Know Your Users via Image Analytics Before Developing Posts : Data-Driven Optimization Framework to Enhance Social Media Engagement

Social media platforms are popular channels for advertisers, enabling firms to advertise products to their audience via posts. Developing these posts with compelling image features and understanding what users prefer to engage with across different platforms is essential for firms to attract new users and keep their current users interested. However, maximizing social media engagement via image features has not been examined rigorously in the literature. To address this critical gap, we propose a novel data-driven optimization framework for analyzing and publishing social media posts across multiple platforms for a planning horizon under a firm's limited budget. To develop the optimization model, we utilize deep learning algorithms and econometrics on social media image posts to distinguish the focal or simple features and the additional or advanced features of images and to estimate the relationship between these features on user engagement. We model the trade-off between the competing costs of social media analytics and the activities associated with developing posts. We demonstrate that the feature-engagement relationship differs between platforms, highlighting the need to develop posts based on each platform's user experience. Our proposed optimization framework incorporates appropriate weights for prioritizing platforms, engagement types, and intensity of engagement on different days for multiple platforms. We provide practically relevant and efficient algorithms for single and multiple platforms that deliver optimal or near-optimal solutions. Our framework and findings have important implications for firms aiming to maximize user engagement on social media platforms