Podcasts are central to media ecosystems, yet no research program engages them systematically as digital trace data. We address this deficiency by demonstrating how podcasts’ RSS-based distribution infrastructure enables podcast data collection and analysis at scale. With a dataset of 6,055 unique Canadian podcasts comprising 655,123 episodes, we provide an initial description of the Canadian podcast ecosystem in terms of its production patterns and political content. Our descriptive analysis unfolds in three parts. First, at the podcast level, we find that the Canadian ecosystem is substantively diverse, primarily urban, and biased towards English. Second, at the episode level, we document exponential growth through the 2010s, a pandemic-era surge in new podcast launches, and that the top 10% of podcasts produce nearly half of all episodes. Third, in terms of content, we find that online-first podcasts exhibit greater variation in sentiment than broadcast formats, that discussions of economic topics during 2023–2024 were more sentimentally negative when focused on inflation, and show how word embedding methods can be used to map how contested issues like housing are discussed across the ecosystem. In addition to our findings and data, we include open tools that others can use to expand this research agenda. Our approach demonstrates that podcasts can and should be treated as data.