Testability defined
Testability is the degree to which a system makes it practical to create a known condition, perform an action, observe the result, and diagnose unexpected behaviour.
Poor testability makes testing slower, more fragile, and more dependent on manual investigation.
Six dimensions
| Dimension | Good testability |
|---|---|
| Observability | Important states, events, errors, and outputs can be inspected. |
| Controllability | Tests can create required states and inputs. |
| Isolation | Dependencies can be stubbed, mocked, or independently exercised. |
| Repeatability | The same test can run again under comparable conditions. |
| Determinism | Results are not dominated by random timing or shared state. |
| Diagnosability | Failures produce useful evidence. |
Poor vs strong testability
Poor
User Action → Opaque System → Failure → Manual Guessing
Strong
Controlled State → Stable Interface → Observable Event/Log → Expected State → Actionable Failure Evidence
Design choices that help
- Stable test IDs
- Accessible APIs
- Structured logs
- Correlation IDs
- Feature flags
- Seedable data
- Dependency simulation
- Health endpoints
- Configurable timeouts
- Repeatable reset
Testability reviews
Ask how the required state will be created, how results will be observed, how failures can be simulated, and how tests can repeat without manual cleanup.
Often the cheapest automation improvement is a product testability improvement.