Media firms such as newspapers compete in a two-sided market, one between content providers and news consumers while the other between content providers and advertisers. I examine the impact of online subscription and advertisement revenues on newspapers' endogenous content choice. An analytical model demonstrates that switching from a subscription to advertising revenue model changes firms' horizontally differentiated content choice. Additionally, I provide an empirical analysis of U.S. digital newspapers and evaluate a structural model of supply and demand. An application of web scraping, Natural Language Processing (NLP), and Machine Learning for data collection and data analysis is presented. The counterfactual estimation suggests that a subscription-based model improves firms' profits.