A shipment can look healthy on a dashboard until the truck fails to arrive.
The ETA is still green. The status says “in transit.” The customer-service team has no exception alert. But at the dock, the appointment window closes, the receiving team moves on and a shipment that appeared under control becomes tomorrow’s problem.
This is how tracking data misleads. Not because someone has deliberately falsified it, but because supply chains turn physical movement into digital signals: scans, carrier milestones, GPS pings, manual updates and probability-based ETAs. Some are facts. Some are estimates. Some are late. Some are missing. Yet many visibility systems display them with the same confidence.
For logistics managers, 3PLs, freight forwarders, retailers, manufacturers and IT teams, that distinction is no longer technical housekeeping. It is the difference between planning and improvisation. A misleading event can send warehouse labor to the wrong dock, trigger an unnecessary customer-service escalation or hide an inbound delay until there is no good option left.
The scale makes the problem harder to dismiss. UN Trade and Development says around 80 percent of the volume of international trade in goods is carried by sea. Pitney Bowes says U.S. parcel volume reached 23.1 billion shipments in 2025. At that size, small defects in timestamps, milestone names or location codes multiply into missed appointments, excess safety stock, failed delivery promises and avoidable expediting.
Customers notice, too. McKinsey found that consumers rank on-time delivery above speed and that about half track orders to make sure shipments remain on schedule. DHL reports that 96 percent of e-commerce retailers say their logistics offering is important to securing sales, while B2B buyers increasingly expect consumer-grade speed, ease and transparency.
The Hidden Defects Behind Tracking
One clue to the depth of the problem is the number of standards bodies trying to fix it. GS1’s EPCIS standard captures the “what, where, when, why and how” of products and other assets, including movement, status and chain of custody. The Digital Container Shipping Association has published cross-carrier track-and-trace standards spanning nine ocean carriers and five shipment phases. UN/CEFACT framed multimodal track and trace around a deceptively simple question: “Where are my goods?”
The first reason tracking data misleads is that much of it is proxy data. Some events are generated directly by sensors or scans, but others are manual, delayed or reconstructed after the fact. International Maritime Organization guidance for shipborne AIS says voyage-related information, including destination and ETA, may need to be entered manually and updated during the voyage. DCSA has said milestone data such as load, discharge, gate-in and gate-out is often communicated inconsistently, with delays or not at all. A polished tracking timeline can therefore be less a live portrait than a cleaned-up translation of fragmented inputs.
The second reason is that missing data is often worse than bad data. In a 2024 study on supply chain visibility, researchers classified data sparseness into noise, bias and missing values. When sparseness reached 90 percent, visibility fell to 52 percent for noise, 65 percent for bias and just 32 percent for missing values. For retailers, distributors and manufacturers that depend on external carriers and upstream partners for truth, the most dangerous signal may be the one that never arrives.
The third problem is semantic confusion. An order is not a parcel. A parcel is not a pallet. A pallet is not a container. And a container movement is not the same thing as a customer delivery promise. GS1 and DCSA standards both try to solve this mismatch by defining common event structures across products, cargo, vessels, equipment and locations.
When companies skip that normalization and jump straight to dashboard labels, one word such as “arrived” can mean arrived at port, arrived at terminal, arrived at a destination city, arrived at the final-mile depot or arrived at the customer’s door. To an operations team, those are different realities.
Then there is the ETA, perhaps the most misunderstood number in logistics. It is not an observation. It is a forecast. In a 2025 study using global shipping-route data, machine-learning ETA models achieved a lower mean absolute error than captain-provided ETAs overall — 16.01 hours versus 22.15 hours — and outperformed captain estimates when arrival was still more than 75 hours away. Yet the captain-entered ETA became more accurate in the final three days because it was being manually refreshed.
The point is not that a model or a human always wins. It is that ETAs should be treated as conditional, time-sensitive probabilities rather than facts engraved into a tracking page.
Why the Errors Matter
For logistics managers, 3PL operators and freight forwarders, low-quality tracking data rarely fails quietly. It pushes labor planning toward guesswork, forces warehouse teams to cushion schedules and turns exception management into manual detective work. Standardized, real-time event exchange can improve decision-making and exception handling. Delayed, missing or inconsistent event data weakens both.
For e-commerce and retail, the damage is commercial as well as operational. A buyer may tolerate a modest delay more readily than a broken promise. That makes tracking data part of the customer experience, not a back-office feed. Delivery and returns now influence conversion, marketplace ratings and repeat purchases. A vague tracking page can create the same business effect as a late parcel: lost trust.
For manufacturers, IT and automation leaders, the same defects appear inside planning systems. Inbound parts may show as moving even when the most recent event is stale. Integrations become fragile because each carrier, ERP, CRM or TMS describes events differently. Automation then accelerates inconsistency instead of reducing it. A workflow triggered by an ambiguous status code can move faster than a manual process and still make the wrong decision.
How to Improve Data Quality
The first step is to treat raw events, not dashboard statuses, as the system of record. Store the original event, timestamp, source, object identifier and event type. Mark clearly whether the signal is observed, inferred or predicted. Every partner should describe the same shipment event in the same way, with the same object ID, location, timestamp and meaning. That is the direction pushed by EPCIS, DCSA’s cross-carrier event standards and UN/CEFACT’s multimodal traceability work.
The second step is to manage data quality as a live operating metric rather than a back-office cleanup task. In supply chain research, Brent Hazen and his co-authors argued that data quality should be acknowledged, measured, monitored and controlled. In practice, for tracking, that means watching completeness, timeliness, semantic consistency, duplicate events, conflicting events and chain-of-custody gaps. If one carrier feed is fast but incomplete, or another is rich but semantically inconsistent, the business should know that before a control tower paints the load green.
The third step is to separate actuals from estimates. A real gate-out scan should not look identical to an inferred milestone, and neither should be displayed like a modeled ETA. Use confidence windows where uncertainty is high. Refresh probabilities as the shipment gets closer to handoff points. A scan, an inference and a forecast are different kinds of truth.
The fourth step is to fix missingness before chasing sophistication. The 2024 sparseness study indicates that missing values do the greatest damage to visibility, which means coverage often matters more than elegant analytics. Sensor overlays and connected tags can add value, but not universally. A 2024 Logistics 4.0 case-study paper found that the payback from tracking and tracing devices depends on transaction volume, the ability to reuse tags, trip duration and supply chain structure.
For freight forwarders, manufacturers and ERP or CRM integration teams, the practical priority is usually more basic: standardize master data, expand carrier coverage, define milestone names and latency rules in data contracts, and use interoperable location identifiers such as UN/LOCODE before buying another glossy dashboard.
A useful test is simple. For every tracking event, ask what object it describes, who generated it, whether it was observed, inferred or predicted, when it happened, when it was received, what system changed it and how much confidence users should place in it. If those questions cannot be answered, the dashboard may still be attractive, but its truth is fragile.
What Visibility Really Means
Perfect visibility is impossible in supply chains shaped by handoffs, congestion, labor shifts, weather and partners running different systems. More data does not automatically mean better data, and real-time telemetry is not always worth the cost. But the evidence points to a clear order of operations: standardize semantics, reduce missing events, monitor data quality continuously and display uncertainty honestly.
The best tracking system does not pretend to be omniscient. It answers the supply chain’s oldest question — where are my goods? — with enough discipline to distinguish what is known from what is inferred and what is still unseen.
In logistics, that distinction is not cosmetic. It is the difference between visibility and illusion.