Corruption is pervasive across the world, yet voters keep electing corrupt politicians. One common explanation is that voters simply lack information on whether candidates are corrupt, yet studies that deliberately provide such information find electoral accountability is weak at best or non-existent at worst. Despite these results, policy-makers still emphasise the importance of transparency and publicity in the fight against corruption. We contribute evidence to this claim by taking a different approach: rather than disclosing corruption itself we explore what kind of readily available information allows voters to identify corrupt politicians. Based on a unique dataset of politicians in Colombia, we first employ machine learning techniques to identify political and personal characteristics that are associated with corrupt practices. We then design an experiment that randomises the provision of this information to evaluate what candidate descriptions enable voters to discriminate corrupt from non-corrupt politicians. Our study aims to contribute to the policy push for greater information disclosure about candidates for public office, by refining exactly what information leads to better voter choices.