We could reframe the existing user-input work as an upstream coverage-design phase, with user inputs as one artifact, and connect it bidirectionally with failure understanding.
Role in Discovery and Delivery
The coverage method remains the same, but its source and purpose change.
| Context | Coverage starts from | Main purpose |
|---|---|---|
| Discovery | Provisional solution behaviour, feasibility and safety hypotheses | Expose whether the proposed behaviour is plausible and where it fails |
| Productization decision | Proposed production scope and known behavioural boundaries | Make residual behavioural uncertainty visible |
| Delivery development | Production Slice Contract and targeted change hypothesis | Compare a candidate with the baseline and detect regressions |
| Release | Committed behaviour, guarantees, invariants, production constraints | Produce release-relevant evidence |
| Production | Active product scope and defined live population | Observe real behaviour, outcomes, drift, and new failure conditions |
| Quality Understanding | Known failure modes, ambiguous cases, and coverage gaps | Refine cases, contrasts, criteria, and regression protection |
In Discovery, a small diagnostic or challenge set may be enough to answer a feasibility question.
In Delivery, the set becomes more stable because it must protect committed behaviour.
In production, the team also needs a declared sampling method. A curated offline case set and a production sample support different claims.