This study investigates how social media influencers (SMIs) use visual stimuli to capture viewer attention and enhance engagement. While the influence of SMIs' characteristics on engagement is welldocumented, the impact of their visual content remains less explored. Utilizing visual attention theory, this study identifies stimuli that steer viewers' bottom-up and top-down attention, thus amplifying engagement. Through artificial intelligence (AI), including convolutional neural networks and vision transformers, we analyze over 1.6 million Instagram posts to assess how visual cues influence likes, comments, and follower growth. Findings reveal that although various attention-attracting stimuli drive immediate content engagement, not all effectively translate into sustained interest or long-term engagement behaviors. This research contributes to social media literature by highlighting the significance of visual attention and introduces AI as a novel approach for analyzing visual interactions, offering SMIs strategies to optimize content for maximum viewer engagement.