Define the right KPIs and metric definitions

Role: Act as a head of analytics defining the KPI framework for a domain or team. Context: Domain and its objective: [DOMAIN_OBJECTIVE]. Candidate metrics in use or proposed: [CANDIDATE_METRICS]. Where definitions currently conflict or are unclear: [CONFLICTS]. Data available and its grain: [DATA]. Who will use the KPIs and how (targets, comp, dashboards): [USE]. Known gaming risks: [GAMING_RISKS]. Task: Produce: (1) a KPI tree - the north-star metric, the 3-5 driver metrics beneath it, and the input/operational metrics beneath those, with the causal logic; (2) a metric dictionary - for each: precise definition, formula, grain, owner, source, refresh, and how it differs from any similar metric; (3) which metrics are for steering vs monitoring vs diagnostics; (4) anti-gaming notes - for each KPI that carries incentive weight, the likely gaming behaviour and the guardrail metric that catches it. Format: Sections: KPI tree | Metric dictionary (table) | Steering / monitoring / diagnostic | Anti-gaming guardrails. ~550 words. Verification: Flag any two metrics that measure nearly the same thing, any definition the data cannot actually support, and any KPI with no owner. Validation: Test the tree: if every driver metric improved, would the north-star necessarily improve? If not, the logic is broken - fix it.