---
title: "Cross-Border Visibility in North America in 2026: Why 91% of Your ETA Error Sits Where GPS Cannot See it"
id: "26572"
type: "post"
slug: "cross-border-visibility-eta-error-north-america-2026"
published_at: "2026-09-15T14:00:00+00:00"
modified_at: "2026-09-15T13:23:27+00:00"
url: "https://locus.sh/blogs/cross-border-visibility-eta-error-north-america-2026/"
markdown_url: "https://locus.sh/blogs/cross-border-visibility-eta-error-north-america-2026.md"
excerpt: "A truck in a customs queue is still moving on your map. Most of the ETA error on a cross-border move lives in the release decision, which telemetry cannot observe and your broker already can."
taxonomy_category:
  - "General"
---

#### [General](https://locus.sh/blogs/category/general/)

# Cross-Border Visibility in North America in 2026: Why 91% of Your ETA Error Sits Where GPS Cannot See it

[Anas T](/author/anas_locus/)

Sep 15, 2026

15 mins read

Cross-border visibility in North America is the practice of tracking a shipment across a US, Canada or Mexico land crossing in a way that reflects its customs status, not just its vehicle position. It differs from domestic visibility because the two things decouple at the border: a truck can be moving toward a secondary inspection bay while the shipment is no closer to release, and it can sit motionless for three hours and be released in the next minute. The result is that most of the arrival-time uncertainty on a cross-border move sits in a release decision that no telemetry feed observes, while the instrumented portion of the journey contributes almost none of it. Locus, the world’s first agentic Transportation Management System, models the crossing as a branch in the shipment’s state rather than as a slow segment of road, so downstream commitments are planned against the outcome that actually governs them.

## Key Takeaways

- On a representative cross-border move, roughly 91% of total ETA error comes from the customs release process rather than from transit, so improving positional data addresses under 9% of the problem.
- Driving line-haul and urban transit error to zero would cut total ETA error by about 4%, because the dominant term is untouched by better telemetry.
- Release time is bimodal rather than noisy, so the mean ETA names a duration that almost no shipment actually takes.
- Knowing which release branch a shipment is on cuts the border-term error by about 74%, and that reduction holds steady whether the secondary referral rate is 5% or 40%.
- Locus joins customs and carrier status into one shipment state and plans downstream commitments per branch, with a leading retailer deployment resolving exceptions in under two hours.

## Why Cross-Border ETAs Fail Differently

The regulatory design of the border is what makes the information asymmetry possible. Under the advance electronic cargo rules codified at [19 CFR 123.92](https://www.ecfr.gov/current/title-19/chapter-I/part-123)
, CBP must receive truck cargo information no later than one hour before the carrier reaches the first US port of arrival, reduced to 30 minutes for shipments qualified under the FAST program. The [rule was established through the ACE truck manifest rulemaking](https://www.federalregister.gov/documents/2007/01/19/E7-762/required-advance-electronic-presentation-of-cargo-information-for-truck-carriers-ace-truck-manifest)
, and it means a determination is being formed against filed data while the truck is still an hour away.

That hour is the crux. A visibility platform watching the vehicle learns nothing during it, because the vehicle is simply driving. A platform joined to the filing and the broker’s status can learn, in that same hour, whether the paperwork is clean, whether a hold has been placed, and whether the shipment is tracking toward a routine release. The determinative information exists before arrival and sits in a system the shipper already pays for.

Wait times themselves are published rather than secret. CBP operates a public [border wait times service](https://bwt.cbp.gov/)
 reporting commercial lane delays by port and lane type, which means port-level congestion is an observable input rather than an unknowable. What is not published, and what dominates the uncertainty, is which individual shipment gets referred onward.

The downstream consequences are expensive because dwell propagates. Research from [ATRI has found detention of six hours or more at 39.3% of stops](https://truckingresearch.org/)
, and a border hold consumes the same driver clock that a receiver’s dock does. Under the federal hours-of-service rules at [49 CFR 395](https://www.ecfr.gov/current/title-49/subtitle-B/chapter-III/subchapter-B/part-395)
, time spent waiting counts against the driver’s on-duty window, so a three-hour secondary inspection can end the day rather than delay it.

| Also Read: Best TMS for Real-Time Visibility and Carrier Management in 2026 |
| --- |

## Where the ETA Error Actually Lives

Decomposing the arrival-time error on a representative cross-border move makes the investment question unambiguous. Take a line-haul transit error of 25 minutes, an urban approach error of 15 minutes and a border release error of 95 minutes, combined in the usual way for independent terms.

| Component | Error contributed | Share of total variance | Observable by telemetry |
| --- | --- | --- | --- |
| Line-haul transit | 25 minutes | 6.3% | Yes |
| Urban approach | 15 minutes | 2.3% | Yes |
| Border release | 95 minutes | 91.4% | No |
| Total | 99.4 minutes | 100% | Partially |

Halving both instrumented terms, which is a substantial engineering achievement, takes total error from 99.4 minutes to 96.1 minutes. Driving them to zero entirely, which is impossible, takes it to 95.0 minutes. The ceiling on everything telemetry can contribute is a 4.4% improvement, and any visibility roadmap that spends its budget on feed frequency and map matching is optimizing inside that ceiling.

This is not an argument against accurate tracking, which matters for many other reasons including theft, driver safety and yard management. It is an argument that on cross-border lanes, the ETA is not the thing tracking improves.

The decomposition can be run on data most shippers already hold, which is what makes it worth doing before the next visibility purchase. Take twelve months of cross-border moves on a single lane and record three timestamps per shipment: departure, arrival at the port of entry, and arrival at the receiving facility. The variance of the first interval gives the transit term, and the variance of the second gives the combined border and final-approach term. Running it per lane rather than per network matters, because a lane through a small port with a single commercial booth behaves nothing like a lane through a major crossing with dedicated FAST processing, and a blended figure will describe neither. Most teams that run this exercise find the split more lopsided than the illustrative numbers used here, not less.

## Release Time Is a Branch, Not a Delay

The border term is not simply large, it is shaped differently from the others. Transit error is roughly symmetric noise around a central value. Release time is bimodal: most shipments clear primary processing quickly, and a minority are referred onward into a process that takes hours.

Model it with 85% of shipments clearing primary at around 20 minutes and 15% referred to secondary at around 210 minutes. The mean release time is 48.5 minutes, and essentially no shipment takes 48.5 minutes. The average describes a duration that does not occur, which is why a single-number ETA on a cross-border lane feels wrong to the people who work the lane even when it is computed correctly.

| Secondary referral rate | Mean release time | Error without knowing the branch | Error knowing the branch | Reduction |
| --- | --- | --- | --- | --- |
| 5% | 30 minutes | 43 minutes | 13 minutes | 70% |
| 10% | 39 minutes | 59 minutes | 16 minutes | 73% |
| 15% | 48 minutes | 70 minutes | 18 minutes | 74% |
| 25% | 68 minutes | 85 minutes | 22 minutes | 74% |
| 40% | 96 minutes | 97 minutes | 26 minutes | 73% |

Two things stand out. Knowing which branch a shipment is on cuts the border-term error by around 74%, against the 4.4% ceiling available from better telemetry. And that reduction is remarkably stable across referral rates from 5% to 40%, which means the case for joining customs status does not depend on knowing your referral rate accurately, a number few operations actually measure.

| Also Read: Cross-Border Route Optimization in North America: Why the Border Is a Queue and Not a Road Segment |
| --- |

## How Status-Joined Cross-Border Visibility Works

### 1. Decompose your ETA error before buying more telemetry

Split historical arrival error into transit and border components on your actual lanes, then compare each against the cost of reducing it. Most teams discover the border term dominates so heavily that the ranking of every other investment changes.

### 2. Treat the crossing as a branch in shipment state, not a road segment

A branch has outcomes with probabilities attached, while a segment has a duration. Modeling the crossing as a segment forces the system to average two populations that behave nothing alike.

### 3. Join customs and broker status into the shipment record

The release signal lives with the broker and in the filing systems, not in the telematics feed. Bringing that status into the same object as the position is the single highest-value integration on a cross-border lane, and it is usually an unglamorous data project rather than a platform purchase.

### 4. Use the advance filing window as your prediction horizon

Because cargo information is filed at least an hour before arrival, that hour is available for prediction rather than for waiting. Systems that begin reasoning about the crossing only when the truck reaches the port discard the window in which the outcome is actually being determined.

### 5. Plan the downstream commitment per branch, not on the mean

Dock appointments, yard slots and onward legs should be planned against the branch the shipment is most likely on, with a prepared alternative for the other. Planning on the blended mean produces a commitment that fits neither population.

### 6. Track driver hours against the branch, not the average

A secondary referral does not merely delay the load, it can consume the remaining on-duty window and end the driver’s day. Hours-of-service feasibility should be evaluated against the secondary branch, because that is the branch that breaks it.

## Why Planning on the Mean Is Never the Right Answer

The practical consequence shows up at the first downstream commitment, usually a dock appointment. Using the same release model, with a primary branch at 20 minutes and a secondary branch at 210 minutes giving a mean of 48.5 minutes, the question is which of those three numbers the appointment should be booked against.

| Dock appointment tolerance | Plan on the mean | Plan on the primary branch | Plan on the secondary branch |
| --- | --- | --- | --- |
| Plus or minus 10 minutes | 0% | 85% | 15% |
| Plus or minus 15 minutes | 0% | 85% | 15% |
| Plus or minus 30 minutes | 85% | 85% | 15% |
| Plus or minus 60 minutes | 85% | 85% | 15% |

The mean never wins. At tight tolerances it fails completely, because the mean sits 28.5 minutes away from where 85% of shipments actually arrive, and at loose tolerances it merely ties the primary-branch plan by accident, succeeding only because the tolerance happens to be wider than that 28.5 minute gap. There is no tolerance at which averaging the two populations beats simply planning for the larger one.

Planning for the primary branch while pre-securing a fallback slot for the referred minority reaches roughly 97%, assuming four out of five fallbacks can be arranged. That is the whole argument in one number: the gain comes from treating the crossing as two populations with a prepared response for each, not from predicting a single arrival time more precisely.

| Also Read: Real-Time Visibility for 3PLs: How to Evaluate |
| --- |

## Position-Based Against Status-Joined Cross-Border Visibility

| Dimension | Position-based visibility | Status-joined visibility |
| --- | --- | --- |
| What is tracked | Vehicle location and movement | Vehicle location plus customs release state |
| Model of the crossing | Road segment with a travel time | Branch with outcomes and probabilities |
| ETA output | Single blended arrival time | Arrival time conditional on release branch |
| Prediction start | When the truck approaches the port | When cargo information is filed, at least an hour out |
| Behavior during a hold | Reports the vehicle as in transit or stopped | Reports the shipment as held pending release |
| Error reduction available | About 4% from perfect telemetry | About 74% on the border term |
| Downstream planning | One appointment on the mean | Primary commitment plus prepared fallback |

| Also Read: How to Evaluate Real-Time Visibility Platforms: 7 Tests |
| --- |

## What to Look for in Cross-Border Visibility Software

**Customs status as a first-class field, not a note.** The platform should hold filing status, hold status and release status as structured states that drive logic, rather than as free text a person reads. If release status cannot trigger a workflow, it is documentation rather than data.

**Branch-aware ETA output.** Ask whether the platform can express arrival as conditional outcomes rather than a single time with a confidence band. A wide confidence interval on a bimodal distribution is technically honest and operationally useless.

**Broker and filing integration on the roadmap you can verify.** Ask which customs brokers and filing platforms are already integrated in production, for which lanes, and what the status refresh interval is. This is where real capability differs most between vendors that look similar in a demo.

**Hours-of-service evaluation against the worst branch.** The system should flag when a secondary referral would make the remaining plan infeasible on driver hours, before the truck reaches the port rather than after.

**Port-level congestion as a separate input.** Published wait times describe the queue, not the individual shipment’s referral risk, and a platform should use them as distinct signals rather than blending them into one delay estimate.

## Status-Joined Visibility in Action

A leading North American retailer moving goods across ocean, rail and road consolidated [six legacy systems into a single execution layer](https://locus.sh/case-studies/retailer-multimodal-logistics-automation/)
 and now resolves exceptions in under two hours, with 99%+ on-time store delivery and 95%+ route compliance. Multimodal flows make the point sharply, because each modal handoff is another place where the vehicle state and the shipment state can diverge, and the two-hour resolution window is what allows a divergence to be caught while options remain.

A Fortune 50 enterprise running 51 sites and a 4,500-strong driver pool across a 120-country network moved [weekly execution rate from 75% to 92%](https://locus.sh/case-studies/fortune-50-parcel-centralized-dispatch/)
 and surfaced more than $14 million in unused capacity. The relevant mechanism is centralized decisioning over a network where no single planner could hold the state, which is the same condition that makes cross-border lanes hard to run from a dashboard.

## Common Cross-Border Visibility Mistakes to Avoid

**Buying feed frequency to fix cross-border ETAs.** More frequent position updates improve a term worth under 9% of the error while leaving the dominant term untouched.

**Reporting a single blended ETA on a bimodal lane.** The mean describes a duration almost no shipment takes, and planning against it produces commitments that suit neither the cleared nor the referred population.

**Treating a moving truck as a progressing shipment.** Movement toward a secondary bay is movement, and a system that equates the two will report progress precisely when the shipment has stopped advancing.

**Starting the prediction at the port.** The filing window opens at least an hour before arrival, and ignoring it discards the only period in which the outcome can be anticipated rather than observed.

## Where Locus Fits on Cross-Border Lanes

Locus, the world’s first Decision-Intelligent, Agentic TMS, holds the crossing as a state transition in the shipment rather than as a segment of road, so a held shipment reads as held even while the vehicle is moving. The [control tower](https://locus.sh/control-tower-software/)
 carries customs and carrier status in the same object as position, and the Carrier, Dispatch and Orchestrator agents replan downstream commitments when the release branch resolves, rather than waiting for a planner to notice a dashboard change. The six governance mechanisms, Explainability, Traceability, Evaluation, Autonomy Levels, Execution Sandbox and Human Review, determine which replans execute automatically and which route to a person, which matters on lanes where a wrong appointment cancellation is expensive. The platform evaluates against 250+ real-world constraints across 1,000+ carriers at 99.99% uptime, with 1.5B+ deliveries executed across 30+ countries.

Locus is [recognized by Gartner for seven consecutive years](https://locus.sh/analyst-recognition/)
, featured in the 2026 Hype Cycle for Supply Chain Execution and Logistics Technologies, named a Leader in TMS by QKS Group in the SPARK Matrix, and ranked #1 in Route Planning on G2’s 2026 Best Software Awards. In October 2025, Ingka Investments, the investment arm of Ingka Group, the world’s largest IKEA retailer, acquired Locus. Locus continues to operate independently.

Cross-border visibility in North America fails when the platform tracks the vehicle and the outcome is decided by the customs release, because around 91% of the arrival-time error sits in a release process telemetry cannot observe and perfect tracking would improve the total by roughly 4%. The release is bimodal rather than noisy, so the answer is not a better average but a branch, and knowing the branch cuts the border-term error by about 74% using information filed at least an hour before the truck arrives. Locus, the world’s first agentic TMS, joins customs and carrier status into a single shipment state and replans downstream commitments per branch. [Request a Locus cross-border visibility assessment](https://locus.sh/schedule-demo/)
 to see how your lane error splits between transit and release.

## Frequently Asked Questions

**What is cross-border visibility in North American logistics?** It is the tracking of a shipment across a US, Canada or Mexico land crossing in a way that reflects customs status alongside vehicle position. The two diverge at the border, because a truck can be moving toward secondary inspection while the shipment is no closer to release. Effective cross-border visibility therefore reports the shipment’s release state, not just where the vehicle is.

**Why are cross-border ETAs so inaccurate?** Because most of the error comes from the customs release process rather than from driving. On a representative lane where transit error is around 25 minutes and release error around 95 minutes, the release term accounts for roughly 91% of total variance. Improvements to GPS, traffic modeling and feed frequency address the smaller share.

**How far in advance must truck cargo information be filed with CBP?** Under 19 CFR 123.92, CBP must receive the cargo information no later than one hour before the carrier reaches the first US port of arrival, reduced to 30 minutes for shipments qualified under the FAST program. That window is the period in which a release determination begins forming, which makes it the natural prediction horizon.

**What is a secondary inspection referral and how long does it take?** A referral moves a shipment from routine primary processing into additional examination, which takes substantially longer and begins after the primary queue has already been served. Because only a minority of shipments are referred, release time is bimodal, and the average across both populations describes a duration that few individual shipments actually experience.

**Does better GPS tracking improve cross-border arrival times?** Only marginally. In the decomposition above, halving both transit error terms improves total ETA error by about 3%, and eliminating them entirely improves it by about 4%. Accurate tracking remains valuable for security, driver safety and yard management, but it is not the lever that fixes cross-border arrival accuracy.

**How should a shipper measure cross-border visibility performance?** Split historical arrival error into transit and release components by lane, and report them separately rather than as one accuracy figure. Then measure how often the release branch was known before the truck reached the port, because that is the metric that predicts whether downstream commitments could be planned rather than repaired.

MEET THE AUTHOR

Anas T

Senior Content Writer - Product Marketing

Anas is a product marketer at Locus who enjoys turning complex logistics problems into simple, clear stories. Outside of work, he’s usually unwinding with a book or catching a good movie or series.

### Related Tags:

[https://locus.sh/blogs/real-time-tracking-cpg-southeast-asia-2026/](https://locus.sh/blogs/real-time-tracking-cpg-southeast-asia-2026/)
#### [General](https://locus.sh/blogs/category/general/)

## [Real-Time Tracking for CPG in Southeast Asia in 2026: Why the Outlet has no Coordinate to Track Against](https://locus.sh/blogs/real-time-tracking-cpg-southeast-asia-2026/)

[Aseem Sinha](https://locus.sh/blogs/author/aseem_locus/)

Sep 15, 2026

Geofenced arrival detection assumes the destination has a reliable coordinate. In dense general trade it does not, and no fence radius produces a verified visit more than 7% of the time.

[Read more](https://locus.sh/blogs/real-time-tracking-cpg-southeast-asia-2026/)

## Cross-Border Visibility in North America in 2026: Why 91% of Your ETA Error Sits Where GPS Cannot See it

- Share
- [Print](javascript:window.print())
- [Download](#)
- [Schedule a Demo](https://locus.sh/schedule-demo/)

### Is your team spending more time on fixing logistics plan than running the operation?

- Agentic transportation management from order intake to freight settlement
- Route optimization built on 250+ real-world constraints
- AI-driven dispatch with automatic execution handling

20%Cost Reduction

66%Faster Planning Cycles

[Schedule a demo](/schedule-demo/)

Insights Worth Your Time

#### [General](https://locus.sh/blogs/category/general/)

## [Locus 2026 UK Consumer Survey: Why Returns Visibility is Now the Conversion Engine for AI-Driven Shopping in UK Retail](https://locus.sh/blogs/returns-visibility-conversion-engine-ai-shopping-uk-retail-locus-q2-2026-consumer-survey/)

[Aseem Sinha](https://locus.sh/blogs/author/aseem_locus/)

May 29, 2026

#### [General](https://locus.sh/blogs/category/general/)

## [Locus 2026 US Consumer Survey: Generative AI isn’t Just Changing How Consumers Shop, it’s Breaking the Demand Patterns US Retail Was Built On](https://locus.sh/blogs/generative-ai-shopping-effect-retail-fulfillment-operations-locus-q2-2026-consumer-survey/)

[Ishan Bhattacharya](https://locus.sh/blogs/author/ishan_locus/)

May 29, 2026

#### [General](https://locus.sh/blogs/category/general/)

## [Embedded vs Bolted-On AI: The Architecture Question European Logistics Buyers Are Asking](https://locus.sh/blogs/embedded-vs-bolted-on-ai-european-logistics-platform-architecture-business-benefits/)

[Aseem Sinha](https://locus.sh/blogs/author/aseem_locus/)

May 21, 2026

#### [General](https://locus.sh/blogs/category/general/)

## [Hybrid Fleet Management: How Owned, 3PL, Gig, ICE, and EV Capacity Actually Operate at Most Enterprises](https://locus.sh/blogs/three-workforce-fleet-reality-owned-3pl-gig-drivers/)

[Aseem Sinha](https://locus.sh/blogs/author/aseem_locus/)

May 7, 2026

#### [General](https://locus.sh/blogs/category/general/)

## [US Returns Hit $850 Billion in 2025: Why US Retailers Are Restructuring Reverse Logistics in 2026](https://locus.sh/blogs/850-billion-us-returns-ai-routing-reverse-logistics-2026/)

[Ishan Bhattacharya](https://locus.sh/blogs/author/ishan_locus/)

May 7, 2026