Is one big project better than several small ones?

Two approaches are often set against each other. The first multiplies small projects, quick and inexpensive, to test and learn. The second launches a big programme, ambitious and structured, to transform the company in one go.

Each has its pitfall. Small projects with no common thread produce tools that do not talk to each other, scattered data and teams tired of experimenting. The big project takes months to show a first result, and confidence wears thin before then. What works lies in between: a clear direction, broken down into steps that are each useful in their own right.

• Isolated small projects add up badly.Each one chooses its own tools and data, and the whole becomes hard to maintain.
• Big programmes lose their backers along the way.When results are slow to come, priorities change and budgets follow.
• A step that is useful in itself protects what follows.It proves, funds and makes the next step credible.

1. 01Write down on one page the overall direction your AI projects must serve.
2. 02Break it down into steps of a few months, each delivering a usable result.
3. 03For every new project, ask which step of that direction it contributes to.

When trials multiply without ever being rolled out, we help link them to a single direction.

• How do we choose our first use cases?→
• Why do our pilots never scale up?→
• Should you know how to stop an AI project?→
• How do you measure the return on investment of an AI project?→