Automated testing does not replace manual testing, its a common misconception. the two testing approaches should complement each other, as they serve for different purposes, covering every aspect what needs to be tested. what is important: there is no automated testing without manual testing, as automated scripts shoudl always be superwised and calibrated. Not everything should be automated, but what can be, maybe should be.
what kind of tests are good candidates for automation:
- repetitive tests: where the same tests needs to be run frequently
- tests where the functionality being tested does not change often
- tests where the expected outcome is clear and easy to verify
- tests what would consume a lot of time run manually
- high risk tests, where human error have serious consequences
test types that are most often automated:
- Regression tests
- Smoke tests
- Performance/load tests
- API tests
- Data-driven test cases
- Cross-browser tests
- CI/CD pipeline checks
what should not be automated:
- One-time or rarely-run tests
- Exploratory testing (where human judgment is needed)
- Visual and UI/UX testing that depends on perception or “feel”
- Tests for rapidly-changing features that would constantly break the automation
- Very short-lived features or experimental code
Examples best left to manual testing:
- Usability testing
- First-time setup and user flow testing
- Accessibility and screen reader testing
- Non-deterministic behavior (randomized outcomes, timing-sensitive features)
benefits | description |
|---|---|
automated tests run much faster and with better consistency than manual tests | |
human errors are eliminated | |
easy to run hundreds or thousands of tests at the same time | |
you can keep testing old features while working on new ones | |
excelletnt for CI/CD pipelines |
limitations | description |
|---|---|
writing automated tests requires time and skills | |
tests can break when the product changes, continous calibrations are needed | |
unstable or wrong tests can cause false positive/negative results | |
often used to replace manual testing, which is a risk |