Promotional Lift
How much extra you sold because of a promotion, above what you'd have sold anyway.
The fork, why two teams get different numbers
The whole metric turns on the counterfactual. BASELINE lift compares promo-period
sales to a modeled or prior "normal" baseline, easy, but the baseline is an
assumption. TRUE INCREMENTALITY compares a treated group to a HOLDOUT/control that
saw no promo, harder, but it isolates cause. The two diverge because baseline lift
counts sales that would have happened anyway, plus PULL-FORWARD (buyers accelerating
a purchase they'd have made next month) and PANTRY-LOADING (stockpiling at the
discount). A promo can show +45% baseline lift and near-zero true incrementality
once pulled-forward and cannibalized volume is removed.
The trap
Baseline lift double-counts demand that only moved in TIME (pull-forward) or in PRODUCT (customers switching from full-price siblings to the promoted SKU), so it flatters the promo and hides a post-promo demand trough and a margin hole.
- The full trap, worked on a real export
- Every formula variant, spelled out
- The reconciliation anchor, what to tie it to and when to refuse
Reference: Trade-promotion / marketing-incrementality measurement convention (reference)