83% Say Their Workforce Isn't AI-Ready. What Should Ocean Teams Measure?

A new survey from the Association for Supply Chain Management (ASCM) and the Digital Supply Chain Institute (DSCI) suggests that supply chain AI adoption is struggling to move beyond planning and analysis. Released October 14, the report found that 83% of respondents considered their employees unprepared to use AI effectively. Among organizations reporting an unready workforce, 64% remained in the pilot or planning stage.

For an ocean operations team, one useful measure of technology’s value is the time between a disruption becoming knowable and someone acting on the affected containers. Most teams already hold pieces of that timeline in carrier advisories, tracking events, emails and booking records. Few have assembled them into a response-time baseline that could help demonstrate the value of AI or visibility investments.

What the ASCM and DSCI Survey Found

The report, The People Paradox in Supply Chain AI Integration, scored AI use across 13 supply chain functions on a scale from 1 (no AI use) to 5 (autonomous execution). The average was 1.94.

Finding	Figure
Respondents who say employees are unprepared to use AI effectively	83%
Average AI adoption across 13 functions (1 to 5 scale)	1.94
Highest-scoring function: demand planning	2.28
Lowest-scoring function: reverse logistics	1.59
Most-cited barriers (tied): workforce skills and resources; unclear business case or ROI	33% each
Next barrier: poor data quality	25%
Reported project success rate, organizations with an AI-ready workforce	71%
Reported project success rate, unprepared organizations	28%
Unprepared organizations still in pilot or planning	64%

Source: ASCM and DSCI, The People Paradox in Supply Chain AI Integration, released October 14, 2026.

Planning functions score highest, while execution-heavy work trails. Ocean exception handling has many of the characteristics that make execution difficult: a vessel delay or terminal closure can require different decisions for containers on the same service, depending on their location, release status and delivery commitments. The survey does not separately measure ocean exception handling, but its findings raise a relevant question about how much of that work remains manual.

The sample is limited. Of 144 survey responses, 107 qualified for analysis, including 11 from C-suite executives. Respondents came from multiple industries, with 62% based in North America and 13% in Europe. The report describes its job-level findings as directional. Workforce readiness and project success were also self-reported, so the 71% versus 28% gap should be treated as an association rather than evidence that readiness alone caused better outcomes.

Response Time Is the Measure the Survey Points To

Asked how companies should judge AI’s return, ASCM CEO Abe Eshkenazi told FreightWaves that headcount reduction is too simple a measure. He suggested response time: how quickly a company recognizes a disruption, weighs the alternatives and acts. DSCI managing director Marko Kovacevic made a similar argument in the report’s release, identifying rapid detection and replanning as capabilities that can distinguish leading organizations.

Response time breaks into intervals a team can measure. The boundaries need consistent definitions, particularly when carrier events and internal records use different clocks:

T0, knowable: the earliest verifiable signal available through a source the team could reasonably access, such as a carrier advisory, terminal notice or container event. The underlying disruption may have begun earlier.
T1, recognized: the first record showing that the team identified the disruption, such as an acknowledged alert or logged exception. Email delivery alone does not prove recognition.
T2, decided: a documented commitment to a response, such as rebooking, retaining a pickup or holding a truck.
T3, executed: confirmation that the instruction was accepted or the change became effective in the carrier’s, terminal’s or trucker’s system.

T0 to T1 measures recognition, T1 to T2 covers evaluating alternatives and deciding, and T2 to T3 measures execution. Visibility tooling can shorten recognition and help identify which containers require attention during the decision interval. Its contribution depends on the sources it covers and how quickly it receives updates.

A shorter response time becomes an ROI argument when the earlier action improves an outcome: avoiding a failed pickup, preventing an avoidable free-time overrun or giving a consignee enough notice to adjust a delivery plan. The decision still needs an owner. As Eshkenazi told FreightWaves, “Ultimately, somebody has to own that decision for the organization.”

Worked Example: IT Recovery and Two Storms at North European Ports

The October 23–25 disruption window in North Europe illustrates how quickly a general alert becomes complicated at container level. Storm Austen affected France and the Netherlands on Friday, October 23, while Dutch weather agency KNMI warned that a second system, Boelo, would bring renewed strong winds from late Saturday into Sunday.

At Le Havre, the situation was already complicated before the storm. An IT incident had disrupted operations at Terminaux de Normandie (TNMSC) since October 13. Kuehne+Nagel’s October 20 update reported that container gate-in and gate-out movements remained unavailable, with no confirmed return to normal operations. By October 22, a notice shared by UMEP Le Havre described a progressive restart of maritime, rail and inland-waterway activities, with container movements gradually returning. An exceptional Saturday morning opening was planned, conditional on the weather.

Kuehne+Nagel’s separate Storm Austen alert on October 23 set out the following terminal schedules. These were expected operating windows at the time of publication, not confirmed records of what ultimately happened.

Terminal	Scheduled status in the October 23 alert
Le Havre, Normandy Terminals (TNMSC)	No handling Friday, October 23. Expected Saturday reopening between 06:30 and 12:00, weather permitting. The terminal was also recovering from its earlier IT disruption.
Le Havre, Terminal de France (GMP)	Closed Friday after limited operations. No vessel operations scheduled 06:00–22:00. Saturday morning operations under consideration, not confirmed.
APM Terminals Rotterdam, waterside	Expected suspension Friday 20:00–Saturday 09:00. Planned Saturday operations 09:00–18:00 with delays anticipated. Another suspension expected Saturday 18:00–Sunday 23:00.
APM Terminals Rotterdam, landside	Expected suspension Friday 22:00–Saturday 06:00 and Saturday 21:00–Sunday 04:00.

Source: Kuehne+Nagel, October 23, 2026. Times are local as published. Actual closures, reopenings and available services depended on weather and terminal conditions.

Recognize

For TNMSC cargo, the first operational warning preceded the storm by more than a week. A team that first learned of the terminal’s problems through Friday’s weather advisory would have missed earlier opportunities to adjust bookings, truck arrangements or customer expectations. The October 22 recovery notice introduced another decision point: could affected containers move during the proposed Saturday opening, or would weather restrictions prevent it?

That makes T0 more demanding than simply recording when an email arrived. The team needs to distinguish the first accessible warning of an IT disruption, subsequent recovery updates and a new weather restriction. It also needs to establish when each container became identifiable as affected. Those timestamps may differ.

Weigh and Act

The Rotterdam schedule shows why the next decision cannot rest on a single port-wide status. In the October 23 forecast, landside operations were expected to be available from 06:00 to 21:00 on Saturday, while waterside handling was planned from 09:00 to 18:00. On Sunday, landside reopening was expected at 04:00, while waterside operations were scheduled to remain suspended until 23:00.

Those differences could matter to containers already discharged and released. A team reacting only to a general “Rotterdam closed” warning might cancel pickups unnecessarily. But an expected landside opening does not guarantee that a particular container can be collected. Gate availability, release status, customs clearance and truck appointments still need confirmation.

Containers still on board require different action: checking revised discharge estimates and updating delivery commitments. At Le Havre, containers affected by the earlier IT incident may have accumulated delays before the storm. Sorting released containers in a yard from those still awaiting discharge, handling or system clearance is the work that determines which actions are available.

Execution means confirming revised truck appointments, accepted booking changes and updated customer ETAs. It also means checking how the interruption affects applicable free-time terms. Weather-related closures do not automatically suspend demurrage, detention or terminal storage charges. The treatment depends on the applicable carrier terms, contract, local rules and the charge involved.

There is a further timestamp complication this weekend. On Sunday, October 25, the Netherlands moves its clocks back one hour at 03:00. The expected Rotterdam landside suspension from Saturday 21:00 to Sunday 04:00 therefore spans eight elapsed hours, even though the local clock times appear seven hours apart. Teams calculating response intervals should normalize timestamps to UTC and retain the original source times.

If your team is piecing together terminal schedules and container status from separate forwarder emails, see how ops teams set up container-level exception alerts on the lanes they choose.

How to Time One Lane

Eshkenazi recommends starting with a high-value process, involving the people who run it and establishing a baseline before setting improvement targets. For an ocean team, one frequently used lane provides a manageable starting point.

Pick one lane with steady volume. Choose an origin and destination pair your team books regularly. Start with eight weeks of history, then extend the period if too few measurable exceptions occurred.
List the exceptions you already handle. Rollovers, blank sailings, transshipment changes, terminal closures and ETA changes beyond your tolerance. A probability-based ETA can help estimate the risk of exceeding an agreed delivery threshold.
Pull T0 to T3 for each exception. Use the earliest verifiable accessible advisory or container event for T0, an acknowledged alert or exception record for T1, and documented decisions and execution confirmations for T2 and T3. Mark missing timestamps rather than estimating them. Normalize the records to one time standard.
Report the median and the long delays. Calculate the median and 90th-percentile duration for each interval, alongside the number of measurable exceptions. A median shows typical performance; the longer delays may reveal the highest operational exposure. Record missed deadlines and avoidable costs separately.
Mark decisions that already follow fixed parameters. Eshkenazi sees room for AI in routine transactional decisions within set limits, with people retaining responsibility for higher-risk calls. On an ocean lane, candidates include notifying a consignee when an ETA crosses an agreed threshold or flagging containers approaching their last free day.

Repeat the measurement after a process or tool change. If recognition time falls while the decision interval remains unchanged, examine ownership, available alternatives and approval requirements. If both intervals improve but avoidable costs remain unchanged, investigate whether the alerts arrive early enough to make a useful intervention possible.

The survey identifies workforce readiness and the business case as major obstacles to AI adoption. Measuring one lane gives a team a way to investigate both: whether the people handling exceptions can use the information effectively, and whether faster decisions produce measurable operational value.

Before Your Next AI or Visibility Review

Do you have a median recognition time, measured from a consistent T0, for at least one lane?
On the last terminal closure that touched your cargo, which source first gave your team an actionable warning?
For North Europe cargo affected by the October disruptions, how many containers were on the ground and released, how many were awaiting handling, and how many remained on vessels?
Which lane decisions already follow fixed rules and could be automated first?
Can you connect shorter response times to avoided costs or improved delivery performance?
When a tool flags an exception, who owns the decision?