When do you use the Multi-Stakeholder Framework (User / Advertiser / Ecosystem / Revenue)?
Use it whenever the question involves quality, ranking, tradeoffs, or ads surfaces. Signals: “relevance,” “spammy,” “value,” “experience,” “quality,” “fairness,” “balance.”
When do you use the Product Health Framework (Upstream → Core → Downstream)?
Use it when diagnosing drops, regressions, trends, or localizing where a system is failing. Signals: “metric dropped,” “complaints increased,” “retention declined,” “what happened?.”
When do you use Metric Decomposition (Users × Frequency × Conversion × Depth)?
Use it when reasoning about volume, engagement, growth, funnels, or success metrics. Signals: “define success,” “measure improvement,” “evaluate performance,” “increase engagement.”
When do you use the Quality Framework (Relevance / Utility / Trust-Safety / Experience)?
Use it when asked about ‘quality’ of anything: ads, feed, search, recommendations, marketplace listings. Signals: “improve quality,” “spammy,” “irrelevant,” “bad experience,” “low-value content.”
When do you use the Tradeoff Framework (User Experience vs Revenue vs Integrity)?
Use it when decisions require balancing outcomes or when optimizing one metric risks harming another. Signals: “revenue impact,” “user complaints but good revenue,” “balance,” “optimize without hurting.”
When do you use the Experimentation Framework?
Use it when designing, evaluating, or interpreting tests or changes. Signals: “test a feature,” “design experiment,” “validate idea,” “prove impact,” “A/B test.”
When do you use the Diagnostic Framework (Segment → Localize → Hypothesize → Validate)?
Use it whenever the question is about problems, issues, drops, lagging metrics, or failures. Signals: “why did this happen?,” “drop,” “decline,” “complaints rising,” “root cause,” “unexpected change.”