OKR & Outcome Architecture

This skill exists to stop: projects and initiatives disguised as Key Results — teams celebrating shipped work that moved no business outcome.

🤖 0. HOW TO USE (agent workflow)

A. AUDIT an existing OKR set ("review these OKRs"):

1. Run the 2 sanity tests (§1) on the set as a whole (all KRs vs the Objective).
2. Grade each Objective against the 4 rules (§2), each KR against the 4 parts + 2 diagnostic tests (§3), scan the 4 metric traps.
3. Check priorities: volume limits (skip if a single Objective), Committed/Stretch labels, ranking, owners (§4).
4. Output: one ❌ current | ⚠️ why it fails | ✅ proposed fix table, one row per issue, + the 60-second checklist result (§7). No generic praise — every remark points at a specific line.

B. WRITE a new OKR set ("help me set this quarter's OKRs"):

1. Ask before writing: what's the strategy / why-now? Which business outcome? Which customer outcome does it serve, and is that outcome underserved? (No answer → jtbd/opportunity-map first, §3b.) Which metrics HAVE baselines? (No baseline → the KR can't be written yet — go measure first.)
2. Draft per §2–§3; every Objective gets ≥1 lagging + 1–2 leading metrics.
3. Self-run the 60-second checklist before presenting.

C. MAP roadmap ↔ OKRs ("which KR does this work serve?"): build the OKR–Initiative matrix (§5), answer its three questions, label non-OKR work explicitly.