The Census Bureau’s mission is to be the nation’s leading provider of quality data about its people and economy. The Census Bureau must do this with limited resources and with care to avoid overburdening survey respondents. The methodology discussed in this paper, cross-survey modeling, allows the Census Bureau to enhance the usefulness of federal data products by using machine learning to bridge the gap between surveys. This method uses data from one survey (typically a smaller survey with a rich set of items) to train a machine learning model to predict an outcome of interest. The model is then applied to another survey (typically a larger survey with fewer items but more statistical power) to estimate how respondents may have answered specific questions if they had been asked. In this way, cross-survey modeling allows for information from a survey with limited geographic detail but rich subject matter to be transferred to another survey with more granular geographic detail. The result is a fused dataset from which new measures can be estimated. The use of cross-survey modeling as part of the Community Resilience Estimates for Heat is an example of how these innovative methods can increase the utility of current data products. To meet data user needs, we used this method to model a variable, whether or not a housing unit had an air conditioning unit from the American Housing Survey to the American Community Survey. This vastly improved the utility of the Community Resilience Estimates for Heat by allowing new information to be integrated into the source data without collecting new survey data.