Cut Out the Noise on Gen AI: Not Moving Forward Strategically and Aggressively Is Your Real Risk of Failure

At a glance

• Noise about "failing" AI pilots is misleading - what's failing is how organizations focus on tactical adoption of low- or no-ROI solutions for proof-point or education purposes instead of developing a true strategic understanding of their high-impact opportunities.
• AI investments and usage are up, but measurable ROI is not. Adoption without impact reflects a flawed approach to breakthrough value-realization opportunities.
• Ideation should prioritize high-impact strategic and productivity opportunities, and pilots should test and validate what to scale.
• Real Gen AI impact comes via reimagining strategies, business processes and organizational capability to unlock new industry- and company-specific competitive advantage in a Gen AI-enabled era.

The recent MIT study on AI pilot programs has sparked numerous questions from our members and prospects; they're asking if headlines about "AI failures" should give them reason to pause their efforts to scale AI's potential impact on their performance. As commentators focus on failure rates, the real story isn't about AI pilots failing to deliver ROI; it's about misunderstandings and missteps in the approach to realizing strategic impact with AI.

Our experience spanning direct project work with global enterprise clients - as well as insights from Innovation Award winners J&J, IBM, McCormick, Roche, Bosch and more - has shown that the enterprises seeing real performance breakthroughs today have a fundamentally different approach to AI.

These organizations are realizing ROI - some exceeding 2500% - with measurable returns within three to 12 months. They do so with an understanding that AI presents a generational opportunity to reimagine performance. These organizations understand that they can now deploy dramatically more valuable automation to infer, reason, learn and act on data and insights to address increasingly complex actions and decisions. They understand that with this capability, they now have exponential intelligence and a new level of workflow productivity at their fingertips. They can leverage more of the enterprise's data and greater human intelligence using large language models further enhanced by domain-specific small language models to improve performance, insights and innovation through agentic AI, with measurable ROI.

These organizations didn't seek to adopt AI to incrementally or tactically improve performance; they set out to reshape their business outcomes and reimagine their processes and organizational capability to deliver new competitive and performance advantages, which resulted in dramatic ROI.

Avoid hammers looking for nails: Adoption without impact

Many organizations approach AI as an enhancement layer and see the goal as making existing processes faster or more efficient. For instance, they use AI-powered chatbots and assistants to improve customer issue resolution rates, or coding assistants to develop boilerplate code or document existing code more quickly. Predictive analytics gets added to existing dashboards, drawing attention to potential risks, concerns or opportunities. These tactical deployments can generate activity, but there's little to show when leaders check top-line growth and operating margins. Usage is up, but returns are limited.

Moreover, despite introducing new AI capabilities, the technology vendor ecosystem reinforces today's flawed approach. "While vendors focus on fast deployment and wide adoption, organizations are discovering the hard (and expensive) way that these ideas are not always aligned with their specific strategic objectives, business processes and performance needs."

A global hospitality executive recently said, "Our lobby is full of technology companies offering to showcase impressive technology capabilities, but they're not bringing us the specifics required to realize measurable value."

Adoption often ignores organization-specific needs and the complexity of scale

The journey from pilot to strategic impact is not as straightforward as many organizations assume. No single AI technology vendor addresses 100% of an organization's actions and decisions in a business process. While these providers have a role, the remaining 20%, 50%, or more of actions and decisions of a business process are often where real performance advantage materializes - from the unique ways the organization acts on insights.

Reaching scale means contending with the state of today's technological infrastructure, a factor that can add considerable complexity. Interoperability protocols, for instance, are still immature, with frameworks like A2A and MCP still evolving. Initially designed for transactional processing, traditional data architectures now need to support real-time inference and learning. And new security models now must address dynamic new threat vectors made possible by autonomous agents.

The good news is that many of today's technology complexities will be tomorrow's afterthoughts, fueled by today's rapid pace of innovation, which continues to eliminate technology complexities as barriers to scaling.

However, focusing on AI adoption can lead organizations to discount or ignore what it takes to scale the organization for new performance advantages. From agent orchestration to interactions between AI systems and human decision-makers, it demands engineering muscle, governance rigor at an enterprise level, and a workforce ready to collaborate with machines. This reality calls for reimagining organizational capabilities and the workforce, outstripping traditional change management - the hidden complexity of scaling AI in an organization.

Today's imperative is industry- and organization-specific ideation

To move beyond these flawed approaches and achieve real impact, organizations must embrace ideation of their industry- and organization-specific strategies, business processes and performance requirements. The Hackett Group® has seen organizations make significant gains in AI-driven performance through ideation. All of these organizations examine outcomes and processes strategically and then explore how AI assists, augments, or - most importantly - can introduce new data sources that result in valuable decision-making capabilities, or even acts autonomously to improve operating execution and performance.

This approach means looking across value chains and pinpointing where AI can eliminate constraints or lower volatility. It calls for understanding processes at the level of individual decisions and information flows, including existing internal and external data sources and technology infrastructure. To succeed here, you must go deep into your formal procedures and the informal workarounds that have evolved and become central to understanding your operational reality.

Ideation takes a team - it is the work of process experts, domain specialists and technology teams working together. Ideation explores the deep operational perspective that can distinguish between activities requiring human judgment and those where a human presence has persisted simply through organizational inertia. It helps identify the specific data needs and technical constraints to address. And it helps identify how impact will be measured through concrete performance indicators and business outcomes.

Platforms like The Hackett Group's AI XPLR™ can streamline this ideation process by rapidly assessing and designing industry- and organization-specific AI solutions and checking feasibility and security concerns before prioritizing the highest-impact scenarios. This moves you from ideas to executable design in days rather than weeks or months.

AI pilots fail, and that's okay

Today's narrative about failing AI pilots misrepresents what pilots are for. Rather than micro-implementations for generating immediate returns, they're better seen as experiments that inform decisions to scale before committing more resources.

That's how leading organizations treat pilots - as extensions of their ideation exercise and as a part of reimagining processes. They use them as instruments to test feasibility within existing architectures. Pilots are used to discover and validate data requirements and establish unit economics. They can also gauge gaps in organizational readiness. Success isn't measured by immediate ROI but by levels of evidence. A pilot succeeds if it answers the question of whether to scale or to stop.

How you move forward matters today

Executive teams today stand at an inflection point. If they continue thinking AI adoption is a matter of accumulating the latest point solutions, then they can expect to enjoy marginal improvements. However, they'll watch competitors use AI to infer, reason, learn and act on data and insights to reshape how they create and deliver value. They'll witness industries transform as competitors use exponential intelligence to innovate at scale and solve problems previously thought too complex or time-consuming.

We've summarized two years of lessons and learnings from our work with organizations and Innovation Award winners into A Readiness Framework for Scaling Gen AI for Performance Advantage. Here are five actions to improve your readiness to move forward with a strategic approach to creating real impact:

• Focus on how you can reimagine your strategies, business and processes, and how your organization creates and delivers value.
• Embrace ideation to examine your needs for how AI assists, augments and acts autonomously in your processes, actions and decision-making.
• Put AI to work for you to accelerate ideation exercises, using solutions like AI XPLR™ to explore the potential impact of AI on your performance.
• Establish a center of excellence to build new technical and organizational capabilities for creating new advantages, rather than depending solely on vendor solutions.
• Treat pilots and POCs as learning investments, and use the insights gained to decide whether to scale or to stop.

We've passed the point where leaders ask themselves whether to invest in AI. Today's questions need to center on readiness to take the steps required to create value. The risks of waiting, of being passive, are too significant.