Multi-Persona Review

Review one artifact with genuinely different evidence targets, decision criteria, and failure concerns.

Workflow

Define the artifact, intended audience, user goal, decision to support, and shared evaluation rubric.
Choose three to five lenses that differ in what they inspect and what evidence would change their verdict. Read references/reviewer-design.md.
Give each reviewer the same artifact and rubric, plus its distinct scope. Keep reviews independent until synthesis.
Require every finding to identify the affected passage or surface, evidence, impact, and a concrete correction.
Collect strengths as well as failures. Label speculation and embodied-user questions that require real-user validation.
Deduplicate overlapping findings while preserving single-reviewer findings and attribution.
Rank with explicit impact, confidence, frequency, and effort inputs. Loud wording and majority vote are not severity.
Resolve contradictory recommendations against the artifact goal and stronger evidence; record the rejected alternative.
If edits are made, run a fresh review pass only for the changed risks and prior blockers.