Ever since the Facebook-Cambridge Analytica data scandal, many observers assume that targeting is an effective campaign strategy. However, we know little about how voters react to targeted campaign messages. I argue that voters react positively to targeted messages at first, but once the targeting gets `too much', they recognize the strategy and punish parties for it.
I tackle this question using a factorial survey experiment in Germany (N=3,200). I take a novel continuous approach to targeting by showing respondents campaign messages that differ in their targeting-degree. The results indeed show that there is a backlash effect of targeting. Voters react positively to messages targeted on one characteristic. However, targeting voters in a high degree, the success of a message decreases and is even significantly lower than not targeting at all. These results hold both for targeting the respondents implicitly or telling them explicitly that they are targeted using personal data. These results question the often-stated manipulation threat of targeting: voters seem to recognize and clearly punish parties that target them `too much'. These findings have important implications for modern election campaigning and its regulations.