Ranked by priority, with a receipt
The ranked board — every request with its score, tier, and duplicate flag.
These requests can't be honestly scored yet — they're missing material fields or ground to more than one meaning. The engine returns REVIEW or REFUSE with the exact gaps, instead of guessing a number.
- ›product area — Request touches multiple product areas; cannot bind one.
- → A human reviewer can disambiguate the flagged field(s); re-ground once a single value is chosen.
The refusal is receipted too
A refusal is a first-class, hash-chained outcome — the decision not to fabricate a score is itself on the ledger, re-provable byte-for-byte.
- ›persona — Request names multiple personas; cannot bind one.
- → A human reviewer can disambiguate the flagged field(s); re-ground once a single value is chosen.
The refusal is receipted too
A refusal is a first-class, hash-chained outcome — the decision not to fabricate a score is itself on the ledger, re-provable byte-for-byte.
- ›delivery confidence — Delivery confidence not provided (0.5 / 0.8 / 1.0).
- ›effort band — No effort signal — cannot size build cost without guessing.
- ›impact band — No impact signal — cannot size business impact without guessing.
- ›reach users per quarter — No reach figure — cannot compute how many users are affected.
- ›value confidence — Value confidence not provided (0.5 / 0.8 / 1.0).
- → Provide the missing field(s) on the request, then re-run.
- → Do not estimate the missing value — source it or leave the request in REVIEW.
The refusal is receipted too
A refusal is a first-class, hash-chained outcome — the decision not to fabricate a score is itself on the ledger, re-provable byte-for-byte.
- ›effort band — No effort signal — cannot size build cost without guessing.
- ›impact band — No impact signal — cannot size business impact without guessing.
- ›persona — Could not identify which user/persona this serves.
- ›product area — Could not bind the request to any governed product area.
- ›reach users per quarter — No reach figure — cannot compute how many users are affected.
- → Provide the missing field(s) on the request, then re-run.
- → Do not estimate the missing value — source it or leave the request in REVIEW.
The refusal is receipted too
A refusal is a first-class, hash-chained outcome — the decision not to fabricate a score is itself on the ledger, re-provable byte-for-byte.
- ›delivery confidence — Delivery confidence not provided (0.5 / 0.8 / 1.0).
- ›reach users per quarter — No reach figure — cannot compute how many users are affected.
- ›value confidence — Value confidence not provided (0.5 / 0.8 / 1.0).
- → Provide the missing field(s) on the request, then re-run.
- → Do not estimate the missing value — source it or leave the request in REVIEW.
The refusal is receipted too
A refusal is a first-class, hash-chained outcome — the decision not to fabricate a score is itself on the ledger, re-provable byte-for-byte.
How this board is built
Every row is ranked by the deterministic priority_score — a 0–100 log-map of reach × impact × confidence ÷ effort, weighted by urgency — computed by the governed model, not by votes. Near-duplicates are collapsed by meaninginto one cluster row, represented by the highest-scoring member; expand a clustered row to see every member and the engine's keep / hold call. The tier buckets are a reading lens over the same score:
The ledger verdict is PASS across 37 entries — every scored, reviewed, and refused outcome is hash-chained. Re-run the corpus and every score, cluster, and hash on this board is identical.