Why do our pilots never scale up?

A pilot is designed to show that an idea works. It is given a controlled scope, clean data, a motivated team, and it is shielded from the rest of the company. That is exactly what makes it succeed, and exactly what stops it growing.

When the time comes to roll it out, everything that had been set aside comes back: less clean data, neighbouring departments that were not consulted, security rules that were bypassed, teams that did not choose the tool. The pilot did not fail. It answered a different question from the one being asked now.

• The pilot measures the task, not the flow.It proves that one step goes faster, without looking at what that changes for the next step.
• Nobody owns it after the demonstration.The project team leaves, and the business inherits a tool it did not design.
• The conditions for rolling it out were never written down.Neither the running cost, nor the level of control, nor the criteria for deciding to go further or to stop.

1. 01Before the next pilot, write down what it will have to prove to be extended, and who will take that decision.
2. 02Identify the departments upstream and downstream of its scope, and involve them from the start.
3. 03Name a business owner who will stay when the project team has left.

We regularly pick pilots up where they stopped, to decide with you whether to relaunch, redirect or stop them knowing why.

• How do you measure the return on investment of an AI project?→
• Should you know how to stop an AI project?→
• Can automating one department unbalance another?→
• ‘A successful pilot proves it will work everywhere.’→