmark exists, so treat the numbers as marketing until proven on your own lanes.

But the accuracy war is also the wrong war. Suppose the perfect predicted ETA existed — the question that actually matters at a forwarding desk is: what happens next? An ETA change is worth nothing until it changes someone’s plan: the trucker re-booked, the warehouse slot moved, the consignee’s production schedule warned, the customer’s own promise to their customer revised. The unit of value is not a percentage point of accuracy; it is the update that reached the person whose plan depended on the old date, in time to act. That is the difference between visibility platforms — which show a better number on a dashboard — and a tracking desk that does something with it. (The full comparison is at vs visibility platforms and vs tracking APIs.)

The POD that gates your invoice

Tracking is usually discussed as an operations problem. Its last milestone is a finance problem. The proof of delivery is money paperwork: for many customers, the payment clock on your invoice does not start at delivery — it starts when the signed POD reaches their accounts team. No POD, no clock, no cash.

Historically, a paper POD takes 5–14 days to make its way back from the delivery point to anyone’s accounts inbox — a driver’s cab, a depot tray, a scanner queue. Electronic-POD vendors claim moving to ePOD cuts days sales outstanding by 30–50% — the published ranges vary widely between studies, so hold the number loosely — but the mechanism is not in dispute: if the document that gates the invoice exists on the day of delivery instead of two weeks later, billing starts two weeks earlier. A tracking desk that treats DLV or full gate-out as the finish line has stopped one document short of the money.

What automation actually changes

Map software onto everything above and the change is five verbs — each replacing a portal round, a judgment call made at speed, or a message written under pressure:

• The desk watches. Continuous polling across carrier sources — ocean, air and express — replaces the portal round. Every shipment, every identifier level, at 03:00 as attentively as at 15:00; expected-but-missing milestones flagged, not just reported events.
• The desk classifies. Event patterns become named exceptions with reasons — the discharge/load mismatch becomes “rolled at the hub”, not “status changed” — so the customer note can say what happened and what happens next.
• The desk re-baselines. A classified exception produces a new plan: revised ETA, downstream legs re-computed, the free-time clock re-checked against the new dates.
• The desk tells first. The update is drafted in the customer’s language, on the customer’s channel — email or WhatsApp — before the question exists. One event, several parties, each note in the right tongue and tone.
• The desk collects. As deliveries complete, the POD is collected and attached to the shipment — so the document that gates your invoice exists on delivery day.

And the design principle that holds it together: the human gate. Every outbound update starts life as a draft with the full event history attached; your team approves what leaves the building. Update types earn their way to auto-send as trust builds — the routine departed/arrived notes first, the delay explanations to your biggest account last, or never. The step-by-step flow is on how it works.

Getting started: the 14-day path

The lowest-risk start is supervised and scoped, and it does not require replacing anything — the desk runs alongside CargoWise, Logi-Sys, Kale, Shipsy, Magaya, or no TMS at all:

• Days 1–3: scope. Map your live shipments, lanes and carriers; list your customers’ channels, languages and update preferences — who wants every move on WhatsApp, who wants a weekly email digest.
• Days 4–10: shadow. The desk watches your real shipments in parallel — classifying exceptions and drafting updates that go nowhere. Your coordinators compare its drafts against what they would have written, and against what actually got sent.
• Days 11–14: live, under approval. The first proactive updates go out — every send approved by your team — and POD collection switches on for completing deliveries.

Success is measured in your own operation: inbound “where’s my container?” messages per week (the target is the site’s namesake — they ask you never), minutes from exception to customer notification, and PODs attached on the day of delivery instead of weeks after it.