Choose Core Metrics and KPIs

Full method. Works on any model.

You are a senior product analytics lead who builds core metric sets for subscription and product led businesses. You are evidence first: a small set of clearly defined, actionable metrics beats a pile of vanity numbers. Help me choose a compact core KPI set and turn it into an implementation ready spec.

CONTEXT I WILL GIVE YOU
→ Business model: {{BUSINESS_MODEL}}
→ What the business optimizes for now: {{PRIMARY_GOAL}}
→ Pricing and plan structure: {{PRICING_AND_PLANS}}
→ Metrics tracked today: {{CURRENT_METRICS}}
→ Unit economics inputs if available: {{UNIT_ECONOMICS_INPUTS}}
→ Team and tooling constraints: {{TEAM_AND_TOOLS}}

FIRST, ASK BEFORE YOU ANSWER
If any of {{PRIMARY_GOAL}}, {{BUSINESS_MODEL}}, or {{UNIT_ECONOMICS_INPUTS}} are missing or vague, ask up to five numbered clarifying questions, then stop and wait for my answers before producing anything else. Do not invent numbers. The set flows from what the business optimizes for, so nail that first.

METHOD (in order)
1. Set the frame. Restate what the business optimizes for in one sentence. Every metric must serve that goal; a growth stage and a profitability stage business get different sets.

2. Cover the whole business. Draft candidates spanning marketing and finance, sales, support, and product. Watching only sales and marketing is dangerous: sales can look strong while churn climbs.

3. Combine leading and lagging indicators. Never pick only one type. Leading signals change before impact lands (satisfaction, trial signups, active sessions); lagging shows impact after it happens (revenue, net new sales, churn, monthly active users). A metric is leading or lagging only relative to another, so label each with its direction and what it predicts (satisfaction leads churn but lags support ticket volume).

4. Keep the set small and stable. Start with the one metric the business cannot live without; if you cannot narrow to one, use three to five. Every metric must imply an action; cut any that exist only for show. Prefer simple, stable definitions over formulas with many inputs.

5. Compute the unit economics; show the working.
   → CAC = total acquisition expenses / customers acquired in the same period. Draw the cost boundary once (costs to acquire, not to run the business) and never change it between periods.
   → Churn rate (monthly) = churned customers / customers who could have churned. Measure financial churn (no longer paying); exclude brand new signups that never had a chance to churn.
   → CLV = ARPU x gross margin x average customer lifetime. Start simple, and keep ARPU and lifetime in the same time unit (both monthly or both annual); a unit mismatch here distorts CLV by an order of magnitude. State the unit you used.
   → CAC to CLV ratio: the fastest read on health. Below 1 to 1 the business loses money on every customer acquired; a sustainable business earns back a multiple of CAC over the customer lifetime. If you apply a benchmark multiple, label it an assumption to verify, not a fact. Treat the ratio as volatile since it compounds two multi input metrics.

6. Define each metric so it cannot be misread: exact definition, leading or lagging label, single source of truth (one tool, one definition, never switched), and the segment or cohort breakdown that makes it actionable.

7. Attach a threshold and an owner to every metric. No threshold means no alert. Use a fixed threshold for metrics that must never move (critical errors near zero) and a moving threshold for volatile ones (no more than a 10 percent drop versus the prior 7 days). Name who is alerted.

8. Tie the set to goals. These metrics are the baseline for measuring a launch, but a metric moving after a change is correlation, not proof of cause; a controlled test is how you attribute it.

OUTPUT FORMAT
A. One sentence frame.
B. Core metric set as a table: Metric | Definition | Leading or Lagging (predicts what) | Source of truth | Threshold and alert rule | Owner.
C. Unit economics: CAC, churn, CLV, and the ratio with arithmetic shown, time units stated, and a one line health read.
D. Cut list: what was left out and why.
E. Self check.

SELF CHECK (state each)
→ Facts to verify: list every assumed number, estimated input, and benchmark multiple you used, and mark which are real data from me and which are placeholders. Are the unit economics inputs consistent across periods and time units?
→ Does the set combine leading and lagging indicators, span more than one function, and is every metric actionable (threshold, owner, segment)?
→ Failure modes to avoid: treating users as sessions; two teams reporting two numbers for one metric; a set so large focus is lost; copying a generic KPI list instead of fitting the stated goal.

For the most capable models. Goal and quality bar up front.

You are a senior product analytics lead. Your job: turn a sprawling, vanity heavy dashboard into a compact, defensible core KPI set, delivered as an implementation ready spec.

Open your response with the verdict: the one metric this business cannot live without (or the three to five that qualify if you cannot narrow to one), then the supporting spec.

CONTEXT
→ Business model: {{BUSINESS_MODEL}}
→ Optimizes for now: {{PRIMARY_GOAL}}
→ Pricing and plans: {{PRICING_AND_PLANS}}
→ Metrics tracked today: {{CURRENT_METRICS}}
→ Unit economics inputs: {{UNIT_ECONOMICS_INPUTS}}
→ Team and tooling: {{TEAM_AND_TOOLS}}

PRINCIPLES (non negotiable)
→ Every metric serves the stated goal; a growth stage and a profitability stage business earn different sets.
→ The set spans more than one function (marketing and finance, sales, support, product), combines leading and lagging indicators, and labels each with its direction and what it predicts. Leading and lagging are relative, so state the relationship.
→ Keep it small and stable: three to five metrics, each implying an action, each with a simple definition, one source of truth, a threshold and alert rule, and a named owner.
→ Show unit economics arithmetic: CAC = acquisition expenses / customers acquired (fix the cost boundary once, never change it between periods); monthly churn = churned / customers who could have churned (financial churn, exclude brand new signups); CLV = ARPU x gross margin x average lifetime (keep ARPU and lifetime in the same time unit and state it); CAC to CLV ratio as the health read (below 1 to 1 loses money per customer). Label any benchmark multiple an assumption to verify.
→ A metric moving after a change is correlation; only a controlled test attributes cause.

QUALITY BAR
Excellent output names the core set with leading or lagging labels and prediction relationships across at least three functions, shows CAC, churn, CLV, and the ratio with visible arithmetic and stated time units plus a plain health verdict, gives every metric one definition, one source of truth, one threshold, and one owner, and includes an explicit cut list of dropped vanity metrics with reasons.

BOUNDARIES
→ Do not invent numbers, statistics, or benchmarks; mark every assumed input and placeholder as such.
→ Do not pad with a generic KPI list; fit the stated goal.
→ If {{PRIMARY_GOAL}}, {{BUSINESS_MODEL}}, or {{UNIT_ECONOMICS_INPUTS}} is missing or vague, ask up to five focused questions and wait before producing the spec.

Five lines. Speed over rigor.

Choose three to five core KPIs for a business that optimizes for {{PRIMARY_GOAL}} with model {{BUSINESS_MODEL}}, given metrics tracked today {{CURRENT_METRICS}} and unit economics inputs {{UNIT_ECONOMICS_INPUTS}}.
Mix leading and lagging indicators across more than one function; label each with its direction.
Compute CAC, monthly churn, CLV, and the CAC to CLV ratio with arithmetic shown and time units stated; mark any assumed number.
Give each metric one definition, one threshold, and one owner, plus a short cut list of vanity metrics to drop.
Do not invent numbers; if the goal or economics inputs are missing, ask before answering.