Political campaigns employ complex strategies to persuade voters to support them. We analyse the contributions of elements of these strategies using data from a field experiment that randomly assigned canvassers to districts, as well as messaging and endorsement conditions. We find evidence for a strong overall campaign effect and show effects for both message-based and endorsement-based campaigns. However, we find little evidence that canvassers varied according to their persuasive ability or that endorser identity matters. Overall the results suggest a surprisingly muted role for idiosyncratic features of prospective persuaders.