Since their introduction in 2016, Instagram Stories have become a core feature of the platform, offering users short-lived posts that disappear after 24 h. Although research on Instagram content is growing, Stories remain understudied, possibly due to challenges in data collection. This chapter addresses that gap by presenting a comprehensive workflow for the computational analysis of Instagram Stories. The workflow outlines steps for data collection, preprocessing, and analysis of textual and visual content, utilising tools such as optical character recognition and automated transcription to capture Stories’ multi-layered nature. By deconstructing each Story into its text, audio, and visual components, researchers can analyse these elements separately or in combination to uncover patterns. The adaptable framework supports various disciplines and research questions, including historical sciences or political communication, the latter being illustrated through a fictional case study of Martian election campaigns.

Additionally, the chapter explores using large language models like GPT-4 to automate content classification, showing how these tools assist in analysing text and images. The workflow emphasises computational approaches while advocating for the inclusion of human annotations to ensure accuracy. This method makes computational multimodal analysis more accessible to researchers with limited technical expertise. Practical guidance, including example notebooks, is available online to help researchers easily apply this methodology.