AI adoption has exploded, but success hasn’t.

MIT’s State of AI in Business 2025 sent shockwaves through boardrooms when it revealed that 95% of AI initiatives deliver zero measurable ROI. Despite billions spent globally, only a small fraction of organisations are seeing tangible results.

At Maxwell Bond’s AI & Data Leaders Roundtable, this finding sparked a lively debate. Why are so many AI pilots stalling before they scale? And more importantly, what separates the 5% that succeed?

 

1.    The Root Cause: Unclear Objectives

The biggest reason for failure isn’t technology, it’s purpose. Many companies launch AI pilots without a clearly defined business problem. The excitement of “doing AI” overtakes the discipline of defining measurable outcomes. Without a target ROI or KPI, it’s impossible to gauge success.

Lesson: Start small, but start with purpose. Ask: “What business problem are we solving?” and “How will we measure success?” Whether it’s reducing churn, improving forecasting accuracy, or freeing time for revenue-generating work, clarity drives adoption.

 

2.    Data Quality and the “Garbage In, Garbage Out” Trap

As one of our Roundtable attendees put it: “AI = Input → Black Box → Output.” If the input is poor, so is the output. Too many pilots overlook the time needed to clean, tag, and structure data before deploying AI models. The result? Models that hallucinate, misclassify, or generate inconsistent recommendations. Research from McKinsey echoes this: 60% of failed AI projects cite poor data quality or access as the main barrier.

Lesson: Invest early in data governance, architecture, and literacy. AI can’t fix messy data, it amplifies it.

 

3.    Culture and Adoption: The Human Factor

Even the most elegant AI model is useless if no one uses it. Our Roundtable leaders agreed that adoption is 70% people and process, 30% tech. Teams need to understand how AI fits into their workflows, and they must trust its outputs.
Change management is vital; leaders must bring people on the journey. As one attendee put it, “AI is as much a cultural shift as it is a technological one.”

Lesson: Make adoption part of your pilot’s success criteria. Train, educate, and celebrate small wins to build momentum.

 

4.    Speed vs. Strategy: The Build Trap

In the race to innovate, many organisations deploy AI too quickly. They buy pre-packaged tools or push proof-of-concepts live without the infrastructure to sustain them. As one leader noted, “AI pilots often work in isolation, but scaling requires integration with the business.” The result? Quick wins that fizzle when scaled.

Lesson: Pilot with scalability in mind. Even if your MVP is small, design it for future integration, not as a dead-end experiment.

 

The Winning Formula: Input, Adoption, Culture

Our Roundtable summarised success into three words: Input. Adoption. Culture.

Get your inputs right, ensure people actually use the tools, and embed a culture that values experimentation. That’s how the 5% win.

 

How Maxwell Bond Can Help

To succeed, you need more than good tech, you need the right people. At Maxwell Bond, our AI, Robotics & Data division helps companies across the UK, EU, and US build high-performing AI teams, from Data Scientists and ML Engineers to AI Product Leads and Heads of Data Strategy.

We also connect senior leaders through our AI Roundtables, where executives share insights, challenges, and strategies that actually deliver results.

👉 Contact Maxwell Bond to build your AI team and strategy for 2026.
👉 Reach out to Andy Holt to join our next in-person or virtual AI Roundtable.