This paper focuses on the increasing prevalence of devices for self-evaluation in the context of social media. It suggests that self-measurement - i.e. aggregates of user activities, received likes, comments, retweets and other algorithmic indicators of reputation and influence - patterns and presents user data in new ways to act upon and modify the self. Furthermore, the technologies involved in self-evaluation are not designed to merely capture user engagement with social media, but act as framing devices, making certain types of insight and action possible while ruling out others. In this sense, self-evaluation can be understood as part of recommendation and premediation culture online and as devices to curate and organise streams of real-time web content. Social media platforms like Twitter and Facebook are based on the continuous engagement of users in activities made possible through their software design and so far these platforms have only offered limited possibilities for users to access and make sense of data they generate in interaction with their contacts. But with the proliferation of user actions, especially Facebook has started to introduce devices that increasingly aggregate and cluster user activities into stories, ranks or numbers. Focusing on Facebook and Twitter, the paper will trace modes of self-evaluation both emerging within social media and from third-party plug-ins, apps and services build on top of platforms. Special attention will be paid to so-called influence measurement devices, most particularly the Klout rank, that attempt to evaluate users’ engagement with and impact on their network. The analysis will explore how the Klout rank encourages users to act on their data and will reflect on the role of numbers, more specifically counts and rankings, seeking to describe them as media of self-evaluation. It will unfold the affective states involved in rankings and will draw on mathematical concepts of ordinality to reflect on their performative capacities. Finally, the paper will position the emergence of self-evaluative media in the context of recommendation cultures, indicating that it not only enables users to transform the self based on data, but is embedded in corporate attempts to channel user engagement into particular directions. In this sense, the investigation of self-evaluative media will draw attention to issues of temporality, numbering and measurement in social media platforms while singling out new modes of pre-structuring user activities through devices.