No measured KPI outcomes are currently stored in this repository. All metrics below are PLANNED and should be treated as placeholders for future implementation.
- Definition: % of reports where user and reviewer agree the recommendation is understandable.
- How to measure: post-report survey (Yes/No or 1–5 scale converted to pass threshold).
- What good looks like (PLANNED target): >= 85% clarity.
- Definition: % of cases requiring full re-run due to missing/ambiguous intake data.
- How to measure: count cases tagged
iteration-requiredbefore first usable recommendation. - What good looks like (PLANNED target): <= 25% after intake form refinement.
- Definition: % of reports that include explicit rationale for recommendation and elimination.
- How to measure: checklist-based QA on report sections.
- What good looks like (PLANNED target): 100% of reports include rationale blocks.
- Definition: % of draft recommendations changed at human review step.
- How to measure: compare draft vs approved recommendation labels.
- What good looks like (PLANNED target): 10-30% (healthy oversight, not blind acceptance).
- Definition: median elapsed time from completed intake to approved decision report.
- How to measure: timestamp intake completion and review approval.
- What good looks like (PLANNED target): <= 2 business days for standard cases.