Tradlinx Insight Report, November 2026

When AI manages the ETA,
how much do logistics operating costs fall?

Average operating cost

15.5%lower

For truck booking, warehouse staffing and customs
Compared with running on the carrier ETA

  • 25%less time trucks and staff spent waiting for cargoUp to 65% in a single case
  • Up to 56%lower operating cost in a single caseCustoms staffing at a Brazilian 3PL
  • 31%smaller error in predicted arrival timeAverage 15.4 h → 10.6 h
  • 64cases that cut operating costs by 30% or moreOut of 320

Based on 320 logistics cases in the US and Latin America (191 shippers, 129 forwarders and 3PLs)

The short version

Every percentage below shows how much lower operating costs were
than when the same cargo was run on the carrier ETA.

  • 15.5%Trucks, staff, customs

    Tasks handled right before arrival saw the largest drop

    • Truck booking, warehouse unloading staff, customs preparation, rail booking: 15.5% lower on average (223 cases)
    • Customer delivery commitments, safety stock orders: 2.7% lower on average
    • All 320 cases: 12.9% lower

    The first group of tasks is confirmed in the last 5 days before the vessel arrives.
    When AI-ETA flags a delay early, the schedule can move before anything is confirmed.

  • 27%Customs and inspections

    By task, customs and warehouses gained the most

    • Customs and inspections: 27% lower
    • Warehouses and CFS (facilities that consolidate small shipments from many shippers): 24% lower
    • Rail connections: 16% lower
    • Shipper industries: machinery and chemicals 18%, automotive and batteries 14%

    All of these assign staff a day or two before arrival.
    An early warning cuts the cost of people waiting for a ship that has not arrived.

  • 15.8%Transshipped cargo

    Transshipped cargo saw close to twice the drop of direct cargo

    • Transshipped cargo (cargo that changes ships at a hub): 15.8% lower
    • Direct cargo: 8.4% lower
    • Leading regions: US Gulf Coast 23%, Central America and Caribbean 21%, South America 17%

    A missed connection at a hub reaches the carrier ETA late.
    Regions with more transshipment therefore gained more from AI-ETA.

  • 93%Kept with one check a day

    Knowing about a delay early is what lowers costs

    • Carrier ETA: sometimes changes days after the vessel falls behind
    • AI-ETA: shows the delay before trucks or staff are confirmed
    • Checking only once a day: still keeps 93% of the savings

    Early knowledge of the delay mattered more than how often anyone checked.

  • About 21%If early-voyage accuracy holds

    15.5% is a conservative figure

    • Assumption used here: AI-ETA is more accurate than the carrier ETA only in the last quarter of the voyage
    • If it proves more accurate earlier too: the figure rises from 15.5% to about 21%

Same vessel, two ETAs

  • Task: trucking import containers from the Port of Savannah to an inland warehouse
  • Problem: the vessel ran more than four days late, and the carrier ETA showed it much later
  • Who: the same thing happens whether the shipper or a forwarder books the trucks

What each planner did at each point, and the cost that built up.

10 days before arrivalVessel falls 4+ days behind
3 days before
Half a day before
2.6 days after planned arrival
3.1 days after
4.7 days after
5.6 days afterContainer ready for pickup
Dispatching on the carrier ETARunning cost$1,410
Dispatching on AI-ETARunning cost$0
Vessel falls 4+ days behind
Container ready for pickup

Delay not visible, original schedule kept

Truck confirmed for the original schedule

Truck arrives, no container. 12 hours of waiting+$960

Truck leaves empty and is rebooked+$450

Carrier ETA finally shows the delay

Rebooked truck picks up the container

Delay spotted. Truck booking moved 4 days laterChange cost $0

Truck confirmed for the new schedule

No waiting

Container picked up as the truck arrivesExtra cost $0

Trucks booked on the carrier ETA$1,410$960 truck waiting time, $450 empty return and rebooking
Trucks booked on AI-ETA$0The truck arrives when the container is ready

About this case

  • Who: a forwarder in Savannah, US, assigning drayage trucks and chassis
  • Route: Busan → Savannah (via the Panama Canal)
  • Leg: Savannah terminal → inland warehouse (by truck)
  • Average across all cargo in this task: operating cost per container $157 → $91

A chassis is the trailer a container sits on.

Knowing before the carrier ETA does lowers costs

Carrier ETAs pick up delays late

  • The ETA can stay unchanged for days after a delay at sea
  • Sometimes it is missing altogether

Missed connections at a hub show up especially late

  • Missing a connection means waiting 3–8 days for the next ship
  • The carrier ETA often changes only when that ship sails

The big gaps sit in a few shipments

  • For most cargo the two ETAs are close
  • Large costs come from the roughly 5% of shipments where the carrier ETA is off by two days or more

Costs fall only if you know before you confirm

  • Changing trucks, staff or alternate transport after confirming them costs money
  • Tasks confirmed right before arrival gain the most

Where AI-ETA’s 4.8-hour accuracy gain comes from

  • Mean error: 15.4 hours for the carrier ETA, 10.6 hours for AI-ETA (Tradlinx AI-ETA, 508,055 predictions validated)
  • 84% of the 4.8-hour gap: from the 4.99% of shipments where the carrier ETA was off by 48 hours or more
4.99% of shipments with carrier ETA errors of 48 hours or moreOther 95%

Why near-arrival tasks gained the most

  • When AI-ETA gets ahead: on a 30-day voyage, it shows delays before the carrier ETA from about a week before arrival (the conservative basis used here)
  • Near-arrival tasks are confirmed inside that week: truck booking, unloading staff and customs preparation are fixed 1 to 5 days before arrival, so there is still time to move the schedule
  • Small misses cost money right away: a waiting truck costs $70–95 per hour, and waiting unloading staff $120–300 per hour
  • Tasks set weeks ahead finish before then: safety stock and customer delivery dates are fixed 2 to 3 weeks out, when the two ETAs barely differ
Both ETAs treated as equally accurate here
AI-ETA more accurate
DepartureArrival
Safety stock orders3+ weeks before arrivalSavings not counted
Customer delivery dates2 to 3 weeks beforeSavings not counted
Trucks, staff, customsLast 5 daysSavings counted

If data confirms early-voyage accuracy, safety stock and customer commitments can benefit as well.
On routes with a hub close to the destination, such as the Caribbean and Panama, missed connections happen inside this stretch, so the effect was especially large.

Savings measured across 320 cases

15.5%

Lower operating cost for trucks, staff and customs

Average of 223 cases confirmed in the last 5 days before arrival
95% confidence interval 15.0–16.0% (five repeated runs)

56%

Largest drop in a single case

Joinville customs staffing, 3PL (transshipment)

12.9%

Lower operating cost across all 320 cases

Including customer commitments and safety stock (2.7% lower)

Number of cases by size of cost drop

Each row includes the rows above it.

30% or more lower
6420%
20% or more lower
10332%
10% or more lower
18056%
5% or more lower
21567%
Confirmed cost drop
24075%
  • Another 51 cases: cheaper and costlier shipments mixed, with no clear difference (35 of them are tasks decided earlier)
  • In 29 cases operating costs rose, for four reasons
    • Customer delivery commitments (14): 2 to 3 weeks out, when the promise is made, AI-ETA is no more accurate, so there was nothing to save. Changing the promise each time AI-ETA moved added cost (promise changes per shipment 0.17 → 0.30)
    • Operations that rebook trucks too often (6): with a rule to rebook whenever the ETA moves six hours, AI-ETA’s three-hourly updates raised rebookings from 0.4 to 1.3 per shipment, and the rebooking cost outgrew the savings
    • Direct services where the carrier ETA is already accurate (4): on these routes AI-ETA’s error was slightly larger (for example 14.3 h against 14.8 h), and following it to change plans cost money
    • Others (5): two safety stock cases, two temporary storage and promotion schedules set weeks ahead, and one Suez routing where AI-ETA accuracy was also set lower for the crisis

The 5 cases with the largest cost drop

CaseRegionRouteCost dropSaving per container
Joinville customs staffing3PL (Joinville)South AmericaTransshipment56.2%$114
Campinas customs staffing3PL (Campinas)South AmericaTransshipment55.1%$111
Valparaíso customs staffingGlobal forwarder (Valparaíso)South AmericaDirect services52.3%$90
Bogotá DC receiving slotsE-commerce company (Bogotá)South AmericaTransshipment52.2%$80
New Orleans customs staffingRegional forwarder (New Orleans)US Gulf CoastTransshipment52.1%$295

What the top cases share

  • Cargo transshipped into Latin America and the US Gulf
  • Staff assigned to customs inspections or distribution center receiving
  • Decisions made a day or two before arrival, inside the stretch where AI-ETA is more accurate

Where does it pay most?

Forwarder and 3PL tasks

  • Customs, inspections and warehouses: costs fell in almost every case
  • Why: staff are assigned 1 to 3 days before arrival, so an early warning cuts waiting costs right away
Customs and inspections17 cases, up to 56% lower
26.8% lower
Warehouse, CFS and transload26 cases, up to 47% lower
24.2% lower
Rail and border connections15 cases, up to 40% lower
16.5% lower
Customer ETA notices24 cases, up to 45% lower
11.3% lower
Rerouting and expedites18 cases, up to 24% lower
11.1% lower
Inland trucking and dispatch29 cases, up to 47% lower
5.0% higher

Bars show the cost drop. Longer to the right means a larger drop; left of zero means costs rose.

  • Inland trucking (moving containers from port to warehouse): costs rose on average
  • Why: it includes 5 operations that rebook trucks whenever the ETA moves by six hours
  • The other cases, with a rule for when to rebook: 21.0% lower

Shippers by industry

  • Machinery, chemicals and automotive: about 92% of operating cost comes from production disruption and expedites
  • How they saved: knowing a delay early, they flew in parts, moved stock from another plant or adjusted output in time
  • Retail and food: more promotions and customer commitments set weeks ahead, so the overall figure is lower

For near-arrival tasks alone, costs fell by 12.9% in retail and 10.5% in food.

Machinery, equipment and chemicals46 cases, up to 47% lower
17.8% lower
Automotive, parts and batteries26 cases, up to 41% lower
14.1% lower
Electronics, appliances and chips27 cases, up to 28% lower
8.2% lower
Retail and e-commerce65 cases, up to 52% lower
8.0% lower
Food, cold chain and pharma27 cases, up to 51% lower
4.0% lower

Bars show the cost drop. Longer to the right means a larger drop; left of zero means costs rose.

By destination region

  • US Gulf, Central America and Caribbean, South America: much of the cargo moves through hubs such as Kingston (Jamaica), Cartagena (Colombia), Balboa (Panama) and Singapore
  • US West Coast: voyages from Asia take about 15 days, so the stretch where AI-ETA is more accurate was set at only about 4 days and the result came out small

If early-voyage accuracy is confirmed, West Coast costs fall by about 22%.

US Gulf Coast23 cases, up to 52% lower
22.7% lower
Central America and Caribbean28 cases, up to 49% lower
21.0% lower
South America90 cases, up to 56% lower
16.7% lower
US East Coast57 cases, up to 42% lower
10.8% lower
Mexico35 cases, up to 31% lower
10.0% lower
Canada9 cases, up to 17% lower
7.0% lower
US West Coast66 cases, up to 26% lower
2.8% lower
US exports12 cases, up to 4% lower
2.5% higher

Bars show the cost drop. Longer to the right means a larger drop; left of zero means costs rose.

  • US exports: mostly customer arrival dates promised two to three weeks ahead, so almost no savings are counted. With early-voyage accuracy confirmed, about 22% lower

By route type

  • Transshipped cargo: close to twice the cost drop of direct cargo
  • Transshipped at Kingston, Cartagena or Balboa, close to the destination: 23.1% lower
Transshipment routes144 cases, up to 56% lower
15.8% lower
Direct services176 cases, up to 52% lower
8.4% lower

Bars show the cost drop. Longer to the right means a larger drop; left of zero means costs rose.

By company size (shippers)

  • Large and mid-sized or smaller companies: similar results
  • Why: a similar task mix (production and expedites make up 83% and 82% of operating cost)

Companies of any size see results when AI-ETA feeds near-arrival tasks.

Large enterprises97 cases, up to 49% lower
14.0% lower
Mid-size and small firms94 cases, up to 52% lower
11.7% lower

Bars show the cost drop. Longer to the right means a larger drop; left of zero means costs rose.

How the savings show up

  • With AI-ETA, planners change plans a little more often
  • Rebooking trucks or using air and other alternate transport costs more, but the losses avoided are far larger
  • The amounts below are savings per container for each task
$25–44

Port-to-warehouse trucking

Fewer trucks leaving empty and less driver waiting time
$42–111

Unloading staff and receiving appointments

Less staff waiting and fewer warehouse appointments rebooked
$18–52

Rail and customs

Saved per container from fewer rail rebookings and less idle customs staff
$326–1,595

Production parts and materials

Saved per shipment from fewer line disruptions
  • Time trucks, staff and equipment spent waiting for cargo: down 25% per shipment, from 8.4 to 6.3 hours on average (169 scheduling tasks)
  • Hours short of production parts: down 15% (expedite tasks)
  • Amount ranges: the middle half of cases for each task, ranked by amount
  • Extra action cost: about $0.20 more spent on truck rebooking and alternate transport for every dollar of loss avoided

Chemical plant in Houston, transshipped at Kingston, Jamaica

Knowing about the missed connection early, the plant adjusted its process

  1. 7 days before arrivalThe mother vessel fell almost six days behind. Both ETAs reflected it two days later.
  2. 1 day after planned arrivalArriving late, the vessel missed its Houston connection in Kingston. The next ship was six days away.
  3. 10 hours laterAI-ETA reflected the missed connection, and the plant changed its process settings well before the decision deadline (5 days after planned arrival).
  4. 7 days after planned arrivalThe carrier ETA changed only after the deadline, around the time the next ship sailed.
On the carrier ETA$503,00096 hours without feedstock
On AI-ETA$25,000Process adjusted ahead of time. This plant saves $5,284 per shipment on average (29%)

Medical device plant in Costa Rica, transshipped at Cartagena, Colombia

Knowing about the connection delay, the plant reassigned receiving staff in time

  1. 3 days before arrivalThe cargo missed its connection in Cartagena and faced five and a half more days of waiting.
  2. 2.5 days beforeAI-ETA reflected the missed connection. The carrier ETA did not change before arrival.
  3. 2 days beforeReceiving staff were locked in. The AI-ETA plan was 17 hours from the actual arrival; the carrier ETA plan was 74 hours off.
On the carrier ETA$1,940Idle staff and reassignment
On AI-ETA$21Staff matched to arrival. Average saving $141 per container (43%)

Battery plant in Georgia, direct service from Busan

Stock from another plant arrived before the line stopped

  1. 13 days before arrivalThe battery cell shipment picked up a delay of almost five days that the carrier did not announce.
  2. 1.5 days beforeAn announced delay of four and a half days followed. On the carrier ETA alone, existing buffer stock looked sufficient.
  3. 3.7 days after planned arrivalAI-ETA captured the hidden delay as well, and the plant moved stock from another plant by truck (36 hours, $13,500 including a rush surcharge).
  4. 9.7 days after planned arrivalThe carrier ETA reflected the delay only after the decision deadline.
On the carrier ETA$674,00096 hours of line disruption
On AI-ETA$15,0000.3 hours of line disruption. Average saving $3,565 per shipment (27%)

Getting more out of AI-ETA

What should a company using AI-ETA focus on to get more from it?
These are the practices that paid off most across the 320 cases.

  • Check tasks confirmed 1 to 5 days before arrival first

    • Results: tasks confirmed right before arrival, such as truck booking, unloading staff and customs preparation, cut operating costs by 15.5% on average, with customs and inspections at 27% and warehouses at 24%
    • Tradlinx AI-ETA: predicts arrival for cargo close to port with a mean error of 9.2 hours, and 92.5% of shipments arrive within ±24 hours of the prediction

    How to use itMake checking AI-ETA the step before trucks and staff are confirmed. Book warehouse and yard slots using AI-ETA’s probability of arrival within 24 hours to cut both double bookings and empty slots.

  • Look more closely at transshipped cargo

    • Results: transshipped cargo cut operating costs by 15.8%, close to double the 8.4% for direct cargo. Cargo transshipped at Kingston, Cartagena or Balboa, close to the destination, reached 23.1%
    • Tradlinx AI-ETA: carrier ETAs that are off by more than two days occur mostly on long-haul and transshipment journeys or amid weather and port congestion, and AI-ETA cut the error on these shipments from 90.0 to 9.2 hours on average. It also fills hub legs where carriers often leave the ETA blank

    How to use itGroup transshipped cargo and review its AI-ETA every day. If the ETA slips sharply at the hub, recheck truck and staff schedules at the destination right away.

  • Check AI-ETA at least once a day

    • Results: checking just once a day captured 93% of the savings. Receiving it automatically every three hours captures the rest
    • Tradlinx AI-ETA: updates every 2 to 4 hours and shows its own prediction where the carrier leaves the ETA blank

    How to use itMake a daily review of AI-ETA for cargo due within a week a standard task at a set time. You can start before system integration is finished.

  • Prepare your response to schedule changes in advance

    • Without a response rule: the 6 cases that rebooked trucks whenever the ETA moved six hours saw rebookings rise from 0.4 to 1.3 per shipment, and operating costs rose by 84%
    • With a response rule: switching to “only when the ETA moves 24 hours or more, at most twice per shipment” turned the result into a 12% average drop (21–31% lower in three cases, little change in the other three)

    How to use itDecide in advance what to do at each size of delay. Leave small shifts alone, rebook trucks and staff once a delay passes a set number of hours, and consider alternate transport or production changes for large delays.

All 320 cases

Amounts are US dollars per shipment, mostly 40 ft containers. Search by city, task or industry.

IDCaseTask or industryDecisionCost on carrier ETASavingDropResult

How we compared operations with and without AI-ETA

  • You cannot ship the same container twice
  • Market conditions change between before and after adoption, and other companies differ in size and routes
  • So we compared two teams running the same cargo
  1. Prepare identical cargo

    Each case sets its route (including transshipment), delay frequency, warehouse and trucking conditions and unit costs to field conditions, with 40,000 to 120,000 shipments.

  2. Both teams operate the same way

    Team A checks the carrier ETA once a day. Team B receives AI-ETA every three hours. Vessels, delays, warehouses and trucks are identical.

  3. Decide only on what was visible then

    Both teams decide on the ETA visible at the decision deadline. Each team’s trigger rule was tuned separately to its own advantage.

  4. Compare the two teams’ costs

    After arrival, all costs are added up and compared, including any cost from acting on AI-ETA when it turned out wrong.

AI-ETA gain set below the measured value

In these cases AI-ETA reduced the mean error by 1.7 hours, less than the 4.8 hours Tradlinx measured.

A narrow stretch for AI-ETA’s lead

AI-ETA was assumed to be more accurate than the carrier ETA only in the last quarter of the voyage (about 7 to 8 days before arrival on a 30-day voyage).

Extra caution on crisis routes

For Red Sea routings, Panama Canal restrictions and hub congestion, AI-ETA accuracy was lowered as well.

A conservative way to average

The cost drop is total savings across cases divided by total operating cost on the carrier ETA. A simple average of each case’s drop would be higher: 17.0% for near-arrival tasks and 13.4% overall.

What is compared
Operations on the carrier ETA (checked once a day) versus AI-ETA (received automatically every three hours)
AI-ETA accuracy
Tradlinx AI-ETA Technical Report (May 2026, 508,055 predictions): mean error from 15.4 to 10.6 hours
Near-arrival tasks
Truck booking, warehouse unloading staff, receiving appointments, customs, rail booking and expedite tasks whose final decision falls in the last 5 days before arrival (or before the goods are needed, for expedites) (223 cases)
Case mix
120 US-focused cases and 200 Latin America and transshipment cases. 191 shippers, 129 forwarders and 3PLs. Task mix for the US-focused 120: 55% scheduling, 21% expedites, 19% customer commitments, 5% safety stock; for the other 200: 56%, 22%, 18% and 4%.
Transshipment
A missed connection at the hub means 3–8 days of waiting for the next ship. The carrier ETA reflects it within about a day half the time, and otherwise around when the next ship sails
Field conditions
Carrier ETA mean error 14–35 hours on direct services and 29–48 hours with transshipment (whole voyage, reflecting lower 2026 schedule reliability). Median baseline on-time rate for stock-out and delivery tasks 94%. Cases such as decisions closed before departure or commitments that cannot change were included at a similar share (7.5%) to the US-focused set (6.7%).
Reading the numbers
The cost drop refers to the operating cost of the task, not total company logistics spend. The same container touches several tasks, so task savings should not be added together. Latin American labor and trucking costs are indexed to US levels, and results will vary with each company’s actual rates.

Unit costs (2026 market data)

ItemMarket rateUsed in the casesSource
LA/Long Beach drayage (port to warehouse trucking)About $471 within 25 miles, about $650 to Ontario, CATruck rebooking $350–500Gateway Lines
Truck detention and dry runs (US)$50–100 per hour, $200–500 per event$70–95 per hour, $350–500 per eventDashdoc, FreightAmigo
Demurrage (US)About $200 a day after free time, about $350 for long stays$240 a day after 4 free days, $384 after day 10 (yard and chassis about $43 a day during free time)FreightFigures
D&D (Brazil imports, 40 ft)5 free days, $90 a day for days 6–10, $175 for days 11–14 (from January 15, 2026)Same tiersMaersk
Warehouse labor (US)Warehousing and storage average hourly earnings $26.87 (production workers $26.02, September 2026, preliminary)Idle unloading staff $120–300 per hourUS Bureau of Labor Statistics
Labor (Mexico and Latin America)Mexico minimum wage 315.04 pesos a day (about $17, 2026)Labor 0.3–0.45x and trucking 0.4–0.45x US levelsBloomberg Law
Air freight (China to LAX)$5.85–7.69 per kg (October 2026)Air replenishment $18,000–45,000 per moveFreight Right, via press
Plant downtime$2.3 million per hour at a large auto plantCritical cargo shortage $800–7,000 per hour (conservative)Siemens 2024

Recent ocean and port conditions

Red Sea and Suez
Even after Houthi attacks on vessels resumed in July 2026, 63% of Asia to Europe capacity returned to the Red Sea routing (September 15 to October 15). Suez transits in 2026 average about 41 a day, below about 74 in 2023. Seatrade Maritime, Ship & Bunker, The Rio Times
Schedule reliability
Global schedule reliability 49.9% in August 2026 (lowest since September 2022); 20% of Asia to North America arrivals on time in September. Sea-Intelligence, Xeneta
Transshipment hubs
Singapore on-time berthing 32.4% in August, connection risk at Colombo, and Caribbean and Panama services changing every 3.9 days on average. Seatrade Maritime, Kuehne+Nagel, The Loadstar
South America
Santos berth waiting time rose from 8–9 hours in 2019 to about 35 hours in 2024, with terminal congestion continuing in 2026. Datamar News
United States
LA/Long Beach rail dwell averaged 7.19 days in September. The Panama Canal limited transits to 32 a day from September 15, easing to 34 in November. C.H. Robinson, Kuehne+Nagel, gCaptain

The figures in this report were calculated for 320 cases in the US and Latin America, using unit costs and ocean and port conditions at the time of writing, together with measured Tradlinx AI-ETA accuracy. Actual savings may vary with a company’s operating practices and decision timing, its cargo mix and routes, cost levels such as freight rates and labor, and market conditions such as port congestion. These figures do not guarantee results for any specific company.

© 2026 Tradlinx. Tradlinx Insight Report

Leave a Reply

Trending