What Data Modeling Is and Is Not (Ch 2 Revised)

Revised Chapter 2 establishing the definition of data modeling. Opens with an e-commerce disaster case study (500-column orders table, 6 inconsistent customer databases, $400K monthly refunds) to ground the discussion.

Reis’s definition: “A data model organizes and standardizes data in a precise, structured representation, enabling and guiding human and machine behavior, informing decision-making, and facilitating actions.” Deliberately emphasizes machines as first-class consumers — a departure from prior definitions (Hoberman, Kent, Burns) that focused primarily on human communication.

What data modeling is NOT: not perfect, not just physical storage (the “Bed of Procrustes” anti-pattern), not a single approach, not a one-time thing, not just for big enterprises or technical teams. Smaller companies benefit more from early modeling because processes are simpler to capture.

RDCO relevance

The machine-consumer emphasis in the definition aligns with our positioning around AI-ready data models. The “Bed of Procrustes” anti-pattern (forcing business logic into physical storage constraints) is exactly what we see clients do with dbt when they start from the warehouse schema instead of the business domain.