Faces vs. Words: Visual and Textual Drivers of Engagement in Charity Campaigns
(A2026-132968)

AUTHORS

Sofia Pansoni, Università di Macerata; Giacomo Gistri, University of Macerata

ABSTRACT

Nonprofit organizations rely on engaging social media content to capture attention and mobilize support. In this context, text overlays are a key strategy to increase message clarity and immediacy. Yet, little is known about how these textual cues interact with other visual elements, or which specific textual characteristics make them effective. Through two complementary studies, we analyzed 1,982 Instagram posts published by three main non-profit organizations, annotated through a generative AI–based multimodal content analysis. Study 1 highlights that text overlays significantly increase engagement, but this effect reverses when human faces are present. Study 2 isolates posts containing text overlays without faces and employs automated machine learning to identify which text features best predict engagement. Results show that engagement is maximized when messages are short, emotionally negative or mixed in tone, visually simple, and positioned centrally or at the bottom of the image.