Political scientists have long been puzzling over why some voters choose parties that they acknowledge as not being the most proximate to their views. This research delves deeper into the determinants of tactical voting by attempting to understad how do personal and systemic characteristics shape the impact of personalised political information on tactical voting. We employ data from the largest voter advice application (VAA) in Spain (N=500,000), run in the context of the July 23, 2023, national elections. This digital tool was combined with a survey experiment were we randomly presented users with their levels of political congruence with different political parties before and after assessing their voting probabilities. Our findings are as follows. First, displaying political congruence with the different political parties increases the likelihood of voting for the own party, while decreases the likelihood for voting for the others--particularly within ideological blocks. Second, individuals scoring lower congruence with their preferred party are more impacted by the treatment. Third, when a different party from the preferred one appears to be first ranked, the treatment advantage for the party aimed to vote vanishes. Fourth, centrist voters are the most affected by the treatment condition. Overall, these findings highlight the role of personalised political information in a fragmented electoral context and confirm that such information reinforces pre-existing electoral intentions but rarely converts them. Our study contributes to understanding how digital tools influence voting behavior, providing evidence of their nuanced effects on voter decision-making in complex electoral settings.