Do campaign innovations scale up, and what challenges do modern campaigns face when trying to implement such innovations? Modern GOTV interventions, which are accompanied by rigorous testing using RCTs, are a textbook example of a promising, effective campaign innovation pioneered by Yale political scientists Donald Green and Alan Gerber in the late 1990s and then adapted and scaled up by academics, political parties, and political groups across the United States, and later across the globe. In this paper, we describe how randomised GOTV interventions spread across time and space. We test how the switch from non-partisan interventions conducted by academics to partisan interventions conducted by political campaigns, in tandem with data privacy restrictions in some states and countries but not in others, condition the effectiveness of GOTV interventions. Partisan interventions are less effective in places that make it harder for campaigns to identify supporters. We also show that political campaigns struggle to meet contact rates set as benchmarks by academics, especially in places where contacting supporters is harder. This decline in effectiveness is accompanied by constant innovation relating to the content of the intervention, which appears to counter some of the decline in effectiveness. Our meta-analysis is based on a comprehensive new dataset of approximately 400 GOTV interventions conducted across the globe between 1998 and 2024.