Patchy distributions: Optimising sample size