Write Key Results

Write Key Results

When to use

When KRs need the outcome test — measurable change in an indicator, not confirmation a deliverable shipped.

What this skill does not do

• Does not draft objectives — route to /okr:draft-objectives.
• Does not set targets — route to /okr:set-targets after baselines known.
• Does not instrument metrics — route unknown baselines to /okr:instrument-metrics.

Preconditions

| Input | If missing | |---|---| | Objective the KR set serves | Ask — every check depends on it | | Candidate KRs (or request to draft) | Ask user to provide or draft from objective |

Provisional mode

Without baseline data: flag Baseline status: unknown per KR; do not imply targets are ready.

Trust spine

• Confidence bands (structured-aggregation):
• High: Outcome-shaped KRs, vanity risks named, set collectively sufficient.
• Medium: Some output-shaped flags with reframes.
• Low: Majority deliverable-shaped — set not ready for targets.
• Failure modes:
• Strategic advice vs. support: Verdicts diagnostic; strategist owns final KRs.
• Client confidentiality: KRs may be pre-release — CONFIDENTIAL header.
• Accountability gap: Deliverables relocated to tasks, not silently kept as KRs.
• Analytical Rigor: Set-level sufficiency check — hitting all KRs proves objective?
• Incentive Gaming: Vanity-metric and sandbagging-by-proxy checks — easy-to-hit numbers that don't prove objective.
• Escalation triggers: Set insufficient even if every KR passes individually — name gap.

Workflow

1. Get the objective — explicit, not implied.
2. Outcome test per KR — deliverable language ("ship," "launch") → reframe to outcome.
3. Vanity-metric check — would unrelated improvement fool readers?
4. Baseline check — unknown → flag for instrument-metrics.
5. Set sufficiency — would hitting all KRs prove the objective?
6. Completeness check before output.

Output format

CONFIDENCE: [defensible recommendation | structured first pass]
LOAD-BEARING ASSUMPTIONS: [if any]
OBJECTIVE: [text]

KR 1: [text]
  Verdict: [PASS | FLAG — output-shaped | FLAG — vanity metric risk]
  Baseline status: [known | unknown — flag for instrument-metrics]

[repeat per KR]

SET-LEVEL CHECK: [would hitting all KRs prove the objective? any gap?]

Worked example

Input: Objective "Improve enterprise retention." KR "Ship onboarding v2."

Expected output (excerpt):

KR 1: Ship onboarding v2
  Verdict: FLAG — output-shaped; suggested outcome: 90-day logo retention ≥ 92%
  Baseline status: unknown — flag for instrument-metrics

SET-LEVEL CHECK: insufficient — single output KR doesn't prove retention objective

Quality checks before delivering

• [ ] Objective explicit at top
• [ ] Outcome test on every KR
• [ ] Vanity-metric check run
• [ ] Baseline status per KR
• [ ] Set-level sufficiency assessed
• [ ] Figures source-tagged or omitted

Propose profile update

When a stable convention surfaces during this run (thresholds, naming, tone, output format, or recurring corrections), propose a profile update: show the exact diff against ~/.claude/plugins/config/claude-for-strategy/okr/CLAUDE.md (org-wide facts go to org-profile.md), ask for confirmation, and write only on yes. Only /okr:practice-setup auto-applies a full profile write.

Outputs

Follows plugin CLAUDE.md § Outputs. Next: instrument-metrics, set-targets, or revise flagged KRs.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: daddia
• Source: daddia/claude-for-strategy
• License: MIT

Install and usage instructions live in the source repository linked above.