Uplift-based audience targeting
Incrementality targeting that finds customers advertising can actually influence—not only those already likely to buy.
Incremental
Spend aimed at buyers ads can actually move
Challenge
Traditional targeting optimized for purchase likelihood but failed to separate shoppers influenced by ads from those who would buy anyway—limiting incremental sales and wasting spend.
What we did
FocusKPI used uplift modeling and rigorous tests across treatment-effect formulations and ensembles (including random forests), selecting the model with the best uplift separation, stability, and generalization. Users were prioritized by incremental impact and the model was embedded for campaign optimization.
Results
Advertisers redirected spend to high-impact audiences, improving efficiency and incremental sales—and reinforcing the network’s performance-focused positioning.
“This changed how we think about targeting—we’re no longer just finding buyers, we’re finding the customers we actually need to influence.”
Why it worked
The work optimized for causal impact rather than propensity alone—tying spend directly to incremental outcomes.
Discuss a similar engagement
We typically start with a scoped proof of concept before production deployment.