Artificial intelligence is causing a 2026. As shown by Sheetin addition to , technology has also caused radical transformations behind the scenes of political marketing.
Among the main innovations is the use of the so-called “synthetic voter”, which partly replaces qualitative research.
Campaigns have long been used in campaigns, but they have become even more important with social networks and AI, which allow the distribution of increasingly segmented messages.
Today it is possible, for example, to target content at middle-class women in the west zone of São Paulo or at voters who are nostalgic for the PSDB’s golden days.
Understand what synthetic voters are, their advantages and limitations.
What is the synthetic voter?
It is a library of virtual profiles built based on data from real groups of voters. Each profile brings together characteristics of a demographic segment. The profiles are created by AI models trained to reproduce opinions, reactions and resistance typical of each group.
What is it for in campaigns?
The synthetic voter partially replaces qualitative research, in which groups of real voters are brought together to give their opinions on certain topics. Based on the results, campaigns test the reception of advertising pieces and candidates’ speeches, for example, and can calibrate speeches and content.
What topics can they give their opinion on?
Synthetic voters can be used, like traditional qualitative research groups, to assess reception of public policy proposals for health, education, housing, etc., or the tone used by the candidate. This makes it possible to modulate the form and content of campaigns.
What is the advantage over traditional surveys?
The main advantages are cost reduction and agility. A well-done qualitative survey with a thousand interviewees can cost R$150,000. The synthetic voter, offered by companies such as SVA Solutions–Galaxies, costs around R$65,000 per month and can be activated at any time, without the need to convene in-person groups.
Do synthetic voters produce reliable data?
The question is whether these profiles are true to reality. The companies that offer the service claim to do so-called grounding: they test the profile responses with real people from each segment to validate the accuracy of the responses.
