The Reason AI Projects Fail, & It's Got Nothing to Do with the Tech

Almost half of UK businesses that started AI projects in2025 have already abandoned them. Not because the technology let them down, but because the people at the top never got involved.

If you're a business leader who's been watching from the sidelines, wondering whether AI is worth your time, here's the good news: it'snot too late. But the research is now very clear on one thing. When leaders engage directly with AI, it works. When they don't, it fails.

The numbers are hard to ignore

Let's start with the headline. According to S&P Global Market Intelligence, 42% of firms abandoned their AI initiatives in 2025. That's more than double the 17% that pulled the plug the year before. At the same time, the British Chambers of Commerce reported that AI adoption among UK SMEs climbed to 35%, up from 25% a year earlier.

So, more businesses are trying AI, but more are failing at it, too.

That's not a technology problem. That's a leadership problem.

What does success actually look like?

Boston Consulting Group wanted to know why some businesses are getting real results from AI while others are burning cash. They surveyed 2,360 executives across 16 countries and found a group of CEOs they call "Trailblazers." These are leaders who spend six or more hours a week personally getting to grips with AI tools.

The result? Their organisations are 12 times more likely to be among the top 5% of companies winning with AI.

Read that again. Not 12% more likely. Twelve times.

And here's the thing. Only 15% of CEOs fall into that Trailblazer category. The other 85% are either cautious followers or sitting somewhere in the middle, investing when they see clear value but not really rolling up their sleeves. The gap between these groups is getting wider, not narrower.

BCG's separate report, The Widening AI Value Gap, reinforces this. Globally, just 5% of companies are what they call "future-built" for AI, consistently generating real value from it. Among those firms, nearly all C-level leaders are deeply engaged with AI. Among the laggards? Just 8%. The pattern repeats everywhere you look, leadership engagement is the single biggest predictor of whether AI delivers outcomes.

Here are the risks, and how to channel your team's enthusiasm into something structured and useful. You can't do that from the boardroom. You have to get on the keyboard.

You can't outsource understanding

If your instinct is to hire someone to handle this, think about what happened to Builder.ai. In May 2025, this AI startup collapsed into insolvency. It was valued at over a billion dollars. Microsoft backed it. It had raised more than $450 million. And overnight, the SMEs that had built their operations around its platform found themselves with applications that no longer worked and data they couldn't access.

That's the sharp end of a broader issue. Many AI vendors selling to SMEs are simply "wrappers." They package up the same technology you could access directly from tools like ChatGPT, Claude, or Copilot, and charge a premium for it. A leader who has spent time using AI personally can spot this in minutes. A leader who hasn't is buying blind.

This doesn't mean you shouldn't work with external partners. It means you need enough first-hand understanding asking the right questions, evaluate what you're being sold, and avoid building your business on someone else's house of cards.

Why starting now actually puts you ahead

Here's the part most people miss. The 42% failure rate isn't just bad news. It's an opportunity.

A huge number of your competitors have tried AI, failed, and retreated. They're more cautious now, more sceptical, and less likely to try again soon. If you start now, not with a massive transformation programme but with genuine personal engagement, you're stepping into a space that others have left.

The BCG research shows that it doesn't take a computer science degree. It takes six hours a week of hands-on use. That's less time than most business owners spend in meetings that could have been emails. And it compounds. The Trailblazer CEOs don't just use AI themselves; they invest 60%of their AI budget on upskilling their workforce. They create a cycle where confidence builds capability, and capability builds results.

Start with one question to your team: "What takes themost time but adds the least value?" That's your first AI use case. Pickone tool, apply it to one problem, and learn alongside your team. You'll understand more about AI's potential and limitations in a week of actual usethan in a year of reading about it.

The bottom line

We're past the point where AI is optional for UK businesses. Research fromMicrosoft and WPI Strategy points to a £78 billion productivity gap opening up over the next decade between SMEs that adopt AI and those that don't.

But the data is equally clear that throwing money at AIwithout leadership engagement is worse than doing nothing. It wastes resources, creates risk, and breeds cynicism in your team. BCG's own advice is telling:70% of the effort in AI transformation should go into people and processes, 20% into technology, and only 10% into the algorithms themselves. The biggest roadblocks aren't technical. They're organisational.

The businesses winning with AI aren't the ones with thebiggest budgets. They're the ones where the leader got on the keyboard, understood the tools, and built the culture around them. It has to come fromthe top, and it starts with you.

It's not too late to be one of those leaders. But it does require you to start.

At ConkerAI, wehelp SME leaders get hands-on with AI. Not through jargon and theory, butthrough practical training built on 21 years of running businesses ourselves. If you're ready to move from the sidelines to the keyboard, get in touch and let's talk about where to start.