discovery-review-criteria

Gate definitions and verdict rubric for the DISCOVER reviewer — applied across 3 lenses (completeness, prioritization, duplicate detection) with 6 gates (G1–G6).

When to use

Entry contract

Exit contract

Invariants

Gate Lens Severity Definition
G1 Completeness HIGH All 3 discovery modes applied, or skipped modes justified
G2 Completeness BLOCKER No open P0/P1 issue absent from the report (top-5 sample-check)
G3 Prioritization HIGH No priority inversions; every P0 has a written justification
G4 Prioritization HIGH Effective capacity respected; no XL issue in the sprint
G5 Duplicates HIGH No undetected duplicate pair (normalised similarity > 80%)
G6 Duplicates MEDIUM All 40–80% pairs flagged in the “Duplicates Detected” section

Similarity thresholds:

Level Similarity Action required
EXACT > 95% One issue labelled status/duplicate linked to the original
NEAR 80–95% Merge recommendation; both issues linked and documented
RELATED 40–80% Flagged as “related”, recommendation documented
DIFFERENT < 40% No action required

Why this shape

Discovery is the least visible phase but the most expensive when it fails: a P0 issue absent from the sprint becomes a production incident or critical debt. Gates G2 and G4 protect against the two most frequent failures — invisible critical issues and over-loaded sprints.

« Maximizing the amount of work not done is essential. » — Reinertsen, D. G., The Principles of Product Development Flow, 2009.

Allowed customisation

See also