The method.
Not a black box.

Brand lift, with the design declared up front.

Method behind every claim. Brand-lift studies on Justify follow the same architecture you'd find in any peer-reviewed marketing-science paper. No EMV, no engagement-as-outcome — every read is an exposed-versus-control comparison with the design declared up front.

• 01
  Survey-panel readsStudies use third-party survey panels weighted to census on age, gender and region. Sample sizes are pre-registered per study and disclosed in the read-out.
• 02
  Matched exposed and controlExposed and control respondents are matched on a pre-declared set of covariates. Residual imbalance is reported in every study.
• 03
  Pre-registered designOutcome metric, cells, minimum detectable effect, power and holdout are locked in a time-stamped document before the campaign goes live.
• 04
  Difference-in-differencesDay-zero baseline subtracted from day-N read. Standard errors clustered. Correction applied to secondary metrics.
• 05
  Adstock and saturationModelled with a structural form declared up front, not fitted after the fact. Parameters estimated from the study data.
• 06
  Audit-ready read-outsEvery study ships with the full design document, the panel composition, and the analysis code that produced the lift number.