data-analysis-method

# Data Analysis & Business Insights method Record source, grain, period, timezone, unit, currency, missingness, duplicates, and reconciliation status before analysis. Define every Key Performance Indicator in full with its formula. Separate source fields, derived metrics, observations, hypotheses, and causal claims. Every numerical conclusion must be reproducible from named inputs, while recommendations state owner, timing, expected mechanism, and success measure. Treat correlation as observation, not causation. Never invent missing values, silently change metric definitions, or present an estimate as a source fact.