I once built a machine learning model to predict which customers would cancel their subscriptions.
Press enter or click to view image in full size![]()
It was technically successful.
89% accuracy. Accuracy: solid. Remember: decent.
I was a hero when I presented it to the team.
The customer success head then questioned, “Okay… so what do we do with this?”
Quiet.
I didn’t have a playbook, but I had constructed a crystal ball.
I discovered then that machine learning isn’t about making predictions.
It has to do with action.
The question is “What should we do about it?” rather than “Will this happen?”
ML Fails When It Lives in a Lab
Too many teams treat ML like a science fair project:
- Train model → Get high score → Show slide → Call it “done.”
But in the real world? A model that doesn’t change behavior is just a fancy calculator.
The magic happens when ML becomes a lever — pulling real outcomes, solving real pain, changing real lives.
“Don’t ask ‘Can we predict it?’…