If your business hasn’t yet implemented AI throughout the organization, enabling reduced costs, greater efficiency and better decision-making, you’re likely feeling behind. The truth is that many businesses are trying, but few are achieving the success they imagined with AI. The problems they’re encountering aren’t typically due to limitations of the technology.  Instead, the lack of success often comes down to multiple issues within two broad categories: an inadequate AI strategy and a mismanagement of change. 

Alongside the AI hype, there’ve also been numerous articles published about companies failing to see ROI with their AI initiatives, the most viral of which are a Gartner study predicting that 40% of AI projects will be abandoned by 2027 and an MIT Report claiming that 95% of AI pilots are failing. With numbers like these, many business leaders may be inclined to take a wait-and-see approach, despite feeling mounting pressure from the ubiquitous AI-related headlines and not wanting to lose strategic advantage. 

The question isn’t whether significant ROI value can be achieved, but how.   

In partnering with numerous clients across the life sciences, CPG, and retail industries, we’ve seen the full spectrum of success, ranging from not knowing where to start, to failed investment, to game-changing business value. Among companies that are struggling, there are several common patterns related to AI strategy and change management. 

Inadequate or Misaligned AI Strategy 

Proper AI strategies go well beyond “we want to improve our productivity and the quality of our decisions with AI.” In practice, an AI strategy is a multi-part approach with interdependent components that include system architecture, organizational design, governance, targeted high-value use cases, and more. Many businesses fall short of checking all these boxes. 

Targeted use cases: Aiming for enterprise-wide AI integration is almost always the wrong approach, but it’s common. Businesses should instead focus on a small number of high-value use cases that drive real business results. For example, a CPG company that relies on manual demand forecasting and inventory management in Excel could benefit from AI that detects disruptions in their supply chain or synthesizes consumer and economic demand signals to automate demand forecast updates. We often suggest identifying these through a Business Value Workshop, finding the greatest pain points, inefficiencies, and growth opportunities with high visibility and that are supported by data and talent readiness. 

System architecture: Despite a common, albeit frustrating perspective, AI isn’t magic. Success requires quality underlying system and data infrastructures. AI does not fix flaws in a company’s data architecture—it magnifies them. 

Governance: Neither use case implementations nor system architectures are one-and-done. They require ongoing monitoring, maintenance, calibration, and security assurance. To govern data and AI systems properly, clear roles and responsibilities must be enforced through robust policies and procedures that ensure master data management, tool testing and validation, security, and issue escalation and backup procedures. 

Organizational Design: Should data architectures and AI systems be centralized for the entire company, or should they be managed by each of the business’ functional areas? There’s no one-size-fits-all solution to organizational design, but success with AI will be difficult to reach if the right data, technology, and talent are not distributed in a manner that aligns with the company’s culture and talent. 

Talent: Even if a business has identified high-value use cases, achieved a gold-standard data and AI infrastructure, and outlined comprehensive governance policies, they will still fall short of success with AI if they don’t have the talent in place to maintain all of it and to effectively use AI products. Employees require not only ongoing training to be able to be able to effectively use AI tools, but companies must also acquire the talent to deploy and update them through either training or hiring. 

Unique Difficulties of Managing Change Surrounding AI 

The other broad category where businesses fall short on their pathway to AI success is change management. While many of the traditional change management best practices apply to AI integration, the overall approach must be reimagined to meet the unique challenges posed by AI.  

While most people have already experimented with ChatGPT, Gemini, Copilot, or other platforms, very few of them understand the technology well enough to be effective in using it. It’s also common for employees to feel threatened regarding their job security, to not trust the technology, or to be resistant to changing their tried-and-true ways of working. 

Technology mistrust: There’s been no shortage of news stories surrounding AI hallucinations and inaccurate or unrealistic model outputs leading to poor business decisions and breaches of trust. While foundational AI models will be prone to occasional errors for the foreseeable future, these can mostly be mitigated through a properly aligned AI strategy and effective tool use according to best practices. 

Job security concerns: Successful AI adoption requires employees who are enthusiastic about leveraging the technology proficiently. Yet many share a common concern that they are going to be replaced by increased automation or by intelligence they cannot match. In some cases, this may be true, and it has both positive and negative implications. However, in our conversations with clients, almost nobody is talking about replacing employees with AI—they simply want them to be more productive and effective at what they are already doing. The employees who survive will be those who embrace the technology and learn to use it effectively, yet many organizations fall short of managing these expectations. Even when expectations are clear, many organizations still don’t provide adequate and targeted resources to upskill their workforce. 

Stakeholder alignment: AI initiatives range from top-down cost-cutting or efficiency-maximizing goals across the enterprise to individual experimentation. The only approach that works is to have stakeholders aligned from the outset, from executives to managers to the most junior employees. Few organizations make the effort to foster this alignment. Executives and department heads must lead by example, change champions and super users should be embedded within different functional areas to provide localized support and alleviate mistrust, and all stakeholders must share common goals through prioritized use cases. 

Process adaptation: AI integration fundamentally changes the methodologies and processes by which work gets done. Quality system architectures, targeted use cases, sophisticated tools, and well-trained employees can still fall short of expectations if processes and working methodologies aren’t adapted upstream and downstream of AI tool use. The most cutting-edge tools used by the most upskilled employees are useless if they don’t have the right inputs coming from upstream business functions or if the tools’ outputs are misaligned with downstream expectations. Critically, this process must be iterative—it’s rare for adaptation to be immediate, requiring an iterative, agile approach. 

Clear communication: It may sound cliché, but vision, expectations, priorities, and procedures must be communicated thoroughly and transparently throughout the organization. Communication through multiple channels is often necessary; email updates, town halls, office hours, frequent updates from managers, and other media should all be leveraged to ensure top to bottom alignment. 

Getting it right 

Building on our years of data strategy and change management expertise, while also having a front-row seat to the common AI pitfalls encountered by businesses, Clarkston is uniquely positioned to guide businesses through AI integration. We start with strategy alignment, ensuring: 

• Prioritized, high-value use cases 
• Sustainable, well-governed system and data architecture to support AI use cases 
• Proper organizational design and talent readiness 

In parallel, we emphasize a change management approach unique to AI implementation, including: 

• Stakeholder alignment from executives to junior employees 
• Clear communication to build trust and clarify expectations 
• Training and enablement 
• Adaptation of workflows to account for AI use 
• Most importantly, an iterative, agile approach 

With this approach, we are confident any organization can find success—greater efficiencies, faster response times, improved productivity, deeper insights—through leveraging the exciting advancements in AI technology. Reach out to Clarkston today to have our AI experts guide your business to success in the age of AI.