Estimating the effect of an action or treatment rather than just a correlation. Covers randomised experiments, difference-in-differences and synthetic controls, propensity scores and instrumental variables, causal graphs, and uplift modelling for product decisions.