What problem does it solve? Analyses often produce a precise, validated answer to the wrong question because the metric, population, or unit of observation was never pinned down before touching the data. This Skill forces a short framing brief — estimand, decision, population, unit, and what result would flip the decision — to be written and signed off before any code runs. ## Core Features & Use Cases - Framing brief: Pins the decision, exact metric (numerator, denominator, unit, time window), population filters, and unit of observation before loading data. - Routing logic: Directs the work to the right next step — pre-analysis plan for confirmatory questions, causal identification for in-data effects, structural estimation for counterfactuals outside the data, or an everyday analysis plan for exploratory work. - Visualization framing: Treats maps, charts, and dashboards as data deliverables, requiring mark-level unit, encoding, and join-cardinality definitions before plotting. - Use Case: A stakeholder asks "did the pricing change work?" The Skill produces a brief defining the treatment, counterfactual, estimand type (e.g., ATT on 30-day retention), and decision threshold, then routes to causal identification before any regression is run. ## Quick Start Before analyzing this dataset, frame the question by defining the exact metric, population, unit of observation, and the decision it informs, then get my sign-off.
