The former director of the Census Bureau, Dr. Robert Groves, emphasizes in his 2011 vision document, The Future of Producing Social and Economic Statistical Information, the importance of the Census Bureau remaining relevant by responding to the threat of competing organizations producing estimates cheaper and faster through passive data sources. More specifically, in addressing this issue in the context of mixed-mode surveys, he states, "€œPrior to contacting them [respondents], we will be ignorant about the mode-preferences of our sample units; we must be able to switch across modes in real-time during the data collection phase to produce timely estimates." He went on to say that, "€œwe need to mount mixed-mode surveys that have real-time, rule-based switching across modes based on statistical analysis of paradata." Some of the potential benefits of switching modes in real-time include reducing lag time for following up with nonrespondents, reduced response burden by providing respondents with a convenient or preferred mode of response, and cost savings by maximizing the number of respondents willing to respond in the cheaper modes of data collection such as Internet and mail. As a first step in addressing this vision, we explore the use of adaptive survey design methods in the ACS by linking administrative data to the 2011 ACS Internet Test samples and developing a model-based assignment of mode switch strategies, focusing only on switching Internet nonrespondents to mail. We conduct a simulation of the adaptive survey design process with the April Internet test sample and then attempt to validate our model-based approach using the November Internet test sample comparing outcomes in Internet response and length of time to Internet nonresponse follow-up.