How to Win the AI Transformation Game?

To say there is enthusiasm for AI would be a trillion dollar understatement. Over 90% of companies have AI in their 2025 strategy in some form or another. Yet for every figure like that, there’s a number raining on its parade. Gartner estimates that over 4 in 10 AI projects never make it. MIT even posits that 95% of AI endeavours never provide any value whatsoever. 

It seems then, that the AI game is currently played by talking the talk, rather than walking the walk. There is a lot of locker room talk. McKinsey boasted having over 12.000 agents. What does that even mean?

Having worked with different companies in their AI transformations, we thought it interesting to look at some of the winning AI strategies so far. 

Disclaimer: we're fully committing to this playful football metaphor. No AI was harmed in the making of this blogpost; they're entirely our own silly comparisons.

The winning strategy

Here’s what winning teams realises:

• All positions play a crucial role. Think human-AI collaboration, not “tools” first.
• Playing together is key.
• Don’t think you’re FC Barcelona.

Let’s unpack that.

Here’s our AI Maturity Model. It looks at all the different perspectives that are important, and need to come together, for AI to truly become an integral part of an organisation.

‍

Technology, of course. But AI is not an IT project. It’s equally about Strategy and vision. But about People and culture too. And an Execution layer as the playmaker.

Behind every dimension, we have a set of questions to probe for an organisation’s maturity level. Taken together, we identify four maturity levels.

‍

But here’s the kicker. The lowest score, i.e. the least developed dimension, determines the organisation's total score.

Given the pressure from the market or the board, companies that are running behind want to close that gap and get ahead. So they over-extend. And they stumble. This is what we mean by not thinking you’re Barcelona or Man City. There are no shortcuts. Just take it one step at a time.

The one-two pass

Companies have another tendency: they tend to overcomplicate things. Luckily, there’s a timeless football move to draw inspiration from. Impactful  in its simplicity. The one-two pass. The push and run.

Here’s a few examples of what we mean:

• Rather than immediately creating an ambition target architecture for tools, be scruffy: try duct taping your existing tools to first prove there’s value there. (Just make sure those tools are compliant and secure.)
• Likewise, encourage people to try different tools. (Again, without sacrificing security or GDPR-compliance.) In exchange for that trust, you can expect them to share their findings. Evaluate tools on a monthly, not yearly basis. After a while, you can start standardising around a few solutions.
• That trust extends to people on a strategic level as well. Give the mandate to people in charge of the different activities in your company. But follow up mandate with responsibility.  At first, positive AI sentiment is an important KPI, but this needs to shift to expecting actual business results.
• As this can be intimidating, don’t roll out this strategy across the entire company. Let a few lighthouse teams or departments lead the way. Their success will ignite the rest. But of course, in a longer time frame, this will change every role and team at your company. This will require change management.
• Finally, also on a strategic level, don’t run before you can walk. Surely, AI will disrupt a lot of different industries. But bite off more than you can chew: first look at your current business, and how that can be done better or faster with AI.

The coaching group

Given the importance of coordination across different disciplines and people, winning companies have one trait in common: they all have some sort of AI coaching group. AI is moving at an unprecedented pace, so it is normal typical corporate structures falter, or slow things down dramatically.

An AI task force (really, call it what you want) is there to keep the overview. What are we doing? What works? What doesn’t? How can we share learnings? How do we build enthusiasm? Which  people do we need to connect?

This group will not map your current hierarchical structure. As an exception to the rule of never comparing things with Apple, here’s a story we like. Steve Jobs would organise a yearly weekend for the ‘100 Most Important People at Apple’. These were often ‘simple’ designers or engineers. But with a love for the craft, highly enthusiastic or well connected.

Make sure your AI coaching group is like that too.

Start with why

And finally, the success will of course depend on the projects you select in the first place. For this, you need to engage in the right discussions. What are we solving? Do we have the right data and how do we turn that into information or action? What happens if we get the right? What if we don’t?  What’s the ultimate goal? 

For this, we’ve developed an AI Value Proposition Canvas. Use it to frame value, risks and metrics, and to move an AI pilot to production with confidence. 

Given the high stakes of the AI game, it shouldn’t be surprising that there’s a lot coming into play. That realisation is crucial in and of itself. But don’t let this intimidate you. Start with simple moves. And look at how the different players in your organisation can score a win-win. 

‍