Rethinking Business Cases in the Age of AI: What Boards Need to Know

In 2015, I created Amazon Web Services’ first comprehensive cloud migration business case tool. The Business Case Accelerator (BCA) went beyond simple IT-focused Total Cost of Ownership (TCO) calculations to include migration costs, data centre decommissioning expenses, Return on Investment (ROI), Net Present Value (NPV) and the payback period. While not revolutionary, the BCA provided Boards with the financial clarity they needed to confidently approve cloud transformations.

Today, artificial intelligence presents a fundamentally harder business case challenge, but with potentially greater rewards. AI’s unprecedented hype cycle is driving rushed investment decisions, while traditional business case frameworks fail to capture AI’s unique value creation patterns. This enthusiasm, however, often outpaces results. According to McKinsey’s latest global AI survey, more than 80% of organisations have not seen a tangible impact on enterprise-level EBIT from their use of generative AI. This disconnect between investment and measurable return highlights just how poorly suited traditional business case models are for evaluating AI’s complex, cross-functional value.

Directors and executives who apply conventional evaluation models risk misallocating resources and missing AI’s true strategic potential.

In this series of articles, I’ll share a new blueprint for building effective AI business cases, starting with understanding why traditional approaches fall short and what Boards need instead. Drawing on my experience with both cloud and AI transformations, I’ll provide practical guidance for navigating this complex landscape.

Cloud vs. AI: A Fundamental Divide

The differences between cloud and AI adoption patterns create the first major challenge for traditional business cases. Cloud adoption typically follows a predictable, sequential path that aligns well with conventional IT financing models.

Cloud’s Sequential Journey

When guiding cloud adoption, I’ve observed a relatively consistent progression through what we called the Cloud Stages of Adoption (SoFA):

1. Project - Organisations started with isolated proof-of-concepts
2. Foundation - They built centralised governance and landing zones
3. Migration - Systematic movement of workloads followed
4. Reinvention - Finally, business transformation leveraged cloud capabilities

Other than some oscillation between stages (mainly due to adoption and migration blockers), this sequential path made financial modelling straightforward. Cloud business cases could rely on predictable cost categories: migration expenses, operational savings, and infrastructure decommissioning. The cloud journey was primarily IT-led, with technical teams at the centre of both the conversation and implementation. TCO discussions were comfortable territory for IT leaders, who could translate infrastructure metrics directly into financial terms. Most importantly, value creation followed a relatively consistent timeline that conventional ROI calculations could capture.

AI’s Parallel Chaos

By contrast, AI adoption resembles a multi-speed, parallel process that defies sequential modelling. As I outlined in my AI Stages of Adoption (AISA) framework — a model I developed to map how different parts of a business adopt AI at different speeds — organisations typically have multiple AI initiatives simultaneously operating at different maturity levels across the business. While the marketing team might be reaching the Scaling stage with AI-driven content generation, the finance department could still be Experimenting with fraud detection models. At the same time, customer service might be Adopting AI chatbots, product development may be Transforming their processes with generative design tools, and operations might be Optimising their workflows with predictive maintenance systems. This creates a complex landscape where a single organisation traverses multiple AI maturity stages concurrently.

Each initiative follows its own timeline, requires different resources, and delivers value on varying schedules. Some machine learning-driven applications, like Large Language Models (LLMs), can deliver ROI in days. Others, such as building custom foundation models, might take many months or years before showing returns.

For Boards accustomed to cloud’s relatively consistent adoption patterns, AI presents a jarring contrast. Consider a hypothetical manufacturing company with a three-year cloud migration that followed a predictable ROI trajectory. Their AI portfolio looks entirely different: multiple concurrent initiatives with timelines ranging from weeks to years, each with distinct risk-reward profiles and requiring different evaluation approaches. The Chief Financial Officer finds it impossible to apply their standard investment criteria uniformly across such varied initiatives.

This multi-speed reality directly challenges conventional business case approaches that rely on uniform assumptions across initiatives. When Boards attempt to apply the same evaluation framework to all AI investments, they inevitably misallocate resources; either missing quick wins by demanding excessive rigour or approving high-risk projects with insufficient scrutiny.

Where Traditional Models Break Down

Traditional business case frameworks suffer from three fundamental limitations when applied to AI investments:

1. They Misrepresent AI’s Complexity

Standard business case calculations (TCO, ROI, NPV, etc.) assume consistent cost structures and benefits across projects. This works well for technology investments like cloud, where migration costs and operational savings follow predictable patterns. For AI use cases, however, different applications have dramatically different profiles.

Imagine a financial services company Board applying the same ROI thresholds to all AI proposals. This approach might green-light a customer service chatbot with easily quantifiable call centre savings while rejecting a fraud detection system that requires more upfront investment with lower initial returns. Yet the fraud solution might ultimately deliver exponentially more value through risk reduction and enhanced customer trust. This one-dimensional evaluation misses AI’s diverse value creation patterns.

2. They Undervalue Capability Building

Traditional business cases struggle to capture the full spectrum of value that technology investments create. When building the BCA, I initially incorporated three pillars of what became the Cloud Value Framework (CVF): Cost Optimisation, Risk Reduction, and Increased Agility. While working with a major financial services organisation on their cloud business case, they requested a time and motion study to quantify the resource efficiencies they stood to gain through cloud adoption. This collaboration led to Resource Efficiency becoming the fourth pillar of the CVF, highlighting how real-world implementation often reveals value dimensions that standard frameworks miss.

In the Five Pillars of AI capability framework, I take this point further by emphasising how sustainable AI adoption requires investment across multiple capability domains: Governance & Accountability, Technical Infrastructure, Operational Excellence, Value Realisation, and People & Culture - which lead to a wide range of business outcomes.

Traditional technology adoption business cases struggle to value these foundational capabilities. They treat initial AI investments as standalone projects (like IT investments) rather than recognising how they build organisational muscles for future initiatives. A retailer’s first computer vision project might show modest returns, but the data architecture, governance frameworks, and skills developed become valuable assets for subsequent AI initiatives.

Conventional business case tools miss these synergies entirely, leading to underinvestment in crucial foundation work that enables future innovation. When every AI project must independently justify its full infrastructure and capability building costs, organisations create siloed initiatives that duplicate efforts and fail to scale.

3. They Over-focus on Cost Reduction

Perhaps most significantly, traditional business cases often fixate on readily quantifiable benefits while undervaluing strategic advantages. This maps directly to the limitations I identified when creating the Well-Advised framework, which balances evaluation across multiple business dimensions beyond simple cost reduction.

While some AI applications like chatbots have found adoption under the Customer Value pillar, I’ve observed that organisations often implement them primarily for cost reduction rather than transformative customer experiences. This tendency to gravitate toward the most easily measurable benefits leads Boards to systematically underweight AI’s potential for innovation, customer growth, and business model reinvention. The misalignment becomes particularly problematic as organisations progress through the AI Stages of Adoption, attempting to move from Optimising to Transforming without appropriate strategic value metrics.

Consider how this might play out: A manufacturing company’s initial AI proposals focus exclusively on operational efficiency improvements with clear cost savings. They struggle to secure Board approval because the ROI, while positive, doesn’t meet standard investment thresholds. By expanding their business case framework to include customer experience improvements and product innovation potential, they could unlock funding for AI applications that might drive market differentiation rather than just incremental cost reduction.

AI’s Unique Valuation Challenges

Beyond the structural limitations of traditional frameworks, AI introduces several valuation challenges that demand new approaches:

Diverse and Unpredictable Costs

AI costs follow fundamentally different patterns than traditional technology investments:

• Data preparation expenses vary dramatically by use case and can dwarf technology costs
• Compute requirements often change unpredictably during development as models evolve
• Talent needs fluctuate throughout the AI lifecycle, from data scientists to domain experts
• Ongoing optimisation costs can be substantial as models require continuous tuning

These patterns represent an opportunity - unlike cloud migrations with their predictable cost curves, AI initiatives can adapt and evolve as business needs change. This flexibility requires different financial models than traditional TCO calculations, but enables organisations to respond quickly to new opportunities. For forward-thinking Boards this represents adaptability as a strategic advantage rather than a budgetary challenge.

Mixed Value Timelines

As I explored in my work on Decision Analytics, AI initiatives generate a mix of leading, lagging, and predictive indicators with different time horizons:

• Short-term operational gains like productivity improvements
• Medium-term customer impacts through experience enhancements
• Long-term strategic advantages via business model innovation

Traditional business cases struggle to compare initiatives with such varied timelines. A generative AI tool might show immediate productivity gains, while a predictive customer churn model delivers greater but delayed revenue impact. Without frameworks to evaluate these mixed horizons, Boards tend to favour short-term returns over strategic advantage.

Capability Synergies

Perhaps the most significant valuation challenge comes from AI’s synergistic nature. Investments in one area often enable value in entirely different domains:

• Data cleaned for one project becomes a foundation for others
• Governance frameworks developed for initial applications scale across the business
• AI expertise built in one department transfers to new use cases

These cross-functional synergies are entirely missed by project-based ROI calculations. An AI Centre of Excellence (AI CoE) further amplifies these synergies, creating value across the enterprise that cannot be attributed to individual projects.

For Boards, this creates a practical challenge: how do you justify investments today when a significant portion of their value will come from enabling future initiatives that aren’t yet defined? Traditional business cases focus on direct, project-specific returns and rely on tools like NPV calculations, but NPV only works when you can define the initiative and quantify its benefits. With AI’s synergistic effects, many valuable future use cases remain undiscovered until the foundational capabilities are in place. The real risk for Boards isn’t missing near-term ROI — it’s missing the strategic scaffolding these investments quietly build for tomorrow’s breakthroughs.

A New Framework for AI Evaluation

The limitations of traditional business cases don’t mean Boards should abandon financial discipline for AI investments. Rather, they need expanded frameworks that account for AI’s unique characteristics while maintaining appropriate governance.

In speaking with Chartered Directors, their Boards, and Chief Transformation Officers, I’ve found successful organisations develop evaluation approaches that:

1. Segment AI initiatives by type - Different evaluation criteria for operational, customer-facing, and strategic AI
2. Balance financial and strategic metrics - Using the Well-Advised framework or something similar to evaluate across multiple value dimensions
3. Consider capability building value - Explicitly accounting for organisational capabilities developed
4. Adopt portfolio thinking - Evaluating AI investments as a collective strategy rather than isolated projects

This multi-dimensional approach enables Boards to make more nuanced investment decisions that align with their organisation’s position in the AI Stages of Adoption. Rather than applying a one-size-fits-all threshold, they can match evaluation criteria to each initiative’s strategic context.

Conclusion: Moving Beyond Traditional Models

Traditional business cases served us well in the cloud era, providing the financial clarity Boards needed to approve transformative investments. For AI, however, these conventional models risk becoming dangerous oversimplifications that miss AI’s complex, multi-faceted value creation.

Boards need new frameworks that account for AI’s parallel, multi-speed nature, diverse value patterns, and capability synergies. These frameworks must maintain financial discipline while expanding beyond simple ROI calculations to capture AI’s full strategic potential.

In the next article in this series, I’ll outline the five essential components every AI business case needs, providing practical guidance for creating evaluation frameworks that truly capture AI’s unique value. We’ll explore how to balance rigour with flexibility, short-term returns with long-term potential, and financial metrics with strategic impact.

About the Author

Mario Thomas is a transformational business leader with nearly three decades of experience driving operational excellence and revenue growth across global enterprises. As Head of Global Training and Press Spokesperson at Amazon Web Services (AWS), he leads worldwide enablement delivery and operations for one of technology's largest sales forces during a pivotal era of AI innovation. A Chartered Director and Fellow of the Institute of Directors, and an alumnus of the London School of Economics, Mario partners with Boards and C-suite leaders to deliver measurable business outcomes through strategic transformation. His frameworks and methodologies have generated over two-billion dollars in enterprise value through the effective adoption of AI, data, and cloud technologies.