In order for tests to run quickly and provide constant feedback, code needs to be organized in such a way that the methods, functions, and classes can be easily replaced with mocks and stubs. A common word for this type of replacements of the actual code is test double. The speed of execution can be severely affected with external dependencies; for example, our code might need to communicate with the database. By mocking external dependencies, we are able to increase that speed drastically. Whole unit test suite execution should be measured in minutes, if not seconds. Designing the code in a way that can be easily mocked and stubbed forces us to structure that code better by applying a separation of concerns.
More important than speed is the benefit of the removal of external factors. Setting up databases, web servers, external APIs, and other dependencies that our code might need, is both time consuming and unreliable. In many cases, those dependencies might not even be available. For example, we might need to create a code that communicates with a database and have someone else create a schema. Without mocks, we would need to wait until that schema is set.