Data normalization in SQL databases is the process of organizing the fields and tables of a database to minimize redundancy and dependency. Its primary goals are to reduce data anomalies and ensure data integrity by dividing large tables into smaller, related tables.

Normalization is important because it:
1. **Eliminates Redundancy**: Redundant data is minimized, which saves storage space and reduces the likelihood of inconsistencies.
2. **Improves Data Integrity**: By enforcing logical data relationships, normalization helps maintain accuracy and consistency across the database.
3. **Facilitates Maintenance**: Updates, deletions, and insertions are simpler and more efficient, reducing the risk of anomalies.

Common normal forms used in database design include:

1. **First Normal Form (1NF)**: Ensures each table column contains atomic (indivisible) values and each record is unique.
2. **Second Normal Form (2NF)**: Achieved when a database is in 1NF and all non-key attributes are fully functionally dependent on the primary key.
3. **Third Normal Form (3NF)**: Achieved when a database is in 2NF and all the attributes are functionally dependent only on the primary key, removing transitive dependency.
4. **Boyce-Codd Normal Form (BCNF)**: A stricter version of 3NF, where every determinant is a candidate key.
5. **Fourth Normal Form (4NF)**: Ensures no multi-valued dependencies other than a candidate key exist.

These normal forms help create a robust, efficient, and easily manageable database system.