The better automation question
Do not start with “Can this be automated?” Start with “Will automating this create enough repeated value to justify building and maintaining it?”
Strong candidates: meaningful risk + repetition + objective result + stable interface + useful feedback.
Strong candidates
- Critical regression paths
- Smoke tests
- Stable APIs
- Data validation rules
- Repeated calculations
- Permission matrices
- Contracts
- Cross-browser checks
- Large combinations
- CI quality gates
Weak candidates
- One-time checks with little expected reuse
- Highly subjective usability judgments
- Rapidly changing prototypes
- Exploratory investigation
- Low-risk scenarios with expensive setup
Automation value model
| Factor | Question |
|---|---|
| Risk | What happens if the behaviour fails? |
| Frequency | How often will this run? |
| Stability | Will the interface remain reasonably stable? |
| Feedback | Does automation produce earlier or more frequent evidence? |
| Manual effort | How expensive is repeated manual execution? |
| Maintenance | How costly will the test be to keep healthy? |
Choose the right layer
Business workflow → selected E2E
UI rendering → UI / visual
API rule → API
Service contract → contract / integration
Calculation → component / unit
Data mapping → data validation
Measure meaningful coverage, such as P1 regression automation—not raw automated test count.