Part 2 · Six weeks, one framework a week · 4 of 7
Activation: Input Ports, Transforms, Output Ports, SLOs (Week 3)
A data product activates once four pieces exist: input ports, transformation steps, output ports, and SLOs — ideally declared, not hand-built.
1 min read
Week 3. Activating the logical model means assembling four pieces:
- Input ports — where data enters, from source systems.
- Transformation steps — the logic that turns raw input into the model's measures/dimensions.
- Output ports — where consumers read the result (often a Gold-layer table; see Medallion Architecture: Bronze, Silver, Gold).
- SLOs — quality and governance conditions on the pipeline.
Without a self-service layer, these four "simple" pieces become hundreds of sub-pieces: credential management, workflow/service configuration from scratch, tool integration, and transformations only a few senior engineers understand. A good self-serve platform lets an analytics engineer just declare input/output locations and transform steps in a spec file (e.g. YAML) and get ready-to-use workflows, monitors, and connectors for free.
Source: 6-Week Data Products Product Playbook.pdf
Where this shows up