Uplift Modeling Lab — Persuadables, Sure Things & Sleeping Dogs

Automated marketing campaigns don't just need to know who converts — they need to know who converts because of the campaign. This simulator generates a synthetic customer population with a real, hidden control-vs-treated conversion probability for each person, plots them in 3D as a live scatter of p(convert|control) against p(convert|treated), and colors them into the four uplift quadrants: Persuadables, Sure Things, Lost Causes and Sleeping Dogs. Set a predicted-uplift targeting threshold and a model-noise level to see how a real two-model (T-learner) uplift model would rank and select customers, and watch the true incremental conversions, sleeping-dog backfire, and Qini coefficient update live as you change the campaign's targeting policy.