This dissertation investigates the U.S. digital news landscape using audience behavioral trace data from 2017 to 2021. In particular, the study combines Web browsing histories and self-reported demographic attributes to characterize 804 digital news domains and document changes in their audience consumption. Three main themes organize the findings: ideology, localness, and inequality. Focusing on ideology, the dissertation suggests that the U.S. digital news landscape did not grow more polarized during the studied period, and audience engagement on ideologically extreme news domains exhibits both reassuring and concerning patterns. Visitors spend less time on these long-tail domains but extremely conservative domains act as rabbit holes that trap audiences who are more likely to vote. Focusing on localness, the dissertation provides a new measure to quantify local news and presents several characteristics of digital local news domains. Local news domains lose audiences to national outlets during elections, their localness declined from 2017 to 2021, and their audience base became more conservative during the studied period. Focusing on inequality, the dissertation documents sharp demographic disparities in who consumes news, and finds that mobile access attenuates some---but not all---forms of inequalities in terms of web access and news consumption. Audiences reading news on mobile devices are more White and educated, but less affluent and politically engaged. Moving beyond static descriptions of news domains, the dissertation utilizes and develops monthly measures, provides longitudinal evidence, and advances the media sociology approach to news consumption by showing that the domain-audience duality is a dynamic process that can be affected by geography, ideology, technology, and social class. Overall, the dissertation leverages an unprecedentedly large dataset on audience behavior and uncovers trends and patterns on how the U.S. digital news landscape has changed in the past four years.