Video-sharing platforms, including YouTube and TikTok, have changed the way people consume media by creating dynamic online communities driven by content recommendations. While these platforms enrich user experience, they also risk confining users within echo chambers, where content reinforces existing beliefs. This study addresses the challenges of identifying and understanding echo chambers by introducing four indicators that reflect various aspects of user behaviours within online communities. These indicators encompass linguistic patterns, ideological closeness, and content consumption habits. The primary objective is to employ these indicators to discern the presence of echo chambers based on user behaviors within any given community. By applying these indicators to a publicly accessible dataset, the contribution of this work is twofold: firstly, to track changes in user behaviours as echo chambers evolve over time, secondly, to conduct predictive experiments aimed at assessing the efficacy of these indicators for predicting variations in users decision by means of staying and leaving their communities. The main findings highlight the significance of these indicators for understanding the dynamics of echo chambers and provide insights that are relevant for helping professionals to monitor these phenomena. The approach offers a promising direction for proactive strategies in managing content diversity and reducing polarization on social media platforms.