Executive Corner - AI ROI: The Strategic Mandate for H22025 by Mark Hewitt

For years, the boardroom discussion around AI centered on one question: “Should we be investing in it?” That era is over. The more pressing question now reverberating in quarterly earnings calls, leadership offsites, and shareholder meetings is,: “We have AI… so where’s the return?”

The global enterprise landscape has reached a critical inflection point. Billions have been invested in artificial intelligence over the past three years, yet most organizations remain mired in AI pilot purgatory. According to McKinsey’s latest workplace AI report, nearly every company has deployed AI in some form, but just 1% believe they’ve reached maturity. The overwhelming majority are stuck in endless experimentation, producing promising demos, but little in the way of measurable business value.

That dissonance between investment and outcome is becoming impossible to ignore. Boards are asking tougher questions. CFOs are pushing for clearer ROI metrics. And the market is drawing a sharper line between those turning AI into value and those running elaborate, and expensive, science projects.

As leaders guiding enterprises through digital transformation, we need to reframe our approach. AI is no longer about possibility. It’s about performance. That begins with identifying what’s getting in the way of measurable returns.

The Hidden Friction Undermining ROI

AI’s challenges are not rooted in the technology itself. They stem from how it is being applied, and misapplied, within enterprise environments.

One of the most common issues we see is a fixation on capability over consequence. Enterprises often fall into what we call the “shiny object” trap: dozens of AI pilots running in parallel, each solving niche use cases without alignment to top-line or bottom-line impact. A chatbot here, an image classifier there, a few predictive models for good measure, yet none designed to move the financial needle.

Worse still, many of these initiatives begin with technology, not business pain. We ask, “Where can we use AI?” instead of, “What problem is bleeding money, and could AI fix it faster or smarter than we can today?” This backward logic leads to fragmented efforts that impress stakeholders in demos, but underwhelm in results.

Measurement further compounds the problem. Too many organizations still evaluate AI success through the lens of model precision, throughput, or deployment velocity. These technical indicators, may look good on internal dashboards, but are often disconnected from customer impact or financial improvement. A highly accurate model that fails to reduce churn or increase margin isn’t a win. It is a distraction.

And then there’s the matter of cost. The infrastructure tax associated with AI is real, and growing. Up to 80% of the effort in AI projects goes into data preparation, system modernization, and post-deployment maintenance. These are not glamorous tasks, but they are essential and expensive. Many organizations fail to factor them into their ROI calculus, resulting in grossly optimistic projections and painful downstream corrections.

Defining What Good Looks Like

So what does successful AI investment actually look like in 2025? The answer lies in a clear, disciplined framework that moves from efficiency to intelligence to transformation.

In the near term (think three to six months), high-performing companies are realizing quick wins through operational efficiency. They are automating manual workflows, streamlining customer support, and optimizing logistics. These early victories are not only self-funding; they create internal momentum and political capital for broader AI programs.

Mid-term returns typically emerge between six and eighteen months. Here, AI enhances decision-making across forecasting, personalization, and strategic planning. This layer of value often proves most lucrative, and recent research from Google Cloud suggests that 74% of enterprises currently seeing returns from generative AI are doing so through improved insights and decision agility.

Long-term impact (eighteen months and beyond) is where the truly transformative value lies. New business models, AI-powered services, and reimagined value chains become possible. But none of this materializes unless the foundation is solid. Without Tier 1 and Tier 2 success, Tier 3 aspirations remain just that, aspirations.

Across each tier, one principle holds: focus, measure, and scale. Leading firms identify a high-impact use case, establish baseline KPIs, demonstrate ROI, and then scale that use case horizontally and vertically. It is not sexy, but it is effective.

Building the Roadmap for AI Value in 2025

Organizations serious about AI returns need to take a far more disciplined and ROI-centric approach to project selection. This begins with a firm stance: no AI initiative should proceed without a 90-day value hypothesis and quantifiable success criteria. If a project cannot prove its worth quickly, it should not move forward.

Measurement systems must also evolve. That means moving beyond IT dashboards to real-time ROI tracking tied directly to business performance like revenue uplift, margin expansion, risk reduction, customer retention. Technical excellence is no longer the benchmark; business contribution is.

Equally important is how the organization governs its AI investments. Value realization must be a shared accountability across technology, operations, and finance. Leading companies are standing up AI value offices or transformation task forces with C-suite sponsorship and a clear mandate: prioritize, track, and terminate as needed.

Modern cloud platforms, such as those offered by Google Cloud, now come with embedded ROI tracking features and measurement frameworks. Tools like these can help enterprises move from gut-driven experimentation to data-driven prioritization, a shift that separates the market leaders from the rest.

The Accountability Era Has Arrived

H22025 will be remembered not for the explosion of AI interest, but for the acceleration of AI accountability. Nearly every enterprise is increasing AI investment this year, but throwing more money at disconnected projects is not a strategy. Without proper measurement, it is waste at scale.

The question you must ask yourself and your leadership team is, “Can we prove, in business terms, the impact of every AI initiative we’ve launched?”

If the answer is no, then it is time to audit your AI portfolio. Deprioritize science projects. Double down on what delivers. Build the measurement muscle now, because your board, your customers, and your market will not wait.

The companies that treat AI as a strategic lever and not just a technical novelty will create lasting advantage. The rest risk becoming the next cautionary tale in a long line of technology fads that never quite delivered.

The window to lead with value is closing. The path is there. The choice is yours.