setting-okrs-goals（设定 OKRS 目标）

Setting OKRs & Goals

Scope

Covers

• Turning strategy (or a North Star) into a small set of team/company OKRs
• Writing objectives that drive weekly execution (not just aspirational statements)
• Designing robust key results (prefer absolute counts; guard against gaming)
• Adding “default-on” systems/habits that make progress inevitable
• Defining review cadence + end-of-cycle grading to create a learning loop

When to use

• “Set our Q2 OKRs.”
• “Write objectives and key results for this team.”
• “We need quarterly goals that actually change behavior week-to-week.”
• “Our metrics are getting gamed / teams are optimizing the wrong thing.”
• “We need an OKR review + grading process.”

When NOT to use

• You don’t have an agreed strategy/North Star at all (use writing-north-star-metrics or defining-product-vision first)
• You need sprint planning or a delivery plan (tickets, estimates, timelines)
• You’re using OKRs primarily for individual performance evaluation
• You only need a single experiment metric for one test
• You need an analytics/event tracking implementation plan from scratch

Inputs

Minimum required

• Planning cycle + horizon (e.g., Q2; annual; 6 weeks) and the team(s) in scope
• Strategy anchor: company goal, North Star, or “why now” narrative for the cycle
• Current baseline for key metrics (or best-available proxy) + where the numbers come from
• Constraints: capacity, must-do commitments, dependencies, risk tolerance
• Stakeholders: decider(s), contributors, approvers, review cadence participants

Missing-info strategy

• Ask up to 5 questions from references/INTAKE.md.
• If still missing, proceed with clearly labeled assumptions and provide 2–3 OKR set options (conservative/base/ambitious).

Outputs (deliverables)

Produce an OKR & Goals Pack in Markdown (in-chat; or as files if the user requests), in this order:

1. Context snapshot (strategy anchor, horizon, scope, constraints, stakeholders)
2. Alignment map (company goal → team objective(s), no more than one step away)
3. Draft OKRs (1–3 Objectives; 2–5 Key Results each) with metric definitions, baselines, targets, owners, cadence
4. Metric robustness + guardrails (anti-gaming checks; ratio/denominator rules; quality guardrails)
5. Systems & habits plan (“default-on” behaviors/processes that make progress recurring)
6. Review + grading plan (weekly check-in; mid-cycle checkpoint; end-of-cycle scoring + learning retro)
7. Risks / Open questions / Next steps (always included)

Templates: references/TEMPLATES.md

Workflow (8 steps)

1) Intake + decision framing

• Inputs: User context; references/INTAKE.md.
• Actions: Confirm horizon, scope, strategy anchor, baseline availability, constraints, and decision-maker(s).
• Outputs: Context snapshot.
• Checks: Everyone agrees what OKRs are for (alignment + learning), and what they are not (performance evaluation).

2) Establish alignment (“one step away”)

• Inputs: Strategy anchor; current company goal/North Star.
• Actions: Write a one-sentence company goal for the cycle; map each proposed team objective to it (no deep cascading).
• Outputs: Alignment map.
• Checks: For every team objective, you can answer: “How does this move the company goal within this horizon?”

3) Draft 1–3 Objectives (outcome-first)

• Inputs: Alignment map; key problems/opportunities.
• Actions: Draft objectives as outcomes + intent (not projects). Keep the set small.
• Outputs: Objective list with short rationale (“why now / why this”).
• Checks: An objective can be understood without reading its KRs; it changes what the team prioritizes weekly.

4) Generate candidate KRs (robust, measurable)

• Inputs: Objectives; baselines (or proxies).
• Actions: Draft 2–5 KRs per objective; define baseline, target, time window, metric owner, and data source. Prefer absolute metrics; if you use a ratio, also include its absolute numerator/denominator KRs or guardrails.
• Outputs: KR table(s) with metric definitions.
• Checks: Two analysts would compute the same number; targets are directionally ambitious but not fantasy.

5) Add systems/habits (default-on execution)

• Inputs: OKRs draft; team operating model.
• Actions: Specify the recurring mechanisms that will produce progress (cadences, routines, gates, customer touchpoints), not just one-off initiatives.
• Outputs: Systems & habits plan.
• Checks: At least one “default-on” system exists per objective, with an owner and cadence.

6) Anti-gaming + guardrails

• Inputs: KRs + systems plan.
• Actions: Identify how each KR could be gamed or cause harm. Add guardrails (quality, trust, margin, volume) and ratio/denominator checks.
• Outputs: Guardrails section + anti-gaming notes per KR.
• Checks: You can name 1–2 failure modes per KR and how you’ll detect them early.

7) Review cadence + grading plan (learning loop)

• Inputs: Full draft OKRs + guardrails.
• Actions: Define weekly review format, mid-cycle checkpoint rules, and end-of-cycle grading (scoring + retrospective questions).
• Outputs: Review + grading plan.
• Checks: The plan produces learning, not blame; it specifies who reviews, when, and what decisions can change mid-cycle.

8) Quality gate + finalize the pack

• Inputs: Entire OKR & Goals Pack.
• Actions: Run references/CHECKLISTS.md and score with references/RUBRIC.md. Add Risks/Open questions/Next steps.
• Outputs: Final OKR & Goals Pack.
• Checks: Pack is shareable as-is; alignment, metrics, guardrails, and cadence are unambiguous.

Quality gate (required)

• Use references/CHECKLISTS.md and references/RUBRIC.md.
• Always include: Risks, Open questions, Next steps.

Examples

Example 1 (B2B SaaS): “Set Q2 OKRs for Activation to improve time-to-first-value for new teams.” Expected: 1–2 objectives focused on new-team success, KRs with baselines/targets, a weekly review cadence, and guardrails (e.g., support tickets/new team).

Example 2 (Growth): “Set quarterly OKRs for Growth; we keep arguing about conversion rate vs volume.” Expected: KRs expressed as absolute numbers (e.g., activated users) plus denominator/quality guardrails to prevent ‘ratio gaming’.

Boundary example: “Write OKRs, but we don’t have a company goal or baseline metrics.” Response: ask for the minimum strategy anchor + baselines; if unavailable, produce 2–3 draft OKR options with explicit assumptions and recommend doing North Star/vision first.