The Canvas: Data Product Design (Week 2)

Week 2. The first real touchpoint between humans and technology.

1. Create a Metric Dependency Tree (MDT) — with a domain analyst, define each metric's logic (formula), associations, and semantics (naming, description, synonyms, tags).
2. Create a logical data model to power the MDT — this doubles as your Data Product Prototype. It has five parts:
   • Entities — core business objects (accounts, transactions, customers, deals).
   • Dimensions — attributes on those entities (purchase_date, last_contacted).
   • Measures — quantified/aggregated dimensions (avg, count, sum).
   • Relationships — associations between entities (1:N customers → transactions).
   • SLOs — data quality / access conditions.
3. Plug into the MDT — wire measures and dimensions into the metric formula (metric = measures × dimensions).
4. Iterate logic with stakeholders to validate the model.
5. Validate the prototype against realistic dummy data before touching production data.