Enter a pool of reps and a KPI baseline to build the plan.
Tools · One-page brief
Training Holdout Planner
Rolling out a sales training, a new call script, a coaching program? Before anyone asks “did it work?”, hand them this page: who waits, what decides, when we read, and how big the holdout has to be to see anything. Fill in the left; the brief on the right rewrites itself.
Build the plan
What the page assumes: reps are assigned at random within strata, everyone is measured the same way in two equal windows, and the read is the trained group’s change minus the holdout’s (a difference-in-differences, analyzed with each rep’s pre number as a covariate). Detectable lift is the smallest true effect the test calls real at 95% significance and the chosen power. Illustrative defaults; the plan is only as good as the pre-window pull.
The holdout is the plan. Getting a leadership team to sequence a rollout instead of blasting it is the same conversation as getting an AI feature an eval set before launch. I run both.