Data modelling and normalisation

Now design your own schema, the set of tables and how they connect. Start from the nouns in the problem, such as customers, orders and products, and how they relate. One to many (one customer, many orders) becomes a foreign key on the "many" side.

Many to many (orders and products) needs a join table such as order_items. Normalisation means storing each fact once. A customer's email lives in the customers table, not copied onto every order, so an update changes one row.

Third normal form is enough for most backends. You denormalise, copying data on purpose, only to fix a measured read problem.