SuperAPI AI Directory

startup-metrics

Data & analytics

View repository

Updated 10-16 07:16

About this skill

Startup Metrics

Make the decision and definition explicit before interpreting a number.

Workflow

  1. Establish the business model, stage, decision, user journey, and natural value interval.
  2. Read metric-system.md to define metrics and sources before analysis.
  3. When values appear inconsistent or are presented for comparison, keep them both, compare definition, entity, cohort, window, source, and as-of date, classify their relationship, and assign any necessary reconciliation.
  4. Read cohorts-experiments.md for retention, segmentation, or tests.
  5. Read pmf-evidence.md before making a product-market-fit claim.
  6. Return the diagnosis, confidence, limitations, missing instrumentation, proposed action, owner, and review date. If ownership is not established, use owner: unassigned; if a calendar date would be invented, use an evidence-based review trigger such as “after the cohort reaches the defined observation window.”

Boundaries

  • A metric without a definition, source, and as-of date is provisional.
  • Do not average values with different definitions or windows.
  • Missing attribution or instrumentation is a finding and an owned corrective action, not permission to infer.
  • Route unexplained customer behavior to customer-discovery, strategic implications to startup-strategy, and weekly commitments to founder-operations.
  • Obtain authorization before changing production instrumentation, billing, live dashboards, or external reporting.