The Absence of Product-Thinking in the Boardroom.

Consider the last board meeting where AI featured in the strategy deck. The slides were polished. The use cases were compelling. The vendor had impressive references. Perhaps you even approved the budget.

Now consider this: MIT research published in 2025 found that 95% of enterprise generative AI solutions deliver zero measurable impact on the bottom line. Not underperforming. Not taking longer than expected. Zero. That figure has been debated - some critics argue the methodology is too narrow. Even if the real number is half that, the signal is unmistakable.

Organisations are not failing at AI because the technology is immature. They are failing because they are asking the wrong questions before they start.

The Spending Paradox: More Investment, Less Impact

The numbers tell a contradictory story. Salesforce’s CIO Trends report shows that full AI implementation jumped from 11% to 42% year on year — a 282% increase. TEKsystems reports that 71% of organisations plan to increase AI spending again in 2026. The acceleration is real.

Yet the returns are not keeping pace. S&P Global Market Intelligence found that the percentage of companies abandoning AI initiatives before production surged from 17% to 42% in a single year. Deloitte’s “AI ROI Paradox” study of 1,854 executives found that only 6% reported payback in under a year, with typical AI ROI taking two to four years — far exceeding the 7–12 month norm for technology investments.

That same TEKsystems research reveals a telling shift in confidence: only 27% of decision-makers now expect digital transformation ROI within six months, down from 42% a year earlier. Expectations are sobering. The question is whether strategy is sobering with them.

The Pilot Graveyard: Where Good Intentions Go to Stall

We have seen this pattern before. McKinsey’s landmark transformation research established that 70% of digital transformations fail. BCG’s 2021 global study of 850+ companies found that only 35% achieved their value targets. The AI era was supposed to be different. It is not.

In fact, BCG’s 2025 AI maturity assessment of 1,250 firms found that just 5% qualify as “future-built” (organisations achieving material AI value at scale). Those 5% are pulling away dramatically: 1.7 times the revenue growth and 3.6 times the total shareholder returns of their peers.

The gap between front-runners and everyone else is not a technology gap. It is a thinking gap.

The Missing Operating System: Why Product Thinking Changes Everything

PwC’s 2026 AI Predictions make a striking assertion: technology delivers only about 20% of an initiative’s value. The other 80% comes from redesigning work. If that ratio is even approximately right, it explains why pouring more money into models, platforms and vendors keeps producing the same disappointing results.

The fix is not better technology. It is better orientation. It is product thinking.

Product-led organisations ask fundamentally different questions before committing to AI. Not “What can this technology do?” but “What customer problem does this solve, and how will we know it worked?” Not “How do we automate this process?” but “What outcome are we optimising for, and who owns it?”

The evidence that this orientation delivers is now substantial. McKinsey’s Operating Model Index, surveying over 400 public companies, found that those in the top quartile of product operating model maturity deliver 60% greater total returns to shareholders and 16% higher operating margins. Crucially, McKinsey found that “ways of working” — the product management disciplines of outcome ownership, cross-functional collaboration and continuous discovery — had the single greatest impact on business performance. 

Yet maturity scores in this area remain among the lowest across industries.

The March 2026 issue of Harvard Business Review puts it directly…

Nelson and Davenport argue that companies must shift from project-based approaches to digital product management - and that emerging technologies like AI may make project-based outcomes even less predictable than before. That is a statement worth reading twice if your AI programme is structured as a series of projects with defined end dates rather than as a continuous product discipline.

Gartner’s 2025 Leadership Vision for Chief Product Officers reinforces the point from the other direction: with generative AI moving past the peak of inflated expectations, CPOs are now being asked to figure out how to deliver actual ROI from AI. That responsibility cannot sit with technology alone. 

It requires the customer compass that product thinking provides.

Three Symptoms Your AI Strategy Is Technology-Led, Not Product-Led

From our work with leadership teams navigating digital transformation at Adastrum, three patterns recur in organisations whose AI investments are drifting towards the 95%.

Silos over dialogue. The AI programme is owned by technology. The product team, commercial leadership and customer-facing functions are consulted, if at all, after the architecture has been decided. The initiative is designed around internal readiness - infrastructure, data pipelines, security — rather than customer readiness. The demo impresses the board. The customer never notices.

Tool-first thinking. The organisation selected a platform before defining the customer problem it would solve. Vendor evaluation preceded problem validation. In practical terms, this means the AI roadmap reads like a technology shopping list rather than a customer journey. If your roadmap describes capabilities rather than outcomes, you have already joined the 95%.

Governance over discovery. The board demands certainty. Proposals are framed for minimal risk, which kills experimentation. Every AI initiative requires a full business case before a single assumption has been tested. This is the opposite of how the 5% operate. Product-led companies test assumptions cheaply and quickly, then invest in what the evidence supports. They use data as a spotlight, not a cage.

Who Sits at the Table Matters More Than What’s on the Roadmap

Product-led AI adoption has implications for who sits around the leadership table. Your CPO and CTO must speak the same language. Customer outcomes should come first, technology choices second. Your CFO must understand customer lifetime value, not just implementation cost. 

If you do not have leaders who think in outcomes rather than outputs…

The most sophisticated AI strategy in the world will stall in pilot.

This is where cross-industry perspective becomes critical. Organisations that benchmark only against their own sector recycle the same assumptions. The breakthrough often comes from a leader who has seen product-led AI succeed in an entirely different industry - someone who brings pattern recognition rather than tribal knowledge.

Before Your Next AI Investment, Check the Operating System Running It

Most organisations are investing heavily in AI capability. 

Far fewer are investing in the product thinking needed to turn that capability into customer value. The technology is not the bottleneck. The operating system around it is.

Three questions for your next leadership meeting:

“Can every member of our executive team articulate the specific customer problem our AI investments are solving, and how we will measure success?”

“Is our AI roadmap structured around customer outcomes or technology capabilities?”

“Who owns the customer problem, and do they have a seat in the room where AI decisions are made?”

If those questions create discomfort, they are working.

If you want an independent, evidence-based view of whether your team is built for product-led AI adoption, you can begin a confidential conversation with Adastrum.