Social media platforms have become fertile spaces for online scam activities, yet existing studies often focus on victim profiles or technological detection this creating a gap on the discursive mechanisms through which scams operate and are resisted. This study aims to expose the deceptive tricks and specifically examines the discourse strategies employed in scam texts on Facebook and WhatsApp, the linguistic features that characterize these strategies, and the counter-discourse practices through which users resist manipulation. Data comprised twelve naturally occurring, multiturn scam conversations, six from Facebook and six from WhatsApp, collected between January and March 2024 through purposive sampling from consenting participants who had encountered and resisted scam attempts. The conversations were subjected to qualitative discourse analysis, guided by Van Dijk’s Socio-Cognitive Model (1993) which integrates discourse structures, cognitive activation, and social context. The findings reveal that scammers strategically construct authority, urgency, and legitimacy through religious and institutional impersonation, while users resist through probing questions, reframing, and strategic delay. By foregrounding both manipulation and resistance within real interactions, The study contributes to knowledge by extending van Dijk’s socio-cognitive framework to the domain of digital fraud, foregrounding user agency alongside scammer manipulation, and offering a platform-sensitive analytical model applicable to discourse studies, digital literacy education, and fraud prevention.