← Control library
R01 · Reliability
Representative performance cases
A polished demo hides failure on real languages, users or operating conditions.
Requirement
Define a versioned test set covering normal, boundary and high-impact cases for the declared Swiss use, including supported languages.
Expected evidence
- Test-set rationale, size and composition.
- Per-case expected result and reviewed outcome.
Proposed verification
- Re-run a disclosed random sample.
- Compare failures by language, channel and high-impact subgroup.
Limit
A sample estimates behaviour only within its composition and confidence limits.