writing-north-star-metrics（编写北极星指标）

Writing North Star Metrics

Scope

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 -> use setting-okrs-goals
• You need a full analytics taxonomy/event tracking plan from scratch
• Stakeholders haven’t aligned on the customer value model / mission at all -> use defining-product-vision first
• You’re choosing a single experiment metric for a one-off test
• You need to diagnose retention or engagement patterns, not define the top-level metric -> use retention-engagement
• You need to assess whether you have product-market fit -> use measuring-product-market-fit

Inputs

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.

Outputs (deliverables)

Produce a North Star Metric Pack in Markdown (in-chat; or as files if the user requests):

1. North Star Narrative (value model, tie-breaker, scope)
2. Candidate metrics (3–5) + selection rationale (evaluation table)
3. Chosen North Star Metric spec (definition, formula, window, segmentation, owner, data source)
4. Driver tree (leading input/proxy metrics + guardrails)
5. Validation & rollout plan (instrumentation checks, dashboard cadence, decision rules)
6. Risks / Open questions / Next steps (always included)

Templates: references/TEMPLATES.md

Workflow (8 steps)

1) Intake + constraints

• 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.

2) Define the qualitative North Star (before numbers)

• 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?”).

3) Generate 3–5 candidate North Star metrics (customer POV)

• 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.

4) Stress-test and pick the North Star metric

• 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.

5) Write the metric spec (make it unambiguous)

• 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.

6) Build the driver tree (inputs + guardrails)

• 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.

7) Define validation + rollout

• 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.

8) Quality gate + finalize pack

• 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.

Quality gate (required)

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

Examples

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.

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.

Boundary example (redirect): “We already have our North Star metric. Now set quarterly OKRs and key results for the team.” Response: redirect to setting-okrs-goals -- this request needs OKR design from an existing North Star, not metric definition work.

Boundary example (reframe): “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.

Anti-patterns

Avoid these common failure modes when defining North Star metrics:

1. Revenue-as-North-Star -- Choosing revenue or profit as the North Star metric. Revenue is a trailing indicator of value delivery; it cannot be a decision tie-breaker for product teams. Use a customer-value metric that leads to revenue.
2. Vanity volume metric -- Picking a metric that only goes up (total users, total messages sent) without a quality or frequency dimension. Always pair volume with a quality or engagement signal.
3. Uncontrollable lagging metric -- Selecting a metric no team can move within a quarter. The North Star must decompose into leading input metrics with realistic levers.
4. Driver tree without levers -- Building a decomposition tree where drivers are described as metrics but no team has a concrete initiative or experiment to move them. Every driver needs at least one actionable lever.
5. Metric without a spec -- Agreeing on a metric name (“weekly active teams”) without defining formula, window, inclusion rules, and segmentation. Two analysts should compute the same number.