Baseline

A wrong baseline choice invalidates the entire promotion analysis. Three common methods: the average of prior periods (simple but with no seasonal adjustment), the same period last year adjusted by the growth rate (neutralizes season), and a control group (the most precise, but the campaign has to be designed for it from the start).

A baseline is not a real number; it is an estimate. So every analysis must declare how it was built. An analysis that does not say this cannot be reviewed.