Enrichment design is a common strategy in personalized medicine in which treatments are given only to patients who are tested positive on a genomic biomarker. Such a targeted trial, as compared to the untargeted trials in which all patients receive the treatments, can substantially reduce the sample size needed for a study as demonstrated in the literature for two-arm trials. To fill in the gaps in the existing literature, we consider trials to evaluate a targeted treatment by comparing pre–post trial outcomes to investigate the intervention effect after the treatment. We investigate the relative efficiency in terms of sample size reduction of an enrichment design against the conventional design, focusing on binary endpoints. The effects of misclassification of the genomic classifier on the relative efficiency are also investigated.