A logistics planner at a mid-cap shipper has twelve browser tabs open. Each is a different carrier portal. The shipment that just left Ningbo is on the third tab. The shipment heading into a customer dock window in Indianapolis is on the seventh. The expedited parcel for the QA sample is on the tenth.
She opens a spreadsheet on a thirteenth tab to consolidate ETAs. The spreadsheet was built last quarter, has a tab for each major carrier, and is the closest thing the team has to an operational view. It's also subtly wrong — ocean ETAs come from the carrier's calculation, road ETAs come from a forwarder's estimate, parcel ETAs come from a tracking page that hasn't updated in 36 hours.
The customer calls. They want to know when their order arrives. The planner promises to check and call back.
This is what "shipment visibility" looks like at most enterprise shippers. It's not a visibility problem. It's a synthesis problem.
Why portal-based tracking can't scale
Carrier portals work fine for one carrier. They start to break when you have more than three.
Each carrier exposes a different schema. MSC's milestone schema isn't Hapag-Lloyd's, which isn't FedEx's, which isn't your domestic LTL provider's. The team builds a mental model of what each portal calls "departed origin" and "in transit" and "out for delivery." Onboarding a new planner takes weeks because that mental model has to be retransmitted.
Worse, the portals don't agree on update cadence. A carrier with EDI sends milestone events within minutes. A carrier with API-only access updates on poll cycles. A carrier with neither requires a portal scrape that runs nightly, and the smaller partners with no API at all simply email you. The planner is looking at three different versions of "now" and being asked to make decisions on them.
By the time the team has 6-8 carriers across ocean, road, parcel, and air, the portal model is operationally broken. The spreadsheet is the only thing holding the operation together — and the spreadsheet only holds together because one specific person knows how to read it. We've written about the deeper structural version of this problem in eliminating visibility gaps across WMS and logistics systems.
The ETA confidence problem
Each carrier portal shows an ETA. Each carrier calculates that ETA differently. Some are pessimistic — they pad against missed cutoffs. Some are optimistic — they assume happy-path execution. Some show the ETA they think the customer wants to see, not the one they actually expect.
None of them come with a confidence band. Building one starts with scoring ETA accuracy against actual arrivals, carrier by carrier.
The planner is asked to make planning decisions against ETAs that have no calibrated accuracy. The result is one of two failure modes:
Over-plan defensively. Assume the optimistic ETAs are wrong, build buffer everywhere, expedite when uncertain. This works — it also costs 15-20% in freight margin you didn't have to spend.
Take the ETAs at face value. Plan tightly against carrier estimates, get burned when they slip, escalate to customer service after the customer has already called. This is the more common failure mode and the more expensive one in customer relationship terms.
ETA confidence scoring — calibrated per carrier, per lane, per service level, based on actual historical accuracy — is the structural answer. The team plans against numbers that come with a trust band attached.
What one shipment record actually means
The mental model shift is from "a shipment exists in each carrier's portal" to "a shipment is one operational record, written into by every carrier involved."
A real shipment record has all four of these properties:
One ID across all systems. The shipment exists once, regardless of how many carriers, modes, or forwarders touch it. Cross-references to carrier IDs, BOL numbers, container numbers, and PO lines hang off the same record, which is the mapping that makes it possible to unify carrier and freight forwarder tracking.
Normalized milestones across modes. Vessel departure, port arrival, customs clearance, dray pickup, DC receipt — every event maps to the same milestone schema, with consistent timestamps and source attribution. The planner reads one timeline, not twelve.
Coverage transparency. When a carrier doesn't post intermediate milestones, the gap is visible. The team knows what's actually tracked and what's an inference. No silent gaps.
Risk attached. Each shipment carries its current risk signal — port congestion, weather, predicted ETA slip, customs hold. The signal updates as the inputs do.
This is what gets unified when shipment data sits in one layer above the carriers rather than in twelve carrier portals. The architecture pattern is the same one we walk through in building a real-time supply chain control tower.
Risk: from receipt to prediction
The shift from "track shipments" to "predict shipment risk" is the conversion that makes the difference for customer service.
In the receipt model, the team finds out a shipment is late when it doesn't arrive. By then customer service has already taken the call. The conversation is reactive — why didn't this arrive? — and the planner is on the back foot.
In the prediction model, the team sees the risk hours before the milestone slips. A customs hold predicted at +6 hours, with the impacted PO, customer, and revenue at stake already attached. The planner notifies customer service. Customer service notifies the customer. By the time the slip happens, everyone affected already knows.
The shift sounds incremental. Operationally it's the difference between an apology and a heads-up. Customers remember the difference.
The signals that drive this are not magic. They're operational data — historical carrier reliability on the lane, current port congestion data, the carrier's posted ETA confidence trend, weather alerts on the route — combined into a risk score that updates as the inputs do. The infrastructure for surfacing those signals across the full visibility layer is the actual product.
How Orkestra fits
Orkestra unifies carrier and freight forwarder tracking across 200+ pre-built integrations into one shipment record. Every milestone normalized. Every ETA carries a confidence band trained on historical accuracy. Every shipment carries a risk score that updates as the inputs do. The carrier-portal jungle becomes one operational view.
A Tier-2 auto supplier uses the same architecture to surface predicted line-stop risk hours before customer impact, cutting expedited freight as a percentage of total freight by a measurable percentage every quarter — a pattern we cover in detail in how Tier-2 and Tier-3 auto suppliers keep OEM lines running. Multi-mode operators reconcile ocean, road, and parcel into one timeline without changing carriers.
See one shipment record · Book a walkthrough
Further reading
- Build a real-time supply chain control tower in 2026
- How Tier-2 and Tier-3 auto suppliers keep OEM lines running
- Eliminate visibility gaps across WMS and logistics systems
And if you'd rather see the architecture in product form: end-to-end performance reporting on the unified data.
