What is a Data Validation?

Validation rules typically check type (is this actually a number), range (is this date plausible), required presence (is this mandatory field populated), and referential integrity (does this foreign key actually point to a record that exists). The goal is to catch bad data at the boundary, as close to its point of entry as possible, rather than letting it flow downstream where it's far more expensive to trace and fix. There's an important distinction between validation and business-rule enforcement: validation checks that data is well-formed, while a separate layer of logic decides whether well-formed data represents something the system should actually accept - an email field can be validly formatted and still belong to a banned domain, for instance. A validation layer that only rejects bad data without explaining why creates its own problem: users and integrations need a specific, actionable error message (which field, what rule, what was expected) rather than a generic failure, or the invalid data just gets resubmitted unchanged.