GLOSSARY

The additional outcome caused by an intervention compared with the outcome that would have occurred without it.

It helps organisations allocate marketing and product investment according to causal impact rather than conversions that would have happened anyway.

An advertiser withholds a campaign from matched regions and compares conversion rates with treated regions to estimate incremental conversions and incremental cost per acquisition.

Results depend on valid assignment, sufficient sample size, limited spillover, stable measurement and a representative test period. Platform experiments may not generalise to other audiences, channels or seasons.

Incrementality is estimated against a counterfactual, commonly through randomised holdouts, geo experiments or other causal designs. Incremental lift compares outcomes for an exposed treatment group with a credible control group while accounting for statistical uncertainty.