*What it measures*: Does this project create a dataset, a feature store, or a model component that other future projects can use? *Why it matters*: If project A takes 3 months but builds a "Customer Transaction Feature Store" that makes Projects B, C, and D take only 2 weeks each, it is arguably more valuable than a standalone project with higher immediate ROI. *Lean goal*: maximize the number of projects that don't start from scratch