During the 2017 French presidential election, the main candidates used data management and analysis tools (NationBuilder, Fédéravox, 50+1) and sought guidance from advisers and service providers in electoral big data. The idea was to target electoral communication, to rationalize activism, and to increase political participation. Data workers aimed to ‘scientifically’ legitimize their new campaign tool offerings, indicating that they could make a real difference in the final result by enabling gains of 1 to 2 points. Some even argued that their services in electoral big data were a way to tackle abstention. By analyzing the words of service providers in electoral big data, we intend to investigate this framing of the use of data in politics as a ‘new electoral science’ and to question its effects on traditional political communication and on the ways in which political parties are perceived and reconfigured. This work uses around sixty interviews conducted with digital communication professionals from the campaign teams of 2012 and 2017 French presidential elections, as well as around thirty interviews carried out with service providers in electoral big data in France.