Teams Are Struggling to Scale AI Across Operations
Most companies say AI matters. Few can prove it. New findings from Coupa's Clarity AI Impact Report show why procurement and supply chain teams are stalling: misaligned expectations, low executive fluency, weak data foundations, and scattered tools.
Here's what matters for operations leaders who need outcomes, not pilots.
The execution gap: strong intent, weak strategy
While 86% of companies see AI as essential, only 29% have a clear, company-wide strategy. That's the first choke point. Teams are executing point solutions without a shared plan, measurable outcomes, or governance.
Result: projects stall or never finish. Budgets get burned. Confidence drops.
Fluency mismatch is killing momentum
Only 5% of executive decision-makers use AI daily, compared to 57% of technical teams. The people funding multi-million budgets often lack the hands-on understanding to set realistic scope, timelines, and constraints.
This gap creates a strategic vacuum: leaders push for results while teams wrestle with integration, training, and process change.
Workforce readiness is the bottleneck
Only 21% of organizations say they have the skills to use AI effectively, and 69% report limited skills and training are slowing adoption. For supply chains that depend on cross-functional coordination, this is a core risk.
Without trained planners, buyers, and logistics teams, even the best models won't lift forecast accuracy, inventory turns, or OTIF.
"Pilot purgatory," unrealistic ROI, and old systems
72% of AI initiatives never make it past the pilot. Despite that, 47% of executives still expect 6-12 month payback. That disconnect fuels rushed launches, thin change management, and poor adoption.
Data is the blocker cited most: 77% point to data quality and system integration as the main barrier. If warehouse, ERP, TMS, and supplier data don't align, you won't get end-to-end visibility or trusted recommendations.
What the market is shifting toward
Organizations are consolidating on unified platforms: 80% prefer to buy AI through external platforms rather than build in-house. But only 2% of AI investment is going to orchestration-exactly what's needed to move beyond isolated tools. Meanwhile, 77% of leaders still focus on simple task automation.
Governance is lagging. 65% prefer human-in-the-loop oversight, yet 56% aren't sure if a formal AI governance policy exists. That human oversight can become a bottleneck if the process and thresholds aren't defined.
"The days of funding AI based on unproven potential are over," says Dennis Bruder, Chief Product Officer of AI at Coupa. "Execution now depends on platform selection."
Playbook: how operations leaders get AI out of pilot and into P&L
- Set business-first outcomes. Tie each initiative to 1-3 KPIs (forecast accuracy, inventory turns, OTIF, working capital, supplier risk). Kill or scale based on impact, not demos.
- Pick a unified platform. Favor solutions with built-in data models, connectors, workflow, and policy controls. Reduce the integration tax and speed time-to-value.
- Invest in orchestration, not just automation. Connect data, decisions, and actions across S&OP, procurement, logistics, and finance. Point tools won't compound value.
- Fix data at the source. Prioritize master data standards, supplier data quality, and event streams from WMS/TMS/ERP. Automate data hygiene where possible.
- Right-size ROI timelines. Expect quick wins in weeks (task automation, assistive insights), with compounding gains over 12-24 months (planning accuracy, inventory reductions).
- Formalize governance. Define model risk tiers, approval thresholds, audit trails, and fallbacks. Make "human-in-the-loop" a clear process, not a vague principle. For a reference point, see the NIST AI RMF framework.
- Upskill the frontline. Train buyers, planners, and analysts on prompts, exceptions, and decision review. Adoption lives or dies with end users.
- Shift to product thinking. Treat AI use cases as products with roadmaps, owners, user feedback, and release cycles. Avoid one-off "projects."
- Make value tracking non-negotiable. Baseline, instrument, and review monthly. Publish results so funding follows impact.
A simple sequencing model
- Phase 1 (0-90 days): Clean priority data objects, deploy assistive AI in two high-volume workflows, set baseline KPIs, launch training.
- Phase 2 (3-9 months): Introduce decision recommendations with guardrails, expand to adjacent teams, embed governance reviews, start vendor consolidation.
- Phase 3 (9-18 months): Orchestrate cross-functional flows (plan-to-procure-to-pay, demand-to-fulfillment), automate low-risk decisions, renegotiate contracts based on realized value.
What to ask your team this week
- Which two AI use cases will move a KPI within 90 days, and what's the baseline?
- Where exactly is data quality blocking decisions, and who is accountable for fixing it?
- What platform choice reduces our integration and governance overhead by half?
- What skills are missing on the frontline, and how do we close that gap this quarter?
Want the source?
Coupa's latest analysis details these gaps and trends. Learn more at Coupa.
Level up team capability
If skills are the blocker, close it fast. Explore role-based AI training for operations, procurement, and supply chain teams at Complete AI Training.
Your membership also unlocks:
