Stop Doing AI Pilots and Start Creating Strategic Impact

Author

Charlene Li
September 22, 2025

If a tree falls in the forest and there's no one to hear it, does it make a sound? Here's a better question. If your AI pilot succeeds brilliantly but has no path to scale, does it create value?

The answer is a resounding no.

I'm writing this from London, where I've been speaking with organizations about a topic that's close to my heart: escaping what I call "pilot purgatory." It’s a familiar cycle—small budget, limited data access, no clear owner, and a focus on technical proof-of-concept rather than business outcomes. Sound familiar?

Here's the uncomfortable truth. MIT research shows that 95% of AI pilots end up failing. But contrary to what many assume, it's not because the technology doesn't work. It's because of how these pilots are constructed.

The Three Fatal Flaws of AI Pilots

I've seen this pattern repeatedly across organizations, and it's always the same three issues that appear time and time again:

🤔 Vague metrics 

When success isn't clearly defined, failure becomes inevitable. "Let's see if this works" isn't a strategy…it's wishful thinking.

🚧 No scaling pathway 

Even when pilots succeed, there's no plan for what comes next. No resource allocation, no process changes, no organizational preparation.

🔄 Endless drift 

Without clear decision deadlines, pilots become organizational zombies—neither alive nor dead, just consuming resources while delivering nothing.

A Different Approach: Strategic AI Initiatives

Here's what my co-author Katia Walsh and I advocate for instead: stop doing AI pilots altogether.

Instead, treat AI initiatives as strategic imperatives. If you're going to use AI, commit fully. Make it count. Link it directly to your most important business objectives.

When I shared this perspective recently, people literally stood up and cheered. They were exhausted by pilots that led nowhere. Committing to well-designed initiatives is good leadership from a strategic angle, and a way to show your people that time they invest into building something new is valued.

The Double-S Prioritization Matrix

So how do you choose which initiatives deserve this level of commitment? First, forget feasibility as your primary filter. Technology evolves too quickly for feasibility to be a significant barrier. 

Instead, allocate resources to your strategic priorities based on two factors:

📈 Size of value: How big is the potential impact?

🔜 Speed-to-value: How quickly can you deliver results?

This creates four categories:

🚀 Momentum makers: High value, high speed

🏆 Quick wins: Lower value but fast delivery 

🎲 Strategic bets: High value but longer timeline 

🥀 Dead ends: Low value, slow delivery 

Designing for Friction

Here's something most organizations get wrong: they try to eliminate conflict during AI implementation. But friction is where innovation happens.

Brené Brown spoke in an interview about having the courage to have difficult conversations and the need to build cultures where people feel safe to bring their “whole selves” and be honest. Disagreement requires safety, trust, and norms that allow dissent. Leaders who foster an environment where disagreement can lead to incredible discoveries are the leaders building winning organizations.

When it comes to implementing AI (in pilot form or otherwise), committing halfway is the killer. If everyone has to agree before you move forward, you'll move at the speed of your most change-averse team member. 

On the flip side, if your team feels empowered to go all-in, you may find yourself in the 5% of AI pilots that make it. Because in AI, speed is everything.

Learning From Industry Wins

Want to see what success looks like? Look at Walmart. They put AI tools in the hands of 1.5 million store associates. They weren’t aiming for glamorous use cases, but for everyday challenges like shift scheduling and customer support. Now, about 900,000 workers use these tools weekly.

That's AI at scale. Not a pilot, not an experiment, but a strategic application delivering real value from day one.

Your Escape Route from Pilot Purgatory

Ready to break free? Here's your action plan:

1️⃣ Pick one business-critical problem where AI can make a substantial difference. 

2️⃣ Define clear metrics and scaling plans. 

3️⃣ Set decision deadlines with regular checkpoints every 4-6 weeks. 

4️⃣ Secure real sponsorship and resources.

Then (and this is crucial!) end or reframe every other pilot that doesn't meet these standards.

You'll find that when you stop hedging your bets with endless pilots and start making real commitments to AI applications that enhance your business strategy, everything changes. Your teams get focused, your stakeholders get serious, and your AI efforts finally start delivering the value everyone's been waiting for.

Remember: If it's truly part of your strategy, feasibility becomes less of a barrier because the resources and commitment follow. If it's not strategic enough to warrant that level of investment, it shouldn't be on your list anyway.

🔊 Your Turn

What AI pilots are you running right now? 

Which ones would qualify as strategic initiatives, and which ones need to be sunset? I'd love to hear about your experiences with pilot purgatory—and your plans for escape.

What I Can’t Stop Talking About

  • If you’re not ready to implement AI, you might be doing more harm than good. My AI readiness checklist can help you feel confident that now’s the time to start using AI to achieve your business goals. 

  • AI isn’t just another SaaS platform to add to your stack. But if executives don’t understand that, AI tools won’t be used to their fullest capacity, leaving resources—and potential—on the table. Empowering leaders to make the most of AI is easier than you think.

My Upcoming Appearances/Travel

If you found this note helpful, please forward and share it with someone who needs the inspiration today. If you were forwarded this, please consider subscribing.

Join me on Facebook, Instagram, LinkedIn, and Twitter.