Affective polarization is often blamed on the rise of partisan news. However, self-reported measures of news consumption suffer serious flaws. We often have limited ability to characterize partisan media audiences outside of the United States. I use a behavioural data set of 728 respondents whose online behaviour was tracked over four weeks during the 2019 Canadian federal election. These data were paired to a survey for a subset of respondents. I find that audiences for partisan media are small, and web traffic is driven by an even smaller share of the population. There are few major partisan differences in news media use, and partisan news exposure is higher among highly attentive, sophisticated news consumers, rather than those with strong political commitments.