Dimensions of Test Categorization

Software tests are often given labels such as “unit,” “smoke,” and “regression.” These labels come from different, independent dimensions — distinct ways to classify a test. Within a single dimension, categories are mutually exclusive (for example, unit vs. end‑to‑end), but categories from different dimensions can be combined (for example, an end‑to‑end regression test). This page summarizes common dimensions and representative categories.

By system granularity

Unit test: Exercises a small unit in isolation.
Integration test: Verifies interactions between components.
End‑to‑end (E2E) test: Tests the entire system from the user’s perspective.

By scope of impact

Regression: Verifies existing behavior still works after changes.
New or modified feature: Validates newly added or changed functionality (including bug fixes).

By breadth of coverage

Smoke / sanity: Quick check of critical paths.
Targeted / feature: Selected features or subsystems.
Comprehensive / full / system: Broad coverage of all features.

By execution method

Manual testing.
Automated testing.

By quality attribute

Functional testing.
Performance testing.
Security testing.
Usability testing.
Reliability and resilience testing.

Example: An automated, end‑to‑end test to verify that performance has not degraded spans four dimensions at once — E2E, regression, automated, and performance.