Work out uplift modelling value instantly with clear inputs, formula shown and shareable results.
Uplift modelling targets the customers whose behaviour the treatment actually changes, rather than those most likely to convert anyway. The measurable quantity is the difference between treated and control conversion, and only that difference creates value: a campaign converting 14 percent against an 11 percent control created 3 percentage points of genuine incremental effect, not 14.
Incremental value
uplift = treated rate - control rate; incremental conversions = customers x uplift; value = incremental conversions x value per conversion
Propensity ranks who will convert, which includes sure things who would have converted without the campaign. Spending on them adds cost with no incremental return.
Yes. Without an untreated control you cannot observe the counterfactual, and the whole metric collapses into ordinary conversion rate.