Video content has become a crucial tool for marketers seeking to engage audiences and strengthen brand recognition. Nonetheless, creating videos that are both memorable and impactful remains a challenge, highlighting the need for robust tools and datasets to evaluate video effectiveness defined as a video’s ability to capture attention, be remembered, and reinforce brand identity. In this paper, we introduce VIDEM (VIDeo Effectiveness and Memorability Dataset), a new dataset designed to advance research on video effectiveness, memorability and brand recognition. Memorability refers to how likely people will remember the media content after viewing it, while effectiveness encompasses both memorability and the video’s success in reinforcing brand identity. VIDEM is the result of a large-scale annotation campaign that employed the Video Effectiveness Measurement Tool (VEMT), which incorporates advanced crowdsourcing techniques and quality control mechanisms. The dataset comprises 424 YouTube videos, annotated by 1403 participants across two main tasks: video memorability and brand memorability. This paper outlines the dataset construction methodology, discusses challenges encountered during the annotation process, and explores the potential applications of VIDEM in video marketing. Additionally, we perform a quantitative statistical analysis to highlight key insights from the dataset. By making VIDEM available at the MediaEval Multimedia Evaluation Benchmark, we foster research on video memorability, brand impact, and marketing analytics, aiding information retrieval and access.