Companies continue to invest heavily in supply chain software, automation and AI, yet many projects miss expected returns, budgets or timelines. New JBF Consulting research suggests the biggest barriers are often governance, data quality and adoption rather than technology capability itself.
Strategy Before Software: Rebuilding the Front End of Tech Decisions
Billions continue to flow into transportation, warehouse, planning, and AI platforms, yet the payoff remains patchy. JBF Consulting’s recent survey found that roughly 89% of supply chain technology implementations captured less than 76% of their projected return, and a similar share missed expectations on time, budget, or core outcomes. The central problem was not software capability but the way decisions were sequenced and governed.
Many organizations begin with a category label rather than an operational hypothesis. The internal brief is to buy a TMS, WMS, or AI engine instead of defining in detail which service, cost, resilience, or working capital problems need to be solved. That approach leads selection teams toward high-profile platforms that dominate analyst quadrants, even if the feature set and complexity do not match the real constraints of the network.
Consultants like JBF are being pulled in earlier to challenge that pattern. The most effective programs now start with an explicit articulation of scenarios, constraints, and strategic priorities: the role of freight in margin protection, targeted cycle time reductions, resilience thresholds, and ESG reporting duties. From there, requirements are broken down by process, role, and data need, and only then do teams test vendors.
When this front-end work is skipped, misalignment shows up late. Systems arrive that cannot support the actual orchestration logic, demand workarounds to mirror the physical network, or introduce latency between planning, execution, and finance. Organizations then face either incremental customizations that erode maintainability or expensive rip-and-replace cycles that restart the ROI clock.
Governance also falters once vendor choice is made. Projects often migrate into pure IT ownership, with operational teams reduced to periodic workshops or user acceptance tests. That structure accelerates technical deployment but disconnects configuration from day-to-day decision patterns, leaving the application technically live but strategically underused.
Change, Data, and AI: The Underfunded Foundations
The JBF research highlights that organizational design and change management remain the most frequent blind spots. Project budgets emphasize licenses, integrator fees, and infrastructure while underfunding program management, role-based training, and adoption support. Employees are expected to contribute to design, testing, and rollout while still managing full operational loads, which slows decisions and weakens ownership.
Change programs are often first on the chopping block when budgets tighten. That decision has predictable consequences: users learn navigation but not how to reframe daily work to exploit the new tools. Vendors typically deliver generic training on screens and menus. What is missing are role-specific playbooks that explain how a planner should use new optimization capabilities, how procurement should interpret new risk signals, or how logistics teams should adjust exception handling.
Industry reports around warehouse automation and AI adoption show a similar pattern. Organizations invest in robotics, computer vision, and algorithmic planning, but utilization rates lag design assumptions because process redesign and workforce transition plans are incomplete. In some networks, expensive automated capacity is manually bypassed during peak disruptions because teams do not trust the data or do not understand how to recover from exception states.
Data quality compounds the problem as AI enters the stack. JBF emphasizes that AI value is gated by governance and master data. Poor item hierarchies, inconsistent location codes, or incomplete carrier performance records feed models that produce weak recommendations. Recent trade analyses of digital supply chain programs show that organizations with mature data stewardship and standardization unlock markedly higher benefits from the same AI tools than peers with fragmented data estates.
The pressure to ‘deploy AI’ quickly can push teams into pilots that run outside core workflows. Dashboards and copilots spring up around the edges of the business, but they do not influence replenishment parameters, routing logic, order promising, or capital allocation. AI remains an experiment rather than a production asset.
The more advanced adopters take a different route. They treat AI as an embedded decision layer rather than a standalone platform. Use cases are tied to specific process outcomes: improved forecast quality in defined categories, faster disruption response for key lanes, or tighter alignment between inventory and working capital targets. Governance councils set thresholds for machine-generated decisions, escalation rules, and auditability so that algorithms reinforce, rather than undermine, control.
Technology Success Depends On Organizational Capacity
As software capabilities become more standardized across vendors, implementation outcomes increasingly depend on an organization’s ability to make decisions, manage change and maintain data discipline. Companies that build these capabilities before major deployments are often able to extract value faster, while those that underestimate them can find themselves owning sophisticated platforms that never become embedded in daily decision-making.
