writing-north-star-metrics
Compare original and translation side by side
🇺🇸
Original
English🇨🇳
Translation
ChineseWriting North Star Metrics
撰写北极星指标
Scope
范围
Covers
- Defining or refreshing a product/company North Star and North Star Metric
- Translating a qualitative value model into measurable, decision-useful metrics
- Creating a simple driver tree: leading input/proxy metrics + guardrails
- Producing a “North Star Metric Pack” teams can use as a decision tie-breaker
When to use
- “We need one metric that defines success.”
- “Teams are optimizing different KPIs.”
- “We’re setting quarterly OKRs and need leading indicators.”
- “We’re launching a new strategy and need a metric that aligns decisions.”
When NOT to use
- You only need OKRs for an already-agreed North Star
- You need a full analytics taxonomy/event tracking plan from scratch
- Stakeholders haven’t aligned on the customer value model / mission at all (do product vision/strategy first)
- You’re choosing a single experiment metric for a one-off test
涵盖内容
- 定义或更新产品/公司北极星及北极星指标
- 将定性价值模型转化为可衡量、对决策有帮助的指标
- 创建简洁的驱动树:前置输入/代理指标 + 约束规则
- 生成“北极星指标包”,供团队用作决策的最终依据
适用场景
- “我们需要一个定义成功的核心指标。”
- “各团队在优化不同的KPI。”
- “我们正在制定季度OKR,需要前置性指标。”
- “我们正在推出新战略,需要一个能统一决策的指标。”
不适用场景
- 仅需为已达成共识的北极星制定OKR
- 需要从零开始构建完整的分析分类/事件追踪计划
- 利益相关方尚未就客户价值模型/使命达成一致(应先制定产品愿景/战略)
- 为单次测试选择单一实验指标
Inputs
输入项
Minimum required
- Product/company + primary customer segment
- The “value moment” (what the customer gets when things go well)
- Business model + strategic goal (growth, activation, retention, margin, trust, etc.)
- Time horizon (next quarter vs next year)
- Measurement constraints (what you can measure today; data latency; known gaps)
Missing-info strategy
- Ask up to 5 questions from references/INTAKE.md.
- If still missing, proceed with clearly labeled assumptions and provide 2–3 options.
最低要求输入项
- 产品/公司 + 核心客户群体
- “价值时刻”(当一切顺利时,客户获得的价值)
- 商业模式 + 战略目标(增长、激活、留存、利润率、信任等)
- 时间范围(下一季度 vs 下一年)
- 测量约束(当前可测量的内容;数据延迟;已知的数据缺口)
缺失信息处理策略
- 从references/INTAKE.md中最多提出5个问题。
- 如果仍有缺失,基于明确标注的假设继续推进,并提供2–3个备选方案。
Outputs (deliverables)
输出成果(交付物)
Produce a North Star Metric Pack in Markdown (in-chat; or as files if the user requests):
- North Star Narrative (value model, tie-breaker, scope)
- Candidate metrics (3–5) + selection rationale (evaluation table)
- Chosen North Star Metric spec (definition, formula, window, segmentation, owner, data source)
- Driver tree (leading input/proxy metrics + guardrails)
- Validation & rollout plan (instrumentation checks, dashboard cadence, decision rules)
- Risks / Open questions / Next steps (always included)
Templates: references/TEMPLATES.md
生成Markdown格式的北极星指标包(可在对话中直接提供;若用户要求,也可作为文件交付):
- 北极星说明文档(价值模型、决策依据、适用范围)
- 候选指标(3–5个) + 选择理由(评估表格)
- 选定的北极星指标规范(定义、计算公式、统计周期、细分维度、负责人、数据源)
- 驱动树(前置输入/代理指标 + 约束规则)
- 验证与推广方案(instrumentation检查、仪表盘更新频率、决策规则)
- 风险/待解决问题/下一步计划(必须包含)
模板:references/TEMPLATES.md
Workflow (8 steps)
工作流程(8个步骤)
1) Intake + constraints
1) 信息收集与约束确认
- Inputs: User context; use references/INTAKE.md.
- Actions: Confirm product, customer, value moment, horizon, constraints, stakeholders.
- Outputs: 5–10 bullet “Context snapshot”.
- Checks: You can explain the customer value in one sentence.
- 输入项: 用户提供的背景信息;参考references/INTAKE.md。
- 行动: 确认产品、客户、价值时刻、时间范围、约束条件、利益相关方。
- 输出项: 5–10条“背景快照”要点。
- 检查: 能用一句话解释客户价值。
2) Define the qualitative North Star (before numbers)
2) 定义定性北极星(先于量化指标)
- Inputs: Context snapshot.
- Actions: Write a North Star statement and value model from the customer’s perspective.
- Outputs: Draft North Star Narrative (template in references/TEMPLATES.md).
- Checks: Narrative can act as a decision tie-breaker (“if we do X, does it move the North Star?”).
- 输入项: 背景快照。
- 行动: 从客户视角撰写北极星声明和价值模型。
- 输出项: 草稿版北极星说明文档(模板见references/TEMPLATES.md)。
- 检查: 该说明文档可作为决策的最终依据(“如果我们执行X,是否会影响北极星指标?”)。
3) Generate 3–5 candidate North Star metrics (customer POV)
3) 生成3–5个候选北极星指标(客户视角)
- Inputs: North Star Narrative + value moment.
- Actions: Propose metrics that measure delivered customer value (not internal activity). Include at least one “friction/absence of pain” option when relevant.
- Outputs: Candidate list with definitions.
- Checks: Each candidate is measurable, understandable, and not trivially gameable.
- 输入项: 北极星说明文档 + 价值时刻。
- 行动: 提出衡量客户交付价值的指标(而非内部活动指标)。若相关,至少包含一个“减少摩擦/消除痛点”的选项。
- 输出项: 带定义的候选指标列表。
- 检查: 每个候选指标均可测量、易于理解,且不会被轻易操纵。
4) Stress-test and pick the North Star metric
4) 压力测试并选定北极星指标
- Inputs: Candidate metrics.
- Actions: Evaluate with references/CHECKLISTS.md and references/RUBRIC.md. Explicitly test:
- Leading vs lagging (avoid “retention as the only goal”; pair lagging outcomes with controllable inputs)
- Controllability within a quarter (proxy/input metrics you can move)
- Ecosystem impact (what breaks if you optimize this?)
- Outputs: Selection table + chosen metric + why others lost.
- Checks: A cross-functional leader could agree/disagree based on definitions and evidence.
- 输入项: 候选指标。
- 行动: 参考references/CHECKLISTS.md和references/RUBRIC.md进行评估。需明确测试:
- 前置性vs滞后性(避免“仅以留存为目标”;将滞后性结果指标与可控的输入指标结合)
- 季度内的可控性(可在数周/数月内影响的代理/输入指标)
- 生态系统影响(若优化该指标,会有哪些环节受影响?)
- 输出项: 选择表格 + 选定指标 + 淘汰其他指标的理由。
- 检查: 跨职能负责人可依据定义和证据表示同意或反对。
5) Write the metric spec (make it unambiguous)
5) 撰写指标规范(确保无歧义)
- Inputs: Chosen metric.
- Actions: Define formula, unit, window, inclusion rules, segmentation, owner, source, latency, and example calculation.
- Outputs: North Star Metric Spec.
- Checks: Two analysts would compute the same number.
- 输入项: 选定的指标。
- 行动: 定义计算公式、单位、统计周期、纳入规则、细分维度、负责人、数据源、数据延迟及示例计算。
- 输出项: 北极星指标规范。
- 检查: 两位分析师计算出的结果一致。
6) Build the driver tree (inputs + guardrails)
6) 构建驱动树(输入项 + 约束规则)
- Inputs: Metric spec + product levers.
- Actions: Decompose into 3–7 drivers; identify leading input/proxy metrics you can move in weeks/months; add guardrails to prevent gaming/harm.
- Outputs: Driver tree table + guardrails list.
- Checks: Every driver has at least 1 realistic lever (initiative/experiment) and 1 measurement.
- 输入项: 指标规范 + 产品杠杆。
- 行动: 分解为3–7个驱动因素;确定可在数周/数月内影响的前置输入/代理指标;添加约束规则以防止操纵/损害。
- 输出项: 驱动树表格 + 约束规则列表。
- 检查: 每个驱动因素至少有1个可行的杠杆(举措/实验)和1个测量方式。
7) Define validation + rollout
7) 定义验证与推广方案
- Inputs: Driver tree + constraints.
- Actions: Plan validation (sanity checks, correlation to outcomes) and operationalization (dashboards, cadence, owners, decision rules).
- Outputs: Validation & Rollout Plan.
- Checks: Plan includes “who does what, when” and works with current instrumentation.
- 输入项: 驱动树 + 约束条件。
- 行动: 规划验证(合理性检查、与结果的相关性)和落地实施(仪表盘、更新频率、负责人、决策规则)。
- 输出项: 验证与推广方案。
- 检查: 方案包含“谁在何时做什么”,且适配当前的instrumentation能力。
8) Quality gate + finalize pack
8) 质量校验与最终定稿
- Inputs: All drafts.
- Actions: Run references/CHECKLISTS.md and score with references/RUBRIC.md. Add Risks/Open questions/Next steps.
- Outputs: Final North Star Metric Pack.
- Checks: Pack is shareable as-is; key decisions and caveats are explicit.
- 输入项: 所有草稿。
- 行动: 参考references/CHECKLISTS.md并使用references/RUBRIC.md评分。添加风险/待解决问题/下一步计划。
- 输出项: 最终版北极星指标包。
- 检查: 指标包可直接共享;关键决策和注意事项均明确标注。
Quality gate (required)
质量校验(必填)
- Use references/CHECKLISTS.md and references/RUBRIC.md.
- Always include: Risks, Open questions, Next steps.
- 使用references/CHECKLISTS.md和references/RUBRIC.md。
- 必须包含:风险、待解决问题、下一步计划。
Examples
示例
Example 1 (B2B SaaS): “Define a North Star metric for a team collaboration tool.”
Expected: a pack that chooses a customer-value metric (e.g., weekly active teams completing the core value moment), plus a driver tree (activation → collaboration depth) and guardrails.
Expected: a pack that chooses a customer-value metric (e.g., weekly active teams completing the core value moment), plus a driver tree (activation → collaboration depth) and guardrails.
Example 2 (Marketplace): “Refresh North Star metric for a local services marketplace.”
Expected: a pack that measures delivered value (e.g., successful jobs completed with quality), plus input metrics for supply/demand balance and quality guardrails.
Expected: a pack that measures delivered value (e.g., successful jobs completed with quality), plus input metrics for supply/demand balance and quality guardrails.
Boundary example: “Our North Star should be retention.”
Response: keep retention as an outcome/validation metric, and propose controllable input/proxy metrics (time-to-first-value, weekly value moments, repeat value delivery) as the operating focus.
Response: keep retention as an outcome/validation metric, and propose controllable input/proxy metrics (time-to-first-value, weekly value moments, repeat value delivery) as the operating focus.
示例1(B2B SaaS): “为团队协作工具定义北极星指标。”
预期输出:一个指标包,选定以客户价值为核心的指标(例如,每周完成核心价值时刻的活跃团队数),并附带驱动树(激活→协作深度)和约束规则。
预期输出:一个指标包,选定以客户价值为核心的指标(例如,每周完成核心价值时刻的活跃团队数),并附带驱动树(激活→协作深度)和约束规则。
示例2(平台型产品): “为本地服务平台更新北极星指标。”
预期输出:一个指标包,衡量交付的价值(例如,高质量完成的服务订单数),并附带供需平衡的输入指标和质量约束规则。
预期输出:一个指标包,衡量交付的价值(例如,高质量完成的服务订单数),并附带供需平衡的输入指标和质量约束规则。
边界示例: “我们的北极星应该是留存率。”
回应:将留存率作为结果/验证指标,同时提出可控的输入/代理指标(首次价值获取时长、每周价值时刻、重复价值交付)作为运营核心。
回应:将留存率作为结果/验证指标,同时提出可控的输入/代理指标(首次价值获取时长、每周价值时刻、重复价值交付)作为运营核心。