Rubrics

Rubrics

Research & evidence policy

2 of 2 in this book · 1 min read

  1. Track every material claim with E### evidence ID. URLs must be real and checked, not plausible-looking inventions.
  2. Record title, publisher, date published, accessed date, geography, sample/segment, exact finding, relevance, limitations, and evidence grade.
  3. Prefer primary reports, official statistics, first-party product docs/price pages, user behavior logs and real interviews with documented consent. Third-party blog summaries are weaker.
  4. Separate FACT (source-backed), INFERENCE (reasoned from facts), HYPOTHESIS (untested) and UNKNOWN.
  5. Time-sensitive claims (pricing, product features, laws, employment data) need current dated verification. Expose outdated or unavailable data.
  6. Survey intent != purchase. Signups != activation. Total users != active users. Traffic != retention. Correlation != causation.
  7. Benchmark competitors by the same geography/audience and pricing unit when possible. Include substitute solutions (manual, generic AI, spreadsheets, social platforms).
  8. No invented interviews, market size, CAC/LTV, viral coefficients or confidence intervals.
  9. Privacy-by-design: avoid collecting identifiable CVs, interview transcripts, employee evaluations or personal information in reports without legitimate basis, consent and controls.
  10. Research conclusions are provisional. A price/retention hypothesis needs empirical testing.