The most common approach to unit testing is "one class = one unit": for each class, you write a unit test.
But this approach lacks robustness. It produces brittle tests which need to be rewritten when you refactor. It encourages developers to simply rewrite their code as a test, meaning their tests mirror code bugs. And over-reliance on mocks/fakes significantly limits the breadth of the test.

This talk presents a different approach. By focusing on user interactions, not code units, we can optimise our testing and largely solve these problems.

Through concrete, practical examples we will learn:
- How to write robust tests which don't need rewriting when you refactor
- How to tailor tests to actions users perform, not specific code implementations
- Why fakes are often better than mocks