01 · PLAN — AI STRATEGY
WhymostAIprojectsnevergolive
9 in 10 never make it past the test. Here’s how to make yours one that does.
The demo works. Then nothing happens.
A team tries AI on a real problem. The demo looks great and everyone is excited. Six months later, nobody is using it.
If that sounds familiar, you’re in good company. It is the most common way AI projects end, in small firms and large ones.
The good news: the reason is rarely the AI itself. Projects stall for the same few reasons, and all of them can be fixed.
Three reasons projects stall
- 1. The data isn’t readyThe demo used a clean sample. Real data is messy, scattered and incomplete, and nobody planned the work to fix it.
- 2. Nobody agreed on successWithout one clear goal and a number to beat, nobody can say “yes, this works”, so the project drifts.
- 3. No plan to run itA demo runs once. A real system runs every day and needs monitoring, updates and a budget. Costs and errors grow when real users arrive.
“A clear no in four weeks is worth more than a maybe after a year.”
How to make yours go live
The projects that make it treat the demo as step one, not the finish line.
- 2–4 WEEKSPlanTest one idea on your own data. Agree the goal first. Accept a clear no as a good result.
- IDEA TO PRODUCTIONBuildBuild for real use from day one: tests, documentation, your systems, your team involved.
- AFTER LAUNCHRunWatch quality and cost every week. Retrain when the data changes. This is the step most projects skip.
Most AI projects don’t fail because the technology isn’t ready. They fail because the plan stops at the demo. Plan for the day after, and yours can be one that goes live.
Is your AI idea stuck at the demo?
Tell us about it. We’ll say plainly whether it can go live, and what it would take.


