A developer's guide to data modelling breaks down moving from SQL queries to designing database tables. The process starts by capturing requirements from users, distinguishing between functional and non-functional needs. This initial step is iterative, requiring constant refinement as the problem is better understood.

The guide outlines a four-stage workflow: conceptual, logical, and physical models. A key insight is normalization—structuring data to avoid duplication. Using a hospital system example, the author shows how creating a Clinic entity prevents repeating addresses across every Appointment record, improving data integrity.

The logical model evolves into an Entity Relationship Diagram (ERD), defining attributes and relationships. Advanced techniques like supertypes (a generic 'people' table) and subtypes (specific 'patients' or 'consultants') prevent data inconsistency. The final physical model translates this design into actual SQL, where performance decisions like indexing become critical.

Ultimately, the author argues that good database design is about thinking, not syntax. While document storage seems convenient, it often leads to harder queries and duplicated data. A relational approach, though more upfront work, yields more maintainable and flexible systems long-term.