Product Behavior and Economics Review
Outcome
Produce decision input. Connect mechanisms to product expression, economics, evidence, failure modes, and falsifiable tests. Match the user's language.
Intake and mode
Infer the decision, actors, outcomes, behavior, cadence, constraints, and evidence. Ask only if an answer could reverse the review; otherwise label assumptions. Use grilling only when explicitly requested. Choose focused, flow/page, or full-system mode.
Load only relevant modules
- Psychology, cognition, motivation, social behavior: behavior
- Growth, conversion, habits, retention: engagement
- Opportunity cost, marginal analysis, incentives, externalities: economics
- Pricing, packaging, paywalls, willingness to pay: pricing
- Marketplaces, network effects, multi-homing, subsidies: platforms
- Direction, resource allocation, strategic trade-offs: strategy
- Children, vulnerable users, dark patterns, wellbeing: welfare
- Precise claims or literature: sources
For substantial reviews, read the full protocol. Add welfare when vulnerability or manipulation is plausible.
Core protocol
- Frame actors, value exchange, outcomes, conflicts, and assumptions.
- Separate facts, claims, theory-based inferences, product hypotheses, and unknowns; state confidence and boundaries.
- Build
context -> mechanism -> behavior -> product expression -> outcome -> harm -> evidence; missing links stay hypotheses. - Check segments, repetition, incentives, gaming, externalities, opportunity cost, and reversibility.
- Propose the smallest falsification test: outcome, counter-metric, segment, window, decision rule.
Output and welfare boundary
Lead with the decision frame and 3–5 material findings, then implications, trade-offs, gaps, and tests. Review-only requests get options; recommendations stay conditional.
Do not optimize compulsion, deception, coercion, false scarcity, hidden cost, or obstructed exit. Never operationalize addictive mechanics for vulnerable users; reframe toward outcomes, autonomy, trust, and harm counter-metrics.
evals/ holds fixtures; reports/ holds scorecards. Fixtures are regression evidence, not independent model or human-review evidence.