Why 95% of Enterprise AI Projects Fail — And How to Avoid the Trap
Having seen first-hand how transformational AI can be when it’s done well, I’ve been surprised at the level of scepticism.
So when I read the new MIT report (The GenAI Divide: State of AI in Business 2025) showing that 95% of enterprise AI projects fail to deliver results, I was shocked.
On closer inspection, the report echoes what I’ve been saying for months:
👉 The businesses that have streamlined and simplified processes are the ones reaping the benefits.
👉 The ones trying to bolt AI onto messy systems are stalling at the first hurdle.
MIT highlights three main reasons why most projects fall short:
Brittle workflows – AI struggles when processes are rigid or broken.
Lack of contextual learning – deployments don’t adapt or improve over time.
Misalignment with daily operations – AI is often layered onto the wrong tasks, adding little real-world value.
In short: AI isn’t failing. Our systems and processes are.
What Businesses Can Do About It
The good news? Businesses can avoid becoming part of that 95% by focusing on the foundations first. Here are three practical steps:
1. Simplify Workflows
Before introducing AI, map how work actually happens. Where are the bottlenecks, the duplications, the unnecessary steps? Simplify and standardise processes first. AI only multiplies what’s already there. If you start with chaos, you’ll just get faster chaos.
2. Structure Your Data
AI is only as good as the information it can access. If your data is scattered across systems, locked in spreadsheets, or poorly maintained, AI will struggle to deliver meaningful insights. Clean, structured, accessible data is non-negotiable.
3. Target the Right Problems
Not everything needs an AI solution. Start by identifying specific, high-value problems where AI can genuinely help:
Automating repetitive admin
Summarising or interpreting large volumes of text/data
Supporting decision-making with predictive insights
When you focus AI on clearly defined pain points, ROI comes into view.
The Bottom Line
AI isn’t a magic wand. It’s a multiplier.
If your systems are simple, standardised, and aligned — AI will accelerate your progress.
If they’re messy, rigid, or misaligned — AI will just amplify the cracks.
The MIT study is a warning, but it’s also an opportunity.
If 95% of projects are failing, there’s a huge advantage for the 5% who get it right.
At N16, this is where we work with business leaders: building clarity in their operations so every investment — whether in people, tools, or AI — delivers real returns.