Between 80% and 95% of organizations report limited financial value from their AI investments, and only 11% of CFOs have been able to concretely measure return on investment from AI, according to Gartner’s 2025 and 2026 data. At the Gartner Data & Analytics Summit in Orlando, Gartner analyst Frances Karamouzis presented a FinOps framework for vetting, prioritizing, and funding artificial intelligence use cases. The framework includes feasibility scoring, business value calculations, and manufacturing-relevant KPIs covering plant utilization, operational downtime, throughput, and supply chain velocity.
Why Do Most AI Investments Fail to Show Financial Return?
A Gartner poll of 1,229 respondents in October 2025 identified use case selection, value assessment, and data strategy as the top three reasons organizations fail to achieve financial value from AI. Respondents ranked selection of the wrong use cases highest, followed by value assessment, data strategy and approach, business function alignment, financial rigor, time horizon and speed, with technology challenges at the bottom.
For manufacturing leaders evaluating AI spending, the ranking points to where to focus. The three most common failure causes all sit with business and data teams. AI investment decisions succeed when they start as business cases tied to specific operational outcomes.
The Gartner framework applies a funding funnel that moves AI initiatives through four stages: ideas, vetting, prioritizing, and funding, with results tracked at the end. Each stage applies a litmus test that maps proposed use cases to revenue impact, cost-efficiency, or risk mitigation before any funding is approved.
How Should Manufacturers Score AI Use Case Feasibility?
According to the session, Gartner’s feasibility scoring framework evaluates AI use cases across six weighted criteria totaling 100 points. The key categories include AI-ready data (hygiene, metadata quality, access), technology maturity and vendor ecosystem, trust and security posture, and skills and implementation capability. Organizations plot each potential use case on a business value versus feasibility matrix to separate investments worth prioritizing from those that require more groundwork.
In the framework’s generic use case assessment, asset predictive maintenance scored as both high business value and high feasibility, placing it in the top-right quadrant that organizations should fund first. Other use cases like autonomous business processes scored high on potential value but lower on implementation readiness.
Any criterion that falls below a defined threshold triggers a structured review: capture the reasons, brief vendor managers, review with the internal team, and provide findings to legal and contract negotiation teams. The system surfaces risk before investment proceeds.
What KPIs Prove AI Value in Manufacturing?
AI value in manufacturing maps to three categories presented at the summit: revenue impact (units sold, growth rate, market share), cost-efficiency (cycle time, throughput, supply chain velocity, plant utilization), and risk mitigation (defect rates, quality compliance, warranty claims, safety metrics). Each requires KPIs that connect AI outputs to financial outcomes. The framework also defines three levels at which organizations can prove financial value, from hard-dollar direct ROI to strategic trade-off visibility.
The three levels of provable financial value determine how manufacturing leaders should build their investment cases. Level one requires a direct, one-to-one cause-and-effect relationship between the AI investment and a measurable financial outcome. Level two builds a credible correlation between operational metrics and business results based on statistical association. Level three creates visibility into business trade-off positions through strategic scenario analysis.
For manufacturing operations, level-one proof is achievable when AI directly improves a tracked metric like plant utilization or operational downtime. Level-two proof applies when AI improves upstream data quality that correlates with downstream production gains. Level-three proof covers strategic scenarios where AI provides decision intelligence that changes capacity planning or capital allocation.
What Does Successful AI ROI Look Like in Production?
A case study presented at the summit described a multinational home improvement goods manufacturer that targeted 10% annual revenue growth while maintaining balanced inventory and production levels. The challenge: frequent product launches and promotions caused volatile demand, production swings, and inventory surges. Quarterly planning cycles and linear forecasting missed feedback loops like the bullwhip effect, leading to supply chain inefficiencies. Leadership needed a long-term view of how policy decisions would affect capacity, capital, and service levels.
The results, as presented by Karamouzis: service levels exceeded 95%, supporting 9% annual revenue growth. Inventory variance dropped 40%, freeing $120M in working capital. Plant utilization rose from 72% to 85%. Deferred plant expansions saved $45M in capital expenditure.
These results represent level-one provable value: direct, measurable financial returns tied to specific operational improvements. Plant utilization gains alone justified the investment by deferring a $45M expansion. The $120M freed from inventory variance reduction went directly to the balance sheet. Service levels above 95% sustained the revenue growth target, demonstrating that AI-driven production and supply chain improvements can deliver returns across multiple financial dimensions simultaneously.
FAQ
1. Why do most AI investments fail to deliver financial value?
A Gartner poll of 1,229 respondents found that the top reasons are selection of the wrong use cases, inadequate value assessment, and poor data strategy, in that order. Technology challenges ranked last. Between 80% and 95% of organizations report limited financial value from AI, and only 11% of CFOs can concretely measure return on investment. The failure pattern sits with business and data decisions, with technology challenges ranking at the bottom.
2. How do you measure AI ROI in manufacturing?
Gartner’s framework defines three levels of provable financial value: direct cause-and-effect ROI, correlation between operational and business metrics, and trade-off visibility for strategic decisions. Manufacturing KPIs include plant utilization, operational downtime, throughput, supply chain velocity, defect rates, and safety metrics. Organizations should identify the top three to five value-impacting areas and connect AI outputs to measurable improvements in those areas.
3. What is a feasibility score for AI use cases?
A feasibility score evaluates AI use case readiness across six weighted criteria totaling 100 points. Key categories include AI-ready data quality and access, technology maturity and vendor ecosystem, and trust, security, and risk management. Organizations plot use cases on a business value versus feasibility matrix to prioritize investments. Any criterion scoring below a defined threshold triggers a structured risk review before funding proceeds.
4. What AI use case has the highest measured ROI in manufacturing?
Asset predictive maintenance scores as both high business value and high feasibility in Gartner’s use case assessment framework. In a separate case study, a multinational manufacturer achieved plant utilization improvements from 72% to 85%, a 40% reduction in inventory variance freeing $120M in working capital, and $45M in deferred capital expansions through AI-driven supply chain and production improvements.
This article is based on a presentation by Frances Karamouzis, Gartner analyst, at the Gartner Data & Analytics Summit in Orlando (2026). Lucian Fogoros of IIoT World attended the event. AI tools were used to help summarize and organize the content. Reviewed and edited by the IIoT World editorial team.
