Skip to content
Back to Idea Validation

MVP Metrics

Choose the 3-5 metrics that actually test your hypothesis — and ignore the rest.

Why

An MVP with 20 tracked metrics tracks nothing — the signal drowns in noise. You need 3-5 metrics that directly test the hypothesis and a clear definition of what each one tells you. The classic mistake is tracking vanity metrics (signups, page views) that go up and to the right regardless of product quality, while ignoring pirate metrics (activation, retention, referral, revenue) that reveal whether the product actually works. The right metrics let you look at a dashboard and know in 5 seconds whether to proceed, iterate, or kill — the wrong metrics let you feel productive while staying blind.

When

Before MVP launch, alongside cut criteria. Takes 1 hour. Review weekly during the MVP test period.

How

1. Start from the hypothesis
   - "devs will pay $15/mo" → the metric is "trial → paid conversion"
   - Every metric must connect to proving or disproving the hypothesis

2. Choose 1 north-star metric
   - The single number that best represents hypothesis success
   - Example: "weekly paying users who completed the core loop"

3. Choose 2-4 supporting metrics
   - Activation: did they reach the "aha moment"? (e.g. generated first changelog)
   - Retention: did they come back? (e.g. used it 2+ weeks)
   - Revenue: did they pay? (e.g. trial → paid conversion)
   - Referral: did they share? (e.g. invited another user)

4. Define each metric precisely
   - Numerator and denominator (e.g. "paid users / trial users, counted at day 14")
   - Time window (!Time window (e.g. "weekly cohort")
   - Where it is tracked (e.g. "Stripe export + Mixpanel event")

5. Explicitly list vanity metrics to ignore
   - Signups (without activation, signups mean nothing)
   - Page views (traffic, not product!Page views (traffic, not product quality)
   - Total registered users (cumulative, hides retention problems)

6!Total registered users (cumulative, hides retention problems)

6. Set up the dashboard
   - Only the chosen metrics, nothing else
   - Update frequency: daily during MVP, weekly after
   - Compare to your cut criteria thresholds

Example

MVP: AI changelog generator.
Hypothesis: "solo devs will pay $15/mo to auto-generate changelogs."

North-star metric:
  Weekly paying users who generated ≥1 changelog (proves both use AND payment)

Supporting metrics:
  Activation: % of signups who generated their first changelog within 24h
  Retention: % of week-1 users who used it in week 2
  Revenue conversion: paid users / trial users at day 14
  Core loop completion: % of sessions that reach "copy changelog" (not just "generate")

Vanity metrics (explicitly ignored):
  - Total signups (without activation, meaningless)
  - Page views (traffic, not product fit)
  - Changelogs generated (activity, not value — a user generating 100 and never paying is not success)

Dashboard (5 numbers, updated daily):
  Week 2: 180 signups, 64% activation, 38% retention, 5.0% conversion, 72% loop completion
  Read in 5s: activation good, retention concerning, conversion at gray-zone edge → fix retention.

Common Problems

  • ⚠Tracking too many metrics — 20 metrics means you watch the ones that look good and ignore the ones that look bad; limit to 5.
  • ⚠Vanity metrics as the north star — "total signups" goes up regardless of product quality; use a metric that requires product!Vanity metrics as the north star — "total signups" goes up regardless of product quality; use a metric that requires product value (retention, revenue).
  • ⚠Metrics without definitions — "activation" means different things to different people; write the exact numerator/denominator.
  • ⚠Updating the dashboard but not acting on it — a dashboard you look at and do nothing with is theater; tie each metric to a decision threshold.

Best Practices

  • ✓Choose a north star that requires both engagement and!Choose a north star that requires both engagement and revenue — "weekly paying users who used the core loop" is stronger than either metric alone.
  • ✓Define each metric with numerator, denominator, time window, and source — ambiguity leads to measuring different things each week.
  • ✓Explicitly list the vanity metrics you are NOT tracking — this prevents!Explicitly list the vanity metrics you are NOT tracking — this prevents someone from adding them back to the dashboard.
  • ✓Tie each metric to a cut criterion — "if retention < 20% at week 2, kill" makes the dashboard actionable, not just informational!Tie each metric to a cut criterion — "if retention < 20% at week 2, kill" makes the dashboard actionable, not just informative.