Rubrics

Rubrics

Evaluation rubric (12 criteria)

1 of 2 in this book · 2 min read

Score 0–5: 0 strong negative evidence; 1 mostly unfavorable; 2 weak/limited signal; 3 credible mixed/moderate; 4 strong positive evidence; 5 independently corroborated strong evidence. Blank means unknown/unresearched. This is not an objective scale.

ID Criterion Weight % Evidence to seek Common mistake
C01 Market Demand 15 Reachable segment size and behavior Huge TAM without reachable users
C02 Pain Point 10 Incidence, frequency, severity Confusing stated interest with action
C03 Willingness to Pay 12 Purchases, deposits, pilot contracts Survey "yes" as purchase
C04 Market Trend 12 Multi-year trend, credible data Extrapolating one viral month
C05 Behavioral Shift 8 Observed new routines Tech adoption without behavior change
C06 Entertainment / Emotion 5 Repeat engagement and emotional drivers Views equal value
C07 Viral Potential 8 Share/invite loops and referral conversion Shares as durable acquisition
C08 Retention 8 Repeat use and cohorts Registrations as active users
C09 Community Effect 5 Peer-created value and moderation viability Assuming users contribute for free
C10 Competitive Advantage 7 Alternatives, switching and defensibility "No competitors" claims
C11 Scalability 5 Incremental cost and operational bottlenecks Ignore variable/AI/moderation costs
C12 Validation Cost 5 Reversible small test and budget Full MVP before testing

Weights = 100. For a pure enterprise utility, emotional/viral dimensions may be less important: change weights before comparing ideas, then document why. Never tune weights to favor a preferred result.

Computation

  • coverage_weight = sum(weight of scored rows) (0–100)
  • weighted_observed_score = sum(weight * score / 5) / coverage_weight * 100 when coverage > 0; otherwise unknown.
  • Do not silently equate 100% observed score at 15% coverage with an overall score of 100.
  • evidence_strength and confidence are separate from score.

Gate conditions

  • Positive aggregate score cannot override a failed critical gate: legal impossibility, unworkable unit economics, no reachable buyer, or severe trust/privacy risk.
  • Require actual live tests before claims about price, retention and referrals.