Retrospective voting is a major concern of scholars worried about democratic accountability. The work attempts to measure the retrospective voting of an important urban policy in Brazil’s largest metropolis, São Paulo - SP. The Bilhete Único policy was the first smart-card for public transportation in Brazil. Commuters with the smart-card were able to save up to three bus fares in a journey, a substantial amount of money for most of São Paulo’s citizens. Using a mixed method approach, I try to test the hypothesis that the policy improvement on welfare positively impacted the electorate performance of the incumbent mayor. My findings suggest that not all bus users rewarded the incumbent mayor for delivering the policy. In particular, students and students’ families seem to be the ones who actually felt the policy effects and cast a retrospective vote. This could be due to delivery network difficulties and to the timing of the policy. Additionally, I employ a normative discussion to establish benchmarks of better public goods. Retrospective voting does not necessarily reward the best policies, therefore the need to discuss benchmarks.