Sentence completion, originally a semi-projective psychological technique, has been used as an effective and lightweight user research method in UX design. More information is yet still needed to understand how different sentence stems probe users’ insights, thereby providing recommendations for effective sentence completion surveys. We used the completion method on a large-scale sample to explore (e-)readers’ experiences and needs. Depending on their reading habits, participants (N=1880) were asked to complete a set of sentences, as part of a web survey. With 14143 user ideas collected in two weeks, our results confirm that remote online sentence completion is a cost-effective data collection method able to uncover feelings, attitudes, motivations, needs, or frustrations. Variation in sentence stems affected collected data in terms of item response rate, idea quantity as well as variety and originality. Building on previous research, this paper delivers actionable insights to optimize the richness of sentence completion outputs.