In order to avoid potentially overloading data collection systems, such as Census Questionnaire Assistance call centers, the 2020 Census will stretch out each of its five mailings over a one-week period and put them into groups of different mail cohorts instead of delivering all of its mailings to the whole country on a single day. The purpose of this study is to establish a method by which to separate addresses into cohorts, such that there is an even distribution of cohorts throughout the country, the burden on the call centers is minimized, and the response rate is maximized. Using analyses and results from mid-decade testing at the Census Bureau, we examine the following aspects of cohort assignment: 1. The geographic level at which the cohorts should be broken up (e.g., Census tract versus block); 2. Operational constraints; 3. Based on their relationship to response rates, which/how characteristics should be used to determine the breakup and timing of the cohorts; and 4. The distribution of the cohorts. This study uses statistical methods to attempt to improve the efficiency and success of the 2020 Census.