Increasingly, social media platforms are understood by researchers to be valuable sites of politically-relevant discussions. However, analyses of social media data are typically undertaken by focusing on 'snapshots' of issues using query-keyword search strategies. This paper develops an alternative, less issue-based, mode of analysing Twitter data. It provides a framework for working qualitatively with longitudinally-oriented Twitter data (user-timelines), and uses an empirical case to consider the value and the challenges of doing so. Exploring how Twitter users place ''everyday'' talk around the sociopolitical issue of UK welfare provision, we draw on digital ethnography and narrative analysis techniques to analyse 25 user-timelines and identify three distinctions in users' practices: users' engagements with welfare as TV entertainment or as a socio-political concern; the degree of sustained engagement with said issues, and; the degree to which users' tweeting practices around welfare were congruent with or in contrast to their other tweets. With this analytic orientation, we demonstrate how a longitudinal analysis of user-timelines provides rich resources that facilitate a more nuanced understanding of user engagement in everyday socio-political discussions online.