how-to-metric-to-scorecardlisted

# how-to-metric-to-scorecard Data-Analytics-layer skill (layer 2 of the phoebe-data-skills 4-layer roadmap). The job is not "make a dashboard" - it is to turn a number into a *decision the board can trust*. Every metric ships with its definition and a validation that shows what it would say if you got it wrong. Showcase walkthrough (Everrest retail case, real computed scorecard): https://github.com/phoebefu6/phoebe-data-skills - `docs/how-to-metric-to-scorecard/` ## Where this sits in the lineage `raw dump -> lake -> warehouse -> marts` **-> executive scorecard**. This skill reads the clean marts (the output of `how-to-schema-and-warehouse`) and turns them into the artifact an executive actually looks at. ## The one rule that separates it from every dashboard tutorial **A metric is a decision made numeric, not just an aggregate.** So every tile carries three things: the number, its definition (formula, grain, filter), and a validation - the naive version and how far off it is. If you cannot state the definition and defend it against the naive version, it is not board-ready. ## The 6 steps ### 1. Input The marts (or the user's tables): orders, order_items, returns, merchants, customers. Note the grain of each and the status field - it decides what counts. ### 2. Sample data (only when real data isn't available yet) Reuse the seeded Everrest marts (seed 42). For a client, skip - use their marts. ### 3. Objective Frame the board question: "what eight numbers does the boa