Most AI pilots in transportation and supply chain fail because teams pick the AI project before confirming there's a business problem worth solving. Gartner reports that AI already accounts for 67% of supply chain digital investment, while 55% of surveyed chief supply chain officers say they're unclear on what return that investment is generating (Supply Chain 247, citing Gartner). JBF Consulting's own 2026 survey of 215 supply chain leaders found the same pattern up close: 78% are pursuing AI without a documented plan, and only 3% check results against the business case that justified the investment in the first place.
That gap shows up in familiar patterns. A team builds and trains an elaborate AI module to generate and summarize a complex report. They cheer their success, but the report never gets used for making faster strategic decisions, as intended – they never engaged the correct stakeholders, who missed the window to provide key requirements and generate training for the end users.
Another team takes the latest AI and points it at an order dataset to hunt for outliers. The project is deemed a success, as the error rate is virtually zero, but they were unaware that an existing upstream rules engine already catches 99.98% of the defective data and automatically corrects the errant records before it ever reaches the in-scope system, let alone is visible by an associate.
Both projects "work". But neither creates meaningful value.
These are both cases where AI was deployed in search of a problem, not situations where a problem was identified and now in search of the right solution. Firms that are actually generating returns today are taking a different approach: they are spending significant effort upfront deciding where limited resources should go to support outcomes that matter to their business rather than on executing unvetted AI projects.
The gap is about which projects get funded, and whether anyone built a business case before AI was chosen as the solution.
How to Find Where AI Actually Adds Value
The better starting point is nothing new: identify the pain and follow it to uncover opportunity. Where is your team burning the most time? Where are defects concentrated? Where are people making high-stakes decisions with unreliable information? Where does a process require sifting through massive amounts of data that even the best associate cannot absorb anything meaningful from it?
This is precisely where the value lives, and it often takes a talented team just to isolate and then qualify these opportunities. "Qualifying" includes identifying and then prioritizing the opportunities, ideally based on expected value and ease-of-implementation. While considering how to solve the pain-point, all solutions should be considered at least at a high level, before locking in the preferred path. Importantly, a painful process is not automatically an AI project.
That discipline is rarer than it should be. JBF's 2026 survey found that 85% of supply chain organizations don't use a consistent, formal process for selecting AI opportunities. Developing or validating the business case was the single most-cited barrier to progress, named by 67% of respondents.
That is not to say that AI won't be the ideal solution. As the technology progresses, AI is more often becoming the better method of solving business problems. Due diligence still needs to be done to ensure that simpler, more consistent approaches are considered, as well.
Another way to look at the value equation is to consider the basics independent of the solution and evaluate them under the lens of business needs and objectives:
- Is there sufficient value here? A painful process that's genuinely low-stakes, low-frequency, or already well-contained isn't worth the investment, however satisfying it would be to automate with or without AI.
- What's the opportunity cost? An hour spent on one initiative is an hour not spent on another. If a competing project carries materially more value, AI capability doesn't change that math.
This requires discipline executed by a knowledgeable individual or team that can evaluate multiple opportunities objectively, while still understanding the company's capabilities, including change management, governance, partnerships, and long-term objectives. Those resources, be they internal or external, can rank opportunities by net value first, then ask whether AI is not just a viable answer, but the best one available.
Firms generating returns right now are the ones deciding, ahead of time, where limited resources should go to support outcomes that matter to their business.
Where AI Now Fits That Didn't Before
Once that discipline is in place, it's time to look beyond the existing backlog of projects and continuous improvement tasks. Many firms are not even considering the new types of opportunities available based on AI capabilities.
Processes shelved years ago that were deemed too complex, too unstructured, or too judgment-dependent to automate, may now be squarely in scope. Unstructured data, ambiguous decision criteria, and pattern recognition across messy datasets are exactly where current AI capabilities are thriving – and many of these projects would have previously failed the value-equation when compared to other potential projects.
None of those fixes replace human judgment, but they do change what people will notice and be able react to, so that associates can intervene earlier while leveraging improved information.
AI can be genuinely valuable, but only if you apply the same rigor: find the pain-point, quantify the value, and confirm AI is the best-fit tool rather than the default one. Following that equation keeps effort from going to low-impact solutions, regardless of the people and technology deployed. Expanding the search to forgotten, ignored, and new areas keeps high-impact opportunities from being missed simply because they didn't make sense using the old cost-benefit model.
What's Missing
AI is not replacing primary skills and resources anytime soon. Enterprise shippers often lack the bandwidth to define and design core aspects of a given initiative:
- KPIs and success criteria
- RACI roles & responsibilities
- Governance structure
- Clear business objectives
- Change management plan
All projects, including AI ones, need those foundational components to function properly and stand a real chance of success. Feedback loops, data & application access limits, escalation paths, model oversight – the elements required to keep an AI process accountable – take real expertise to define properly, and many internal teams are either already stretched thin running the business or lack the appropriate training.
JBF's own survey backs this up directly: only 11% of supply chain organizations report having a formal, actively maintained governance framework for AI performance after launch, and 22% say accountability for a live AI initiative varies by initiative rather than sitting with a defined owner.
Taking a higher-level look at today's business environment, people are still a critical component to successful AI deployment, but there are often very few employees dedicated to supporting the very projects that typically deploy AI as a solution. Associates that would be prioritizing projects, driving change management, refining & defining new policies, and training team members on new processes are being eliminated based on the very technology they should be cultivating.
Research on AI business adoption backs this up, according to Gartner: roughly 80% of organizations piloting autonomous AI have already cut headcount, but those cuts show no correlation with actual ROI (Gartner newsroom). The firms seeing real returns invest more, not less, in the skills and structures needed to guide the systems they've deployed. An AI model without that support structure carries an unmanaged risk with an uncertain return on investment, whatever it achieved in testing.
Mitigation and Next Steps
Value identification, prioritization, and governance design are not tasks most internal teams do well without support. The cost of getting it wrong is a shelf full of AI projects that never earn their keep, failing to generate meaningful value while diverting resources from more promising projects.
One of the most reliable methods to prevent such mistakes is engaging an objective partner well before those projects are defined – one with no incentive to sell a specific AI product, whose job is to help identify and prioritize the opportunities worth pursuing, and has the ability to apply an appropriate governance model that ensures value is achieved using the best-fit solution. This is the discipline behind JBF's four-phase approach: Strategic Design, Orchestration, Delivery, and Continuous Improvement, with each phase building the accountability structure that JBF's own research shows most organizations still lack. Executing that process well means every initiative uses the best-fit solution that delivers measurable outcomes that hold up well into the future — be it AI or a more traditional approach.
The results show a clear gap between AI activity and organizational readiness.
About the Author
Chris Doersen is a Principal of Client Engagement at JBF Consulting, bringing more than 20 years of experience designing, modeling, and implementing logistics solutions for complex transportation networks. He partners with global shippers to drive efficiency, optimize fleet and carrier operations, and enable technology adoption that delivers measurable impact. Chris has deep expertise with platforms including Blue Yonder, Descartes, Llamasoft, and Appian, and has led large-scale network design and TMS implementations for leading manufacturers and distributors. Known for his analytical rigor and operational insight, Chris helps organizations turn transportation strategy into sustained results.
FAQs
Most failed AI pilots trace back to teams choosing an AI project before confirming a business problem worth solving. Gartner reports that supply chain organizations already put 67% of digital investment into AI, and 55% of chief supply chain officers say they can't identify the return that spending produces. JBF Consulting's own 2026 survey of 215 supply chain leaders found the same root cause directly: 78% are pursuing AI without a documented plan, and only 3% check results against the business case that justified the investment.
Rank the opportunity by expected value and the cost to implement it before checking whether an AI tool exists for it. Ask two questions: how much value is at stake, and what else that time could fund. JBF's 2026 survey found that 85% of supply chain organizations don't use a consistent, formal process for selecting AI opportunities, which is exactly the gap this ranking discipline closes.
Gartner's research found that roughly 80% of organizations piloting autonomous AI have already reduced headcount in that area, with those cuts showing no link to actual return on investment. Firms seeing measurable returns tend to add headcount in the roles that define governance, KPIs, and change management around AI, even as competitors cut those same roles.
Gartner's research found that roughly 80% of organizations piloting autonomous AI have already reduced headcount in that area, with those cuts showing no link to actual return on investment. Firms seeing measurable returns tend to add headcount in the roles that define governance, KPIs, and change management around AI, even as competitors cut those same roles.
A painful process qualifies as an AI project only after it clears three checks: enough value at stake to justify the resource, an opportunity cost that favors it over competing initiatives, and confirmation that AI is the best available method for solving it. Simpler, more consistent approaches deserve the same evaluation.
An objective partner with no product to sell works best for this step, someone whose job is to size and prioritize opportunities across the business and then apply governance once a project starts. JBF's 2026 survey found that only 11.4% of supply chain organizations report having the strategy, governance, and capabilities needed to scale AI successfully on their own, which is the gap an outside partner is built to close.
It's JBF Consulting's term for the distance between how fast organizations adopt AI and how well they've built the structures needed to prove it's working. JBF's 2026 survey of 215 supply chain leaders found the gap across the AI lifecycle: 78% pursue AI without a documented plan, 40% remain stuck at prototype or pilot, and only 10.5% have a formal governance framework for AI once it goes live.
