Significant theories of public opinion, polarization, and campaign finance hinge on how and whether campaign contact and advertising persuades Americans to vote differently. We argue that the best estimate of the marginal effects of campaign contact on Americans' candidate choices in general elections is zero. First, a systematic meta-analysis of 36 experiments estimates an effect of campaign contact on vote choice in general elections of zero. Second, we present a series of nine original experiments that together increase the quantity of statistical evidence in the literature about the persuasive effects of personal contact in general elections by 10-fold. In both the literature and our studies, we show an intriguing pattern wherein persuasion in general elections typically only appears when outreach is conducted far in advance of an election and measured immediately---but that this early persuasion decays rapidly, while contact close to election day does not even have immediate effects. Our findings raise new questions about the generalizability of studies conducted far from elections that measure outcomes immediately. Substantively, they may help explain why campaigns have turned away from trying to win over moderate voters and toward rousing the enthusiasm of existing supporters to turn out.