Who Pulls the Plug on Failing AI Projects? A C-Suite Guide

Artificial intelligence (AI) initiatives, while promising transformative benefits, don’t always deliver. Determining when to terminate a failing AI project and, crucially, who should make that call, is a growing challenge for organizations. The decision isn’t straightforward, as different C-level executives often view “failure” through distinct lenses. This article explores the perspectives of the CIO, CFO, and CEO, and outlines a framework for making informed decisions about AI project termination.

Defining AI Project Failure

Dovi Geretz, CTO at SlickTrip, emphasizes that defining AI failure hinges on several key factors: scalability, reliability, data quality, and security.¹ However, these technical considerations often differ from the priorities of other C-suite leaders. Chief Financial Officers (CFOs), for example, typically assess failure based on financial metrics – missed return on investment (ROI) targets, rising costs, or a lack of clear economic value.¹ Meanwhile, the Chief Executive Officer (CEO) tends to evaluate failure in terms of strategic impact, such as whether the initiative supports business transformation or competitive differentiation.¹

The Shifting Balance of Power

The influence over the “kill switch” for an AI project isn’t static. it shifts depending on the reason for the failure. If the issue stems from technical feasibility, data readiness, or integration challenges, the Chief Information Officer (CIO) usually takes the lead in the decision-making process.¹ Conversely, if costs escalate without a corresponding ROI, the CFO’s influence grows.¹ However, the CEO retains ultimate authority, particularly when the project is tied to long-term strategy, brand impact, or competitive positioning.¹

Shared Accountability and Predefined Metrics

Experts agree that a collaborative approach is crucial. Steeve Lavoie, CTO at Allied Scientific Pro, notes that the CFO often has the most influence due to control over funding.¹ However, Geretz advocates for a joint decision-making process. He suggests that each AI project should have predefined success metrics, stage gates, and clear kill criteria agreed upon by IT, finance, and the business units involved.¹ When these criteria aren’t met, a technical assessment led by the CIO, a financial impact assessment by the CFO, and a strategic evaluation by the CEO should collectively inform the decision.

Preventing Failure Through Upfront Alignment

Greg Fletcher, CTO at Ocula Technologies, stresses the importance of proactive planning. He recommends defining tangible checkpoints before launching an AI initiative, including internal adoption rates, accuracy thresholds, and cost benchmarks.¹ This structured approach minimizes political friction and ensures decisions are based on objective data. Misaligned expectations are a primary cause of delays in AI projects, so leadership must share a common understanding of the tool’s capabilities and limitations.¹ Regular progress reviews should focus on comparing results against these shared benchmarks.

The Value of Learning from Failure

Even unsuccessful AI projects can yield valuable insights. Ashish Verma, chief data and analytics officer at Deloitte, points out that failures offer data and experience to inform future strategies.¹ Organizations shouldn’t shy away from ambitious AI bets, even if they carry a risk of failure, recognizing that testing and learning are fundamental to innovation.¹

Key Takeaways

• Define Failure Upfront: Establish clear success metrics, stage gates, and kill criteria before project initiation.
• Shared Accountability: Involve IT, finance, and business leaders in the decision-making process.
• Shifted Influence: Recognize that the decision-making power shifts based on the nature of the failure (technical, financial, or strategic).
• Embrace Learning: View failures as opportunities to gather valuable data and refine future AI strategies.

As AI continues to evolve, organizations must develop robust frameworks for evaluating and terminating failing projects. A collaborative, data-driven approach, with clear accountability and a willingness to learn from setbacks, will be essential for maximizing the return on AI investments.

¹ https://www.informationweek.com/it-leadership/should-the-cio-cfo-or-ceo-hold-the-kill-switch-on-ai-